JP5899764B2 - Noise removal apparatus, noise removal method, noise removal program, and recording medium - Google Patents

Description

  The present invention belongs to the technical field of a noise removal apparatus, a noise removal method, a noise removal program, and a recording medium. More specifically, the present invention relates to a technical field of a noise removal apparatus and a noise removal method for removing noise included in a digital image formed by a plurality of pixels, a noise removal program for the removal, and a recording medium on which the noise removal program is recorded. Belongs.

  2. Description of the Related Art In recent years, for example, an image pickup device incorporated in a digital camera or the like (for example, an image pickup device such as a CCD (Charge-Coupled Device)) has been remarkably improved in performance such as high density. Along with this, development of techniques for reducing or eliminating noise included in an image signal output from an image sensor or an image signal obtained by processing the image signal has been actively performed. .

  Here, the noise included in the imaging signal output from the above-described imaging element includes, for example, an imaging including an original image (including a moving image or a still image; the same applies hereinafter) due to the structural reason of the imaging element. There is noise that will be included in the signal. In addition, noise included in an image signal obtained by processing an imaging signal includes, for example, so-called block noise caused by a block encoding process for generating a digital image being performed on the imaging signal.

  And with respect to such various kinds of noise, the original image has a low amplitude component without impairing the high amplitude component of the original image (that is, the shape of the contour of the image component (subject) included in the original image). As a technique for removing noise and distortion included as an amplitude component, for example, there is a technique disclosed in Patent Document 1 below. Note that the distortion specifically includes, for example, distortion at a block boundary corresponding to the block compression encoding method and distortion around the contour in the original image. The technique disclosed in Patent Document 1 is a noise removal filter using a so-called ε-separated nonlinear digital filter. The ε-separated nonlinear digital filter is disclosed in detail, for example, in Non-Patent Document 1 below.

Here, the outline of the noise removal processing by the ε-separated nonlinear digital filter will be described. In the noise removal processing, noise and the like that are low-amplitude components are removed without damaging the high-amplitude components in the digital image as the original image by the following six-stage processing from (1) to (6). In other words, the noise removal process is a process of smoothing the low amplitude component without damaging the high amplitude component (see FIG. 4 of Patent Document 1).
(1) Among digital images input as image signals, a square area composed of M × N pixels (M and N are natural numbers) is set as a “filter target area” and included in the filter target area. The brightness of a pixel is obtained for each pixel.
(2) Temporarily storing luminance information indicating the luminance of the central pixel of the filter target area.
(3) A difference between the luminance indicated by the stored luminance information and the luminance of a plurality of other pixels other than the central pixel of the filter target region is detected for each other pixel. In general, in order to simplify the processing, the difference between the luminance of the central pixel stored in the above (2) and the luminance of all the pixels (including the central pixel) in the filter target area is uniformly displayed. Ask for. In this case, the difference between the luminance of the central pixel stored in (2) above and the luminance of the central pixel as a part of all the pixels in the filter target area is naturally zero.
(4) A pixel to be subjected to noise removal processing is determined using the difference from the other pixels. The determination process is hereinafter referred to as “noise determination process”. That is, among the other pixels, the other pixels whose detected difference is equal to or greater than a preset noise threshold are assigned to the other pixels whose luminance belongs to the high amplitude component (that is, the other pixels Are excluded from the target pixel of the noise removal process. On the other hand, other pixels whose difference from the luminance of the central pixel is less than the noise threshold are regarded as noises because the luminance of the other pixels belongs to the low amplitude component (that is, the other pixels include noise or the like). The target pixel for the removal process.
(5) A process of multiplying the difference for the target image for noise removal processing by a preset filter coefficient is performed for each target image, and all the multiplication results for each target image are added. In this case, the filter coefficients multiplied by each of the target pixels are all the same.
(6) The above addition result is added to the luminance of the central pixel indicated by the luminance information stored in (2), and the addition result is the luminance after noise removal at the central pixel.

  By performing the six-stage processing from (1) to (6) described above repeatedly using all the pixels constituting the digital image sequentially as the central pixel, the ε-separated nonlinear digital filter for the digital image is used. A noise removal process is executed. In the noise determination process (4), depending on the noise threshold setting method, other pixels whose detected difference is larger than the noise threshold are excluded from the target of the noise removal process, and the difference is equal to or less than the noise threshold. Other pixels may be the target pixel for noise removal processing.

  On the other hand, in the noise removal filter disclosed in Patent Document 1, the noise removal strength of the entire noise removal filter is controlled by making the noise threshold used for the noise determination process (4) variable. ing. However, in Patent Document 1, the intensity of noise removal processing performed on the target image after the target pixel of noise removal processing is determined by the noise determination processing of (4) above (that is, noise as the noise removal filter itself). No mention is made of changing the removal intensity.

  On the other hand, in an actual image sensor, there are a wide variety of noise removal intensities to be applied depending on the contents of a digital image (moving image or still image) that is the target of noise removal processing. Therefore, in addition to making the noise threshold variable as in the technique described in Patent Document 1, if the noise removal strength as the noise removal filter itself can be changed, a highly versatile and more effective noise removal filter can be obtained. Will be obtained.

Here, the noise removal intensity as the noise removal filter itself generally correlates with the range (size) of the filter target region. That is, the noise removal intensity in the noise removal filter itself disclosed in Patent Document 1 or Non-Patent Document 1 is obtained by assigning the filter coefficient corresponding to the center pixel to all the pixels (including the center pixel) in the filter target region. Correlate to the value divided by the sum of the corresponding filter coefficients. The filter coefficient is expressed as “predetermined coefficient A _{i, j} ” in Patent Document 1, and is expressed as “filter coefficient a _k ” in Non-Patent Document 1. The value obtained by dividing the filter coefficient corresponding to the center pixel by the sum of the filter coefficients corresponding to all the pixels in the filter target area is the “weight of the filter coefficient” at the center pixel, resulting in noise removal. The noise removal strength in the filter itself correlates with the weight of the filter coefficient in the center pixel. Specifically, for example, a filter target area of 3 pixels × 3 pixels (number of pixels is 9), a filter target area of 5 pixels × 5 pixels (number of pixels is 25), and a filter target area of 7 pixels × 7 pixels (number of pixels is 49) If the filter coefficients corresponding to all the pixels are “1” in the comparison in 49), the weight of the filter coefficient in the center pixel of the 3 × 3 filter target area is 1 ÷ (3 × 3) ≈0.11. Similarly, the weight of the filter coefficient in the center pixel in the case of 5 pixels × 5 pixels is 1 ÷ (5 × 5) ≈0.04, and the weight of the filter coefficient in the center pixel in the case of 7 pixels × 7 pixels is 1 ÷ (7 × 7) ≈0.02. The smaller the filter coefficient weight in the center pixel, the stronger the influence of other pixels around it (see FIG. 4 of Patent Document 1). As a result, the noise removal filter itself related to the center pixel As a result, the noise removal strength is also increased. In the example of the three filter target areas described above, the noise removal strength as the noise removal filter itself, that is, the noise removal strength for smoothing the low amplitude component without damaging the high amplitude component in the image signal is the center pixel. The filter target area of 7 pixels × 7 pixels having the smallest weight of the filter coefficient at is the strongest.

  As described above, the noise removal strength as the noise removal filter itself can be increased by expanding the range of the filter target region, while the noise removal strength can be reduced by narrowing the filter target region.

Japanese Unexamined Patent Publication No. 2000-102020 (FIG. 1, FIG. 4 etc.)

Harashima et al., “Ε-Separation Nonlinear Digital Filter and its Applications”, IEICE Transactions, Vol. 65, NO. 4, pp. 297-304, Apr. 1982

  However, when the noise removal intensity as the noise removal filter itself is controlled by expanding or narrowing the range of the filter target region, a plurality of types of storage capacities corresponding to the range (width) of the filter target region are included. It is necessary to prepare memory. This is because when the range of the filter target region changes, the number of other pixels to be processed in the above (1) to (3) also changes, so the memory capacity of the memory necessary for these processes also increases. Because it will change. Therefore, when trying to control the noise removal strength as the noise removal filter itself by the above method, for example, there is a hardware limitation such that the storage capacity of the memory mounted on one chip must be changed later. There are many problems, and there are many problems that are impossible in practice.

  Therefore, the present invention has been made in view of the above-described problems and the like, and its problem is that it is highly versatile and effective noise removal processing for a digital image while requiring no hardware change. It is an object of the present invention to provide a noise removal device and a noise removal method capable of performing the above, a noise removal program for the removal, and a recording medium on which the noise removal program is recorded.

In order to solve the above problem, the invention according to claim 1 is characterized in that at least one other pixel around one pixel in a digital still image formed by a plurality of pixels, A detection unit such as a difference calculation unit that detects a difference value of a signal between the pixel and a comparison unit such as a threshold determination unit that compares each detected difference value with a threshold value; and the comparison unit Based on the comparison result, an identification means such as a threshold value determination unit that identifies the other pixel to be subjected to noise removal processing, and an area of the digital still image that includes dark current noise, the digital still image A dark part detecting unit such as a dark part determining unit that detects the dark part of the image, and when the detected dark part includes the one pixel, the detected difference value of each of the other pixels specified Multiplied by The filter coefficients, based on the detected dark position in the digital still image, and setting means of the filter setting unit for setting for each said other pixels, for each of the other pixels identified was the detection A multiplication unit such as a calculation unit that multiplies the set filter coefficient by a difference value; and an addition unit such as a calculation unit that adds and outputs the multiplication result of the multiplication unit to the signal of the one pixel; The dark portion is formed by detection processing by the detection unit, comparison processing by the comparison unit, specification processing by the specification unit, multiplication processing by the multiplication unit using the set filter coefficient, and addition processing by the addition unit. Bei as the one pixel of each of the pixels, and control means of the arithmetic unit or the like is repeated for each of the pixels forming the dark portion, the to That.

In order to solve the above-mentioned problem, the invention according to claim 4 is a noise removal method executed in a noise removal apparatus for removing noise included in a digital still image formed by a plurality of pixels. A detecting step of detecting a difference value of a signal between each of at least one other pixel around one pixel in the image and the one pixel, and each detected difference value and threshold value; A comparison step for comparing each of the digital still image, a specifying step for specifying the other pixel to be subjected to noise removal processing based on a comparison result in the comparison step, and the digital still image including dark current noise A dark portion detection step of detecting the region of the digital still image as a dark portion of the digital still image, and when the detected dark portion includes the one pixel, A setting step of setting a filter coefficient to be multiplied by the detected difference value for each of the other pixels based on the position of the detected dark portion in the digital still image, and the specified other A multiplication step of multiplying the detected difference value by the set filter coefficient for each of the pixels, and an addition step of adding the multiplication result in the multiplication step to the signal of the one pixel and outputting the result. Detection processing in the detection step, comparison processing in the comparison step, identification processing in the specific step, multiplication processing in the multiplication step using the set filter coefficient, and addition processing in the addition step. A control step of repeating each of the pixels forming the dark portion with each of the pixels to be formed as the one pixel; Including the.

In order to solve the above-described problem, the invention described in claim 5 is directed to a computer included in a noise removal apparatus for removing noise included in a digital still image formed by a plurality of pixels. Detecting means for detecting a difference value of a signal between each of at least one other pixel around the pixel and the one pixel, and comparing each detected difference value with a threshold value Comparing means, specifying means for specifying the other pixels to be subjected to noise removal processing based on a comparison result in the computer functioning as the comparing means, and the digital still image including dark current noise A dark part detecting means for detecting an area as a dark part of the digital still image, and the other specified when the detected dark part includes the one pixel. Setting means for setting a filter coefficient to be multiplied to the detected difference value of each pixel for each of the other pixels based on the position of the detected dark portion in the digital still image, the specified other Multiplication means for multiplying the detected difference value by the set filter coefficient, and a multiplication result by the computer functioning as the multiplication means is added to the signal of the one pixel and output. Using the computer that functions as the detecting means, the comparing process by the computer that functions as the comparing means, the specifying process by the computer that functions as the specifying means, and the set filter coefficient Multiplication processing by the computer functioning as the multiplication means An addition process by the computer acting as a fine said adding means, as the one pixel of each of the pixels forming the dark portion, control means for repeating for each of said pixels forming the dark portion, to function as a.

According to the invention described in any one of claims 1, 4, and 5 , a region of a digital still image including dark current noise is detected as the dark portion, and the one included in the dark portion is detected. Since the filter coefficient to be multiplied by the difference value of the signal corresponding to the other pixels around the pixel is set for the other pixel based on the position of the dark portion in the digital still image, for each digital still image including the dark portion By setting the optimum filter coefficient for the digital still image, even if it is a digital still image including a dark part, highly versatile and effective noise removal processing can be performed on the digital still image.

In order to solve the above problem, the invention according to claim 2 is the noise removal device according to claim 1, and corresponds to each of the other pixels included in the dark part in one digital still image. At least a filter coefficient table including the filter coefficient and a filter coefficient table including the filter coefficient corresponding to each of the other pixels included in a region other than the dark portion in the one digital still image. Storage means such as a table memory is further provided, and the setting means switches the stored filter coefficient table based on the detected dark part and outputs it to the multiplication means, and the multiplication means outputs the output The multiplication process is performed using the filter coefficient included in the filter coefficient table.

According to the invention described in claim 2 , in addition to the operation of the invention described in claim 1, the filter coefficient table including the filter coefficient corresponding to each of the other pixels included in the dark part, and the area other than the dark part At least a filter coefficient table including filter coefficients corresponding to each of the other pixels included, and switching them based on the dark portion of the digital still image for use in multiplication processing. The coefficient can be changed.

In order to solve the above problems, the invention according to claim 3, in the noise removing device according to claim 1 or claim 2, wherein the other pixel, for the one pixel in said digital still image The pixels at the symmetric positions, and the filter coefficients are configured to be the same at the other pixels at the symmetric positions.

According to the invention described in claim 3 , in addition to the operation of the invention described in claim 1 or 2 , the other image is a pixel in a symmetrical position with respect to one pixel in the digital still image, Since the filter coefficients are the same in other pixels at symmetrical positions, the multiplication means and the addition means can be simplified.

In order to solve the above problem, according to the invention described in claim 6 , the noise removal program according to claim 5 is recorded so as to be readable by the computer.

Therefore, by causing the computer to read out and execute the noise removal program, a digital still image area including dark current noise is detected as the dark part, and the area around one pixel included in the dark part is detected. the filter coefficients to be multiplied to the difference value of the signal corresponding to some other pixels, since the computer functions so as to set the the other pixels on the basis of the dark part of the position in the digital still image, digital still containing dark portion By setting an optimal filter coefficient for each image, even for a digital still image including a dark part, a highly versatile and effective noise removal process can be performed on the digital still image.

According to the present invention, to detect the dark part of the digital still images, the filter coefficients to be multiplied to the difference value of the signals corresponding to other pixels surrounding one of the pixels included in the dark portion, the dark portion in the digital still image The other pixels are set on the basis of the positions .
Therefore , by setting an optimal filter coefficient for each digital still image including a dark part, even if a digital still image including a dark current area including a dark current noise is included, one unnecessary and quality wear-specific modifications, can be a versatile and effective noise removal process performed on the digital still images.

It is a figure explaining the principle of this invention, (a) is a figure which shows the filter object area | region which concerns on this invention, (b) is a figure which shows the 1st example of distribution of the filter coefficient which concerns on this invention, (C) is a figure which shows the 2nd example of distribution of the filter coefficient which concerns on this invention, (d) is a figure which shows the 3rd example of distribution of the filter coefficient concerning this invention, (e) is this invention FIG. 10 is a diagram illustrating a fourth example of the distribution of filter coefficients according to the present invention, FIG. 10F is a diagram illustrating a fifth example of the distribution of filter coefficients according to the present invention, and FIG. It is a figure which shows the 6th example of. It is a figure which shows the noise removal filter which concerns on 1st Embodiment, (a) is a block diagram which shows the schematic structure of the said noise removal filter, (b) is a flowchart which shows the noise removal process in the said noise removal filter. . It is a figure which shows the noise removal process which concerns on 1st Embodiment. It is a figure which shows the noise removal filter which concerns on 2nd Embodiment, (a) is a block diagram which shows the schematic structure of the said noise removal filter, (b) is a flowchart which shows the noise removal process in the said noise removal filter. . It is a figure explaining the dark part determination process which concerns on 2nd Embodiment thru | or 4th Embodiment, (a) is a figure which illustrates the image which is the object of dark part determination, (b) is the result of dark part determination. It is a figure illustrated. It is a figure which shows the noise removal filter which concerns on 3rd Embodiment, (a) is a block diagram which shows the schematic structure of the said noise removal filter, (b) is a flowchart which shows the noise removal process in the said noise removal filter. . It is a figure which shows the noise removal filter which concerns on 4th Embodiment, (a) is a block diagram which shows the schematic structure of the said noise removal filter, (b) is a flowchart which shows the noise removal process in the said noise removal filter. . It is a figure explaining the area | region determination process which concerns on 5th Embodiment and 6th Embodiment, (a) is a figure which illustrates the image which is the object of area | region determination, (b) is the 1st of the result of area | region determination. It is a figure which shows an example, (c) is a figure which shows the 2nd example of the result of area | region determination. It is a figure which shows the noise removal filter which concerns on 5th Embodiment, (a) is a block diagram which shows the schematic structure of the said noise removal filter, (b) is a flowchart which shows the noise removal process in the said noise removal filter. . It is a block diagram which shows the schematic structure of the noise removal filter which concerns on 6th Embodiment. It is a figure explaining the dark current noise detection process which concerns on 7th Embodiment and 8th Embodiment, (a) is a figure which shows the 1st example of the image which is the object of dark current noise detection, (b) is It is a figure which shows the 2nd example of the image which is the said object, (c) is a figure which illustrates the result of a dark current noise detection. It is a figure which shows the noise removal filter which concerns on 7th Embodiment, (a) is a block diagram which shows the schematic structure of the said noise removal filter, (b) is a flowchart which shows the noise removal process in the said noise removal filter. . It is a block diagram which shows the schematic structure of the noise removal filter which concerns on 8th Embodiment. It is a block diagram which shows the schematic structure of the noise removal filter which concerns on 9th Embodiment. It is a block diagram which shows the schematic structure of the noise removal filter which concerns on 10th Embodiment.

  Next, modes for carrying out the present invention will be described based on the drawings. The embodiment described below is an embodiment when the present invention is applied to a noise removal filter that removes noise and distortion included in a digital image composed of a plurality of pixels. Moreover, in the following description, the same name etc. are used about the concept etc. which were demonstrated in the background art mentioned above.

(I) Principle of the Present Invention First, before describing specific embodiments, the basic principle of the present invention will be described with reference to FIG. FIG. 1 is a diagram for explaining the principle of the present invention.

  Similar to the above background art, also in the present invention, a square area in a digital image input as an image signal is set as a filter target area. For example, as illustrated in FIG. 1A, a filter target area F including a pixel Px of 5 pixels × 5 pixels is set for each central pixel, with each pixel included in the digital image as an image signal as a central pixel. . In FIG. 1A, a total of 25 pixels are shown as pixels Px m, n (m and n are natural numbers of 5 or less), respectively. In the filter target region F illustrated in FIG. 1A, the pixel Px33 is the central pixel. Then, the luminance of each pixel Px is acquired from each pixel Px in the filter target region F.

  As described above, in the noise removal processing in the background art, the filter coefficients corresponding to the respective pixels Px in the filter target region F are all the same as illustrated in FIG. 1B (the above processing (5 )reference). In FIGS. 1B to 1F, the luminance of each pixel Px is indicated by a number in ().

  In contrast to the background art as described above, in the present invention, in addition to the background art illustrated in FIG. 1B, each pixel Px is provided as illustrated in FIGS. 1C to 1F. The corresponding filter coefficient can be set (changed) for each pixel Px. Specifically, for example, in the example of the filter target region F shown in FIG. 1C, the value of the filter coefficient corresponding to the center pixel Px33 is set to, for example, “16”, and four adjacent pixels in the vertical and horizontal directions with respect to the center pixel Px33. The filter coefficients corresponding to the pixel Px23, the pixel Px32, the pixel Px34, and the pixel Px43 are, for example, “8”, and the eight pixels Px13, the pixel Px22, the pixel Px24, the pixel Px31, and the pixel Px35 further outside the center pixel Px33. The filter coefficients respectively corresponding to the pixel Px42, the pixel Px44, and the pixel Px53 are “4”, for example. Further, the filter coefficients corresponding to the pixel Px12, the pixel Px14, the pixel Px21, the pixel Px25, the pixel Px41, the pixel Px45, the pixel Px52, and the pixel Px54 are set to “2”, for example, and the pixel Px11, the pixel Px15, the pixel Px51, and the pixel Px55 The corresponding filter coefficient is, for example, “1”. Similarly, as illustrated in FIGS. 1D to 1F, the filter coefficient corresponding to each pixel Px constituting the filter target region F is set for each pixel Px. Then, combinations of 25 filter coefficients illustrated in FIGS. 1B to 1F are stored in advance in the noise removal filter as “filter coefficient tables”, and these filter coefficient tables are stored in advance. Depending on the content of the digital image, it can be switched manually or automatically.

  Thereby, for example, in the case of the filter coefficient table according to the background art illustrated in FIG. 1B, the weight of the filter coefficient in the central pixel Px33 is about 0.04 as described above, but in FIG. In the case of the filter coefficient table exemplified, the weight of the filter coefficient is 16 ÷ (16 + 8 × 4 + 4 × 8 + 2 × 8 + 1 × 4) = 0.16. In the case of the filter coefficient table exemplified in FIG. The weight of the filter coefficient is 4 ÷ (4 + 2 × 4 + 1 × 20) = 0.125. Further, in the case of the filter coefficient table illustrated in FIG. 1E, the weight of the filter coefficient is 4 ÷ (4 + 2 × 4 + 1 × 8 + 0 × 12) = 0.2, and the filter coefficient table illustrated in FIG. In this case, the weight of the filter coefficient is 1 ÷ (1 + 2 × 8 + 4 × 16) ≈0.012. In the cases illustrated in FIGS. 1B to 1F, the ranges (sizes) of the filter target regions F are all the same. That is, in the filter coefficient tables illustrated in FIGS. 1B to 1E, the noise removal strength as the noise removal filter itself is shown in FIG. The filter coefficient table illustrated in FIG. 1B becomes weaker in the order of the filter coefficient table illustrated in FIG. 1C, the filter coefficient table illustrated in FIG. 1D, and the filter coefficient table illustrated in FIG. ing.

  As described above, according to the present invention, the weight of the filter coefficient in the center pixel Px33 as each filter coefficient table, that is, the noise in the center pixel Px33 as the noise removal filter itself, without changing the range of the filter target area F at all. The removal intensity can be changed. Then, by switching the filter coefficient table exemplified in FIGS. 1B to 1F manually or automatically according to the contents of the digital image, for example, the clarity of the contour of the image component in the digital image can be improved. The noise removal processing can be effectively performed while freely controlling.

  Note that the filter coefficient values in each filter coefficient table may have the same filter coefficient for pixels located symmetrically with respect to the center pixel Px33 in the filter target area F, as illustrated in FIG. In the case illustrated in FIG. 1G, the same filter coefficient is set for the pixel Px23, the pixel Px32, the pixel Px34, and the pixel Px43, which are in a symmetric position with respect to the central pixel Px33. Similarly, the pixel Px22, the pixel Px24, the pixel Px42, and the pixel Px44 have the same filter coefficient, and the pixel Px13, the pixel Px31, the pixel Px35, and the pixel Px53 have the same filter coefficient, and the pixel Px12, pixel Px14, pixel Px52, and pixel Px54 have the same filter coefficient, and pixel Px21, pixel Px25, pixel Px41, and pixel Px45 have the same filter coefficient, and pixel Px11, pixel Px15, pixel Px51, and pixel Px55. The filter coefficients are the same. By using a filter coefficient table including filter coefficients having such values, the hardware scale as a noise removal filter can be greatly reduced as will be described later.

(II) First Embodiment Next, a first embodiment according to the present invention based on the above-described principle will be described with reference to FIGS. 2 is a diagram illustrating the noise removal filter according to the first embodiment, and FIG. 3 is a diagram illustrating the noise removal processing according to the first embodiment. Further, it is assumed that the filter target area F in each of the following embodiments is a filter target area F including 25 pixels Px of 5 pixels × 5 pixels illustrated in FIG. Further, in FIG. 3, for example, the luminance of the pixel Px11 is expressed as “P ₁₁ ”.

  As shown in FIG. 2A, the noise removal filter S1 according to the first embodiment includes a memory unit 1, a difference calculation unit 2 as an example of a detection unit, an example of a comparison unit, and an example of a specification unit. An example of a threshold determination unit 3, an example of a multiplication unit, an operation unit 4 as an example of an addition unit and an example of a control unit, a noise threshold setting unit 5, a filter setting unit 6 as an example of a setting unit, and an example of a selection unit And an operation unit 7. The filter setting unit 6 includes a table memory 60 as an example of a storage unit and a filter coefficient setting unit 61. At this time, for example, the filter coefficient tables T1 to Tn (n is a natural number) illustrated in FIGS. 1B to 1F are stored in the table memory 60 in a nonvolatile manner.

  In this configuration, the memory unit 1 receives an image signal Sin corresponding to a digital image including the filter target area F, as shown in FIGS. The memory unit 1 acquires the luminance of each pixel Px (including the center pixel Px33) constituting the filter target area F in the digital image from the image signal Sin and temporarily stores it. These processes correspond to the processes described in (1) and (2), respectively (step S1 in FIG. 2B). Then, the memory unit 1 outputs the stored luminance information indicating the luminance of each pixel Px to the difference calculation unit 2 as the memory signal Sm.

Next, the difference calculation unit 2 acquires the luminance information indicating the luminance of each pixel Px in the filter target region F including the central pixel Px33 from the memory unit 1, and executes the process corresponding to (3) above. That is, as schematically shown in FIG. 3, the difference calculation unit 2 detects the difference Δ between the luminance of the central pixel Px33 and the luminance P of each pixel Px for each pixel Px, and the detection result is represented by the pixel Px. And output to the threshold value determination unit 3 as the difference signal Sdif. Difference computation section 2 for example, the pixel Px11 includes a luminance P _11, the luminance P ₃₃ of the center pixel Px33, it calculates the difference delta _11, and outputs the threshold decision unit 3 as a differential signal Sdif. Difference computation section 2 for another example pixel Px44 includes a luminance P _44, the luminance P ₃₃ of the center pixel Px33, it calculates the difference delta _44, and outputs the threshold decision unit 3 as a differential signal Sdif. The difference calculation unit 2 executes the above processing for all the pixels Px in the filter target region F as shown in FIG.

On the other hand, the user grasps the contents of the digital image, for example, by displaying a digital image corresponding to the image signal Sin before being input to the memory unit 1 on a display (not shown). The operation unit 7 executes an operation of arbitrarily selecting the filter coefficient table T (see FIGS. 1B to 1F) based on the above. An operation signal Soc corresponding to this operation is output from the operation unit 7 to the filter coefficient setting unit 61 of the filter setting unit 6. Accordingly, the filter coefficient setting unit 61 selects the filter coefficient selected by the user at that time from the filter coefficient table T stored in the table memory 60 based on the operation signal Soc corresponding to the user's operation. The table T is set, and the filter coefficient included in the set filter coefficient table T is output to the calculation unit 4 for each pixel Px as the filter coefficient signal Sc (step S2 in FIG. 2B). In FIG. 3, for example, the filter coefficient corresponding to the pixel Px11 is represented as “C ₁₁ ”.

  In parallel with this, for example, based on the content of the digital image displayed on the display, the user performs an operation on the operation unit 7 to set a noise threshold value related to the noise determination process of (4). An operation signal Sot corresponding to this operation is output from the operation unit 7 to the noise threshold setting unit 5. Thereby, the noise threshold value setting unit 5 sets a noise threshold value related to the noise determination processing corresponding to the above (4) based on the operation signal Sot, and outputs the noise threshold value signal Sth to the threshold value determination unit 3. Thereby, the threshold determination unit 3 executes the noise determination process corresponding to the above (4) (step S3 in FIG. 2B). That is, the threshold value determination unit 3 compares the difference Δ for each pixel output as the difference signal Sdif with the threshold value input as the threshold signal Sth, for example, the pixel Px corresponding to the difference Δ of a value equal to or greater than the threshold value. Are excluded from the target pixels of the noise removal processing according to the first embodiment (FIG. 2B, step S3; non-noise). On the other hand, the threshold determination unit 3 sets, for example, the pixel Px corresponding to the difference Δ between the values less than the threshold as the target pixel of the noise removal processing according to the first embodiment (FIG. 2 (b), step S3; noise). Then, the threshold determination unit 3 adds a flag or the like indicating that the pixel Px is the target pixel, and calculates a difference Δ corresponding to each pixel Px as a pixel signal Spx for each pixel Px. 4 is output.

  Next, the calculation unit 4 uses the filter coefficient C output from the filter coefficient setting unit 61 as the filter coefficient signal Sc and the difference Δ output from the threshold determination unit 3 as the pixel signal Spx, and A multiplication / addition process corresponding to (5) and an addition process corresponding to (6) above are executed (step S4 in FIG. 2B). That is, the calculation unit 4 first applies to all other pixels Px other than the pixel Px to which a flag indicating the target pixel is added, out of the difference Δ output from the threshold determination unit 3 as the pixel signal Spx. One value of the corresponding difference Δ or the filter coefficient C corresponding to all the other pixels Px is set to zero. Next, as indicated by a broken line in FIG. 3, the calculation unit 4 multiplies the difference Δ corresponding to each pixel Px and the filter coefficient C for each pixel Px, and adds all the multiplication results for each pixel Px. At this time, unlike the background art, the calculation unit 4 is included in the filter coefficient C (that is, the filter coefficient table T selected by the user for each digital image) output from the filter coefficient setting unit 61 as the filter coefficient signal Sc. Multiplication processing is performed using the filter coefficient C). In this case, for all the other pixels Px other than the pixel Px that is the target pixel for the noise removal process, either the value Δ of the difference Δ or the filter coefficient C corresponding to the other pixel Px is set to zero. Therefore, as a result, the multiplication result is significant only for the pixel Px that is the target pixel of the noise removal processing.

  The arithmetic unit 4 adds the luminance of the center pixel Px33 to the addition result for all the pixels Px, as shown by a one-dot chain line in FIG. 3, and (in other words, the center pixel Px33 for the center pixel Px33). Is the output signal Sout after the noise removal processing according to the first embodiment (for the filter target region F centered at).

  Thereafter, the calculation unit 4 performs the processing from step S1 to step S4 shown in FIG. 2B with all the pixels Px in the digital image as the central pixel (in other words, for all the filter target regions F in the digital image). It is confirmed whether it has been performed or not (FIG. 2 (b), step S5). When the processing from step S1 to step S4 shown in FIG. 2B has been completed for all the pixels Px in the confirmation in step S5 (FIG. 2B, step S5; YES), the calculation unit 4 calculates the image signal Sin. Assuming that the noise removal processing is completed for all the pixels in the digital image input as, the output signal Sout for all the filter target regions F is output corresponding to the digital image including the filter target region F. The signal Sout is output, and the noise removal processing according to the first embodiment is finished.

  On the other hand, if there is a pixel Px in which the processing from step S1 to step S4 shown in FIG. 2B is not completed in the confirmation in step S5 (FIG. 2B, step S5; NO), the calculation unit 4 completes the processing. In order to execute the processing from step S1 to step S4 shown in FIG. 2B with the new pixel Px not being the center pixel, the processing returns to step S1 (step S6).

  Note that the processing in steps S1 to S6 described above is processing for one filter target region F. After the processing of step S1 to step S6 has been completed for all of the one filter target region F, the pixel Px adjacent to the center pixel Px33 (specifically, the pixel Px22, the pixel Px23 illustrated in FIG. A new filter target region F having the pixel Px24, the pixel Px32, the pixel Px34, the pixel Px42, the pixel Px43, or the pixel Px44) as the central pixel is set in the memory unit 1 and the difference calculation unit 2, and the new filter target The processing from step S1 to step S6 is performed again on the area F. Then, the processing of steps S1 to S6 is completed for all the filter target regions F that can be set in one digital image, so that the noise removal processing according to the first embodiment for the one digital image is completed. To do. At this time, the filter coefficient table T according to the first embodiment can be configured to be switched for each filter target region F, for example, or configured to be switched for each of a plurality of continuous filter target regions F, for example. You can also

  As described above, according to the operation of the noise removal filter S1 of the first embodiment, the filter coefficient C multiplied by the difference Δ in the noise removal processing for the filter target region F included in the digital image is set to the content of the digital image. Therefore, by setting the optimum filter coefficient C for each digital image, highly versatile and effective noise removal processing can be performed on the digital image.

  In addition, since the filter coefficient C can be set by the user's operation, an effective noise removal process according to the user's preference or the like can be performed on the digital image.

  Further, a plurality of types of filter coefficient tables T are stored, and these are switched based on the contents of the digital image and used for the multiplication process. Therefore, the filter coefficient C can be changed with a simple configuration.

Note that, in the multiplication process corresponding to the above (5) executed in the calculation unit 4 according to the first embodiment, the difference Δ corresponding to each pixel Px and the filter coefficient C are multiplied for each pixel Px (FIG. 3). For example, when the filter coefficient C is considered for each pixel Px as illustrated in FIGS. 1B to 1F, the number of pixels Px constituting the filter target region F (in the case illustrated in FIG. 1) 25 multipliers are required in the arithmetic unit 4. However, as illustrated in FIG. 1G, in the case where the pixels Px that are symmetrical with respect to the center pixel Px33 in the filter target region F are set to the same filter coefficient C, pixels corresponding to the same filter coefficient C are used. If all the differences Δ for Px are added and then multiplied by the same filter coefficient C, the result is the same as when multiplying each pixel Px one by one. Therefore, when illustrated in FIG. 1 (g), since the filter coefficients C can put a filter coefficient C ₃₃ corresponding to the center pixel Px33 it is the sum seven multiplier requires only seven. Furthermore, since the difference delta ₃₃ corresponding to the center pixel Px33 is naturally is zero, considering this, the multiplier in the calculation unit 4 will be able to reduce further up to six. As described above, for example, when the filter coefficient table T in the mode illustrated in FIG. 1G is used, the filter coefficient C is the same in the other pixels Px at positions symmetrical to the center pixel Px33. Therefore, the hardware scale required for the noise removal filter S1 (specifically, the number of multipliers) can be greatly reduced.

(III) Second Embodiment Next, a second embodiment, which is another embodiment according to the present invention, will be described with reference to FIGS. FIG. 4 is a diagram illustrating a noise removal filter according to the second embodiment, and FIG. 5 is a diagram illustrating dark part determination processing according to the second embodiment and third and fourth embodiments described later. In describing the noise removal filter according to the second embodiment, the same member number and step number are used for the same member and process as those in the noise removal filter S1 according to the first embodiment. Detailed description is omitted.

  In the noise removal filter S1 according to the first embodiment, with respect to the filter coefficient C corresponding to each pixel Px, the optimal filter coefficient for each filter target region F while the user actually sees the digital image including the pixel Px. The table T is set. On the other hand, in the second embodiment and third and fourth embodiments described later, the difference corresponding to each pixel Px in consideration of the type of noise included in the digital image and the characteristics corresponding to the type. A filter coefficient C to be multiplied by Δ is set.

  That is, as described in the background art above, noise included in the image signal Sin includes noise that is included in the digital image mainly due to the structure of the image sensor, and block coding processing for digitization. There is block noise due to being applied. In this case, the former is typically what is called dark current noise. In the following description, the filter target region F having the pixel Px including dark current noise as the central pixel is referred to as a “dark portion” in the digital image. In the second to fourth embodiments described below, more effective noise removal processing is performed by switching the filter coefficient C in consideration of the characteristics of each of these noises.

As shown in FIG. 4A, the noise removal filter S2 according to the second embodiment is different from the noise removal filter S1 according to the first embodiment in the dark current among the pixels Px in the filter target region F. a dark part determining section 10 as an example of a dark portion detecting means for detecting a pixel Px noisy, the dark area threshold value serving as the threshold value in detecting the dark part determining section 10 pixels Px, including the dark current noise, in the operation unit 7 And a dark part threshold value setting unit 11 that is set based on a user's operation.

  In this configuration, in addition to the noise threshold setting operation similar to that of the noise removal filter S1 according to the first embodiment, the user sets the dark portion threshold based on, for example, the contents of the digital image displayed on the display. The operation to be performed is executed in the operation unit 7. An operation signal Sod corresponding to this operation is output from the operation unit 7 to the dark part threshold value setting unit 11. Thereby, the dark part threshold value setting part 11 sets the said dark part threshold value based on the said operation signal Sod, and outputs it to the dark part determination part 10. FIG. Based on the luminance information output as the memory signal Sm from the memory unit 1, the dark part determination unit 10 uses the dark part threshold set by the dark part threshold setting unit 11 in the digital image (in other words, in the filter target region F). A dark part having a dark luminance as a central pixel is detected from a pixel Px that seems to contain dark current noise, and a dark part signal Sdc indicating a position of the detected central part in the digital image of the central pixel, for example, is detected. It generates and outputs to the filter coefficient setting unit 61 (FIG. 4B, step S10). Note that the operation signal Soc output from the operation unit 7 to the filter coefficient setting unit 61 in the noise removal filter S1 according to the first embodiment is not output from the noise removal filter S2 according to the second embodiment.

  Here, the dark part determination processing in the dark part determination unit 10 will be described with reference to FIG. For example, when a digital image G as illustrated in FIG. 5A is input as the image signal Sin, the dark part determination unit 10 to which the corresponding memory signal Sm is input detects the dark part in the digital image G. In FIG. 5B, a plurality of dark portions D are detected in the lower left portion, the lower side portion, and the lower right portion in the digital image G. Then, the dark part determination unit 10 generates a dark part signal Sdc indicating, for example, the position of the central pixel in the digital image G in the detected dark part, and outputs the dark part signal Sdc to the filter coefficient setting unit 61.

  Accordingly, the filter coefficient setting unit 61 sets one of the filter coefficient tables T from the filter coefficient table T stored in the table memory 60 based on the dark part signal Sdc, and the set filter coefficient table T Is output to the calculation unit 4 for each pixel Px as a filter coefficient signal Sc (FIG. 4B, steps S11 and S12). Specifically, the filter coefficient setting unit 61 has, for example, a filter coefficient table Tm set in advance as a filter coefficient table T to be associated with a dark part, and a filter target area F other than the dark part (in other words, a bright part is determined). When the filter coefficient table Tn is preset as the filter coefficient table T to be associated with the filter target area F), the filter coefficient table Tm for the filter target area F determined to be a dark part by the dark part signal Sdc. Is output as a filter coefficient signal Sc to the calculation unit 4 (step S11 in FIG. 4B). Further, the filter coefficient setting unit 61 is considered to include a filter target region F other than the filter target region F that is determined to be a dark part by the dark part signal Sdc (in other words, block noise other than dark current noise is included). For the filter target region F), the filter coefficient C included in the filter coefficient table Tn is output to the calculation unit 4 as the filter coefficient signal Sc (FIG. 4B, step S12).

  Thus, the calculation unit 4 executes the multiplication / addition process corresponding to (5) above and the addition process corresponding to (6) above based on the output filter coefficient signal Sc (step (b) in FIG. 4). S4).

  As described above, according to the operation of the noise removal filter S2 according to the second embodiment, in addition to the effect by the operation of the noise removal filter S1 according to the first embodiment, a dark part included in the digital image G is detected. Since the filter coefficient C is set based on the detected dark part and the content of the digital image G, more effective noise removal processing is performed in consideration of the characteristics of dark current noise and block noise. Can do.

(IV) Third Embodiment Next, a third embodiment which is another embodiment according to the present invention will be described with reference to FIG. FIG. 6 is a diagram illustrating a noise removal filter according to the third embodiment. In the description of the noise removal filter according to the third embodiment, the same member numbers and step numbers are used for the same members and processes as those in the noise removal filter S2 according to the second embodiment. Detailed description is omitted.

  In the noise removal filter S2 according to the second embodiment, the filter coefficient C corresponding to each pixel Px is switched according to the determination result of whether or not it is a dark part in the dark part determination unit 10. On the other hand, in the noise removal filter according to the third embodiment, the noise threshold used for the noise determination process in the threshold determination unit 3 is switched according to the determination result of the dark part determination unit 10 as to whether it is a dark part. Note that the filter coefficient C corresponding to each pixel Px in the noise removal filter S3 according to the third embodiment is switched by the same configuration as the noise removal filter S1 according to the first embodiment.

  That is, as illustrated in FIG. 6A, the noise removal filter S <b> 3 according to the third embodiment is the same as the noise removal filter S <b> 2 according to the second embodiment except that the threshold includes a noise threshold setting unit 5 and a threshold memory 51. A setting unit 50 is provided. At this time, the threshold memory 51 stores a plurality of noise thresholds H1 to Hn (n is a natural number), which are different noise thresholds, in a nonvolatile manner. The dark signal Sdc in the noise removal filter S2 according to the second embodiment is output to the noise threshold setting unit 5 instead of the filter coefficient setting unit 61 in the noise removal filter S3 according to the third embodiment.

  In this configuration, in addition to the selection operation of the filter coefficient table T similar to the noise removal filter S1 according to the first embodiment, the user operates the dark part threshold value setting operation similar to the noise removal filter S2 according to the second embodiment. This is executed in part 7. The dark part determination unit 10 performs dark part determination processing similar to the dark part determination process according to the second embodiment, generates the dark part signal Sdc, and outputs the dark part signal Sdc to the noise threshold setting unit 5 (step S10 in FIG. 6B). .

  Thereby, the noise threshold value setting unit 5 sets one of the noise threshold values H from the noise threshold value H stored in the threshold value memory 51 based on the dark part signal Sdc, and sets the set noise threshold value H as the noise threshold value. The signal Sh is output to the threshold value determination unit 3 for each dark part (FIG. 6B, steps S15 and S16). Specifically, for example, the noise threshold value setting unit 5 has a mth noise threshold value Hm set in advance as the noise threshold value H to be associated with the dark portion, and the noise threshold value H to be associated with the filter target region F other than the dark portion. When the n noise threshold value Hn is set in advance, the mth noise threshold value Hm is output to the threshold value determination unit 3 as the noise threshold signal Sh for the filter target region F that is determined to be a dark part by the dark part signal Sdc (FIG. 6). (B) Step S15). The noise threshold value setting unit 5 outputs the nth noise threshold value Hn as the noise threshold signal Sh to the threshold value determination unit 3 for the filter target region F other than the filter target region F that is determined to be a dark part by the dark part signal Sdc. (FIG. 6B, step S16).

  Accordingly, the threshold determination unit 3 executes the noise determination process corresponding to the above (4) for each filter target region F based on the output noise threshold H (step S4 in FIG. 6B).

  As described above, according to the operation of the noise removal filter S3 according to the third embodiment, in addition to the effect by the operation of the noise removal filter S1 according to the first embodiment, a dark part included in the digital image G is detected. Since the noise threshold H is set based on the detected dark part and the content of the digital image G, more effective noise removal processing can be performed.

(V) Fourth Embodiment Next, a fourth embodiment which is another embodiment according to the present invention will be described with reference to FIG. FIG. 7 is a diagram illustrating a noise removal filter according to the fourth embodiment. In the description of the noise removal filter according to the fourth embodiment, the same member numbers and step numbers are used for the same members and processes as those in the noise removal filter S3 according to the third embodiment. Detailed description is omitted.

  In the noise removal filter according to the fourth embodiment, the processing in each of the noise removal filter S2 according to the second embodiment and the noise removal filter S3 according to the third embodiment is combined, and the processing unit 4 corresponds to the above (5). Both the filter coefficient C used for the multiplication processing to be performed and the noise threshold used for the noise judgment processing corresponding to (4) in the threshold judgment unit 3 are switched according to the judgment result of the dark part judgment unit 10 as to whether it is a dark part. .

  That is, as shown in FIG. 7A, in the noise removal filter S4 according to the fourth embodiment, the dark part signal Sdc in the noise removal filter S3 according to the third embodiment is also output to the filter coefficient setting unit 61. . In the noise removal filter S4 according to the fourth embodiment, the operation signal Soc from the operation unit 7 to the filter coefficient setting unit 61 and the operation signal Sot to the noise threshold setting unit 5 are not output.

  In this configuration, the user performs only the dark part threshold value setting operation similar to that of the noise removal filter S2 according to the second embodiment on the operation unit 7. The dark part determination unit 10 performs dark part determination processing similar to the dark part determination processing according to the second embodiment, generates the dark part signal Sdc, and outputs the dark part signal Sdc to the noise threshold setting unit 5 and the filter coefficient setting unit 61 (see FIG. 7 (b) Step S10).

  Accordingly, the noise threshold setting unit 5 can select any one of the noise thresholds from the noise threshold H stored in the threshold memory 51 based on the dark part signal Sdc, similarly to the noise removal filter S3 according to the third embodiment. H is set, and the set noise threshold value H is output as a noise threshold signal Sh to the threshold value determination unit 3 for each dark portion (steps S15 and S16 in FIG. 7B). Further, the filter coefficient setting unit 61 selects any one of the filter coefficients from the filter coefficient table T stored in the table memory 60 based on the dark part signal Sdc, similarly to the noise removal filter S2 according to the second embodiment. The table T is set, and the filter coefficient C included in the set filter coefficient table T is output to the calculation unit 4 for each pixel Px as the filter coefficient signal Sc (FIG. 7B, steps S17 and S18).

  Thus, the threshold determination unit 3 executes the noise determination process corresponding to the above (4) for each filter target region F based on the output noise threshold signal Sh (step S4 in FIG. 7B). The computing unit 4 executes the multiplication / addition process corresponding to the above (5) and the addition process corresponding to the above (6) based on the output filter coefficient signal Sc (FIG. 7B, step S4). ).

  As described above, according to the operation of the noise removal filter S4 according to the fourth embodiment, in addition to the effect of the operation of the noise removal filter S1 according to the first embodiment, a dark part included in the digital image G is detected. Since the noise threshold value H and the filter coefficient C are set based on the detected dark part and the content of the digital image G, more effective noise removal processing can be performed.

  In the second embodiment and the fourth embodiment described above, the filter coefficient C used for the multiplication processing corresponding to the above (5) in the calculation unit 4 is the determination result of whether or not it is a dark part in the dark part determination unit 10. It was set as the structure switched according to. In the third embodiment, the noise threshold value used in the noise determination process corresponding to the above (4) in the threshold value determination unit 3 is switched according to the determination result of whether or not it is the dark part, and the above (5) The filter coefficient C used in the multiplication processing corresponding to) is switched by the same configuration as the noise removal filter S1 according to the first embodiment. However, other than this, for example, the noise removal processing according to the first to fourth embodiments is performed only on the filter target region F determined to be a dark part according to the determination result of whether or not the dark part determination unit 10 is a dark part. And the noise removal process is not performed on the other filter target areas F other than the filter target area F determined to be a dark part (more specifically, for the other filter target areas F, for example, the first implementation) It is also possible to configure so that the processing after the difference calculation unit 2 for the memory signal Sm in the noise removal filter S1 according to the embodiment is bypassed. In this case, there may be a processing time difference caused by the presence or absence of the noise removal process between the filter target area F determined to be a dark part and the other filter target areas F. For this reason, when such a processing time difference occurs, for example, the processing unit 4 needs to perform processing for canceling the processing time difference so that the digital image G is not partially lost or the like. Keep in mind.

(VI) Fifth Embodiment Next, a fifth embodiment which is another embodiment according to the present invention will be described with reference to FIGS. FIG. 8 is a diagram for explaining region determination processing according to the fifth embodiment and a sixth embodiment described later, and FIG. 9 is a diagram illustrating a noise removal filter according to the fifth embodiment. In the description of the noise removal filter according to the fifth embodiment, the same member numbers and step numbers are used for the same members and processes as those in the noise removal filter S1 according to the first embodiment. Detailed description is omitted.

  In the noise removal filter S1 according to the first embodiment, with respect to the filter coefficient C corresponding to each pixel Px, the optimal filter coefficient for each filter target region F while the user actually sees the digital image including the pixel Px. The table T is set. On the other hand, in the fifth embodiment and the sixth embodiment to be described later, a filter coefficient C for multiplying the difference Δ corresponding to each pixel Px is set for each region in the digital image.

  In other words, the filter coefficient C according to the present invention can be applied to the entire digital image G with respect to the digital image G illustrated in FIG. On the other hand, in the fifth embodiment and the sixth embodiment described later, as illustrated in FIG. 8C, the digital image G is divided into, for example, eight image areas Ga to Gh, and the divided image areas Ga are divided. The filter coefficient C is changed and set every time. In the fifth and sixth embodiments described below, more effective noise removal processing is performed by switching the filter coefficient C and the like in consideration of the image area Ga and the like in the digital image G.

  As shown in FIG. 9A, the noise removal filter S5 according to the fifth embodiment is configured to determine an image region Ga or the like in the digital image G with respect to the noise removal filter S1 according to the first embodiment. As an example, a region determination unit 20 is provided.

  In this configuration, in addition to the noise threshold setting operation similar to that of the noise removal filter S1 according to the first embodiment, the user can use the image area Ga or the like based on the content of the digital image displayed on the display, for example. The operation unit 7 executes an operation for setting the boundary of the. An operation signal Soa corresponding to this operation is output from the operation unit 7 to the region determination unit 20. Thereby, the region determination unit 20 generates a region signal Ssc for identifying the region Ga and the like in the digital image G based on the operation signal Soa and outputs the region signal Ssc to the filter coefficient setting unit 61 (FIG. 9B). Step S20). Note that the operation signal Soc output from the operation unit 7 to the filter coefficient setting unit 61 in the noise removal filter S1 according to the first embodiment is not output from the noise removal filter S5 according to the fifth embodiment.

  Accordingly, the filter coefficient setting unit 61 selects one of the filter coefficient tables T associated with the image area Ga or the like from the filter coefficient table T stored in the table memory 60 based on the area signal Ssc. The filter coefficient C included in the set filter coefficient table T is output to the calculation unit 4 for each pixel Px as the filter coefficient signal Sc (FIG. 9B, steps S11 and S12). Specifically, the filter coefficient setting unit 61 has a filter coefficient table Tm set in advance as a filter coefficient table T to be associated with, for example, the image area Ga illustrated in FIG. 8C, and also supports other image areas Gb. When the filter coefficient table Tn is set in advance as the filter coefficient table T to be used, when the image area Ga is indicated by the area signal Ssc, the filter coefficient for the filter target area F included in the image area Ga is set. The filter coefficient C included in the table Tm is output to the calculation unit 4 as the filter coefficient signal Sc (FIG. 9B, step S11). In addition, when the image signal Gsc is indicated by the region signal Ssc, the filter coefficient setting unit 61 uses the filter coefficient C included in the filter coefficient table Tn for the filter target region F included in the image region Gb. It outputs to the calculating part 4 as filter coefficient signal Sc (FIG.9 (b) step S12).

  Accordingly, the calculation unit 4 executes the multiplication / addition process corresponding to the above (5) and the addition process corresponding to the above (6) based on the output filter coefficient signal Sc (step (b) in FIG. 9). S4).

  As described above, according to the operation of the noise removal filter S5 according to the fifth embodiment, in addition to the effect of the operation of the noise removal filter S1 according to the first embodiment, the filter coefficient C is an image area in the digital image G. Since the filter coefficient C is set on the basis of the contents of the digital image G and the image area Ga and the like, the filter coefficient C is set in consideration of the characteristics of the image area Ga and the like of the digital image G. Effective noise removal processing can be performed.

(VII) Sixth Embodiment Next, a sixth embodiment which is another embodiment according to the present invention will be described with reference to FIG. FIG. 10 is a diagram illustrating a noise removal filter according to the sixth embodiment. In the description of the noise removal filter according to the sixth embodiment, the same member numbers and step numbers are used for the same members and processes as those in the noise removal filter S5 according to the fifth embodiment. Detailed description is omitted.

  In the noise removal filter S5 according to the fifth embodiment, the filter coefficient C corresponding to each pixel Px is switched according to the image region Ga determined by the region determination unit 20 or the like. In addition to this, in the noise removal filter according to the sixth embodiment, the noise threshold used for the noise determination processing in the threshold determination unit 3 is also switched according to the image region Ga determined by the region determination unit 20 and the like.

  That is, as illustrated in FIG. 10, the noise removal filter S6 according to the sixth embodiment is different from the noise removal filter S5 according to the fifth embodiment in that the threshold setting includes a noise threshold setting unit 5 and a threshold memory 51. Part 50 is provided. At this time, in the threshold memory 51, as in the case of the noise removal filter S4 according to the fourth embodiment shown in FIG. 7, a plurality of noise threshold values H1 to Hn (n is a natural number), which are different noise threshold values, are nonvolatile. Is remembered. The region signal Ssc in the noise removal filter S6 according to the sixth embodiment is output not only to the filter coefficient setting unit 61 but also to the noise threshold setting unit 5.

  In this configuration, the user performs an image area setting operation on the operation unit 7 similar to that of the noise removal filter S5 according to the fifth embodiment. Then, the region determination unit 20 performs the same processing as the region determination unit 20 according to the fifth embodiment, generates the region signal Ssc, and outputs the region signal Ssc to the noise threshold setting unit 5 and the filter coefficient setting unit 61.

  Thereby, the noise threshold value setting unit 5 sets one of the noise threshold values H from the noise threshold value H stored in the threshold value memory 51 based on the region signal Ssc, and sets the set noise threshold value H as the noise threshold value. The signal Sh is output to the threshold determination unit 3 according to the image area Ga and the like. Specifically, the noise threshold value setting unit 5 has an m-th noise threshold value Hm set in advance as a noise threshold value to be associated with, for example, the image region Ga illustrated in FIG. 8C, and is associated with another image region Gb. When the nth noise threshold value Hn is set in advance as the power noise threshold value and the image area Ga is indicated by the area signal Ssc, the mth noise threshold value for the filter target area F included in the image area Ga. Hm is output to the threshold determination unit 3 as a noise threshold signal Sh. Further, when the image signal region Gb is indicated by the region signal Ssc, the noise threshold value setting unit 5 determines the threshold value for the filter target region F included in the image region Gb with the nth noise threshold value Hn as the noise threshold signal Sh. Output to part 3. Further, the filter coefficient setting unit 61 selects any one of the filter coefficients from the filter coefficient table T stored in the table memory 60 based on the region signal Ssc, similarly to the noise removal filter S5 according to the fifth embodiment. The table T is set, and the filter coefficient C included in the set filter coefficient table T is output to the calculation unit 4 for each pixel Px as the filter coefficient signal Sc.

  Thus, the threshold determination unit 3 executes the noise determination process corresponding to the above (4) for each filter target region F based on the output noise threshold signal Sh. Similarly, the arithmetic unit 4 executes the multiplication / addition process corresponding to (5) above and the addition process corresponding to (6) above based on the output filter coefficient signal Sc.

  As described above, according to the operation of the noise removal filter S6 according to the sixth embodiment, in addition to the effect of the operation of the noise removal filter S5 according to the fifth embodiment, the contents of the digital image G and the image area concerned Since the filter coefficient C and the noise threshold are set based on Ga and the like, more effective noise removal processing can be performed in consideration of the characteristics of the image area Ga and the like of the digital image G.

(VIII) Seventh Embodiment Next, a seventh embodiment which is another embodiment according to the present invention will be described with reference to FIGS. FIG. 11 is a diagram for explaining dark current noise detection processing according to the seventh embodiment and an eighth embodiment to be described later, and FIG. 12 is a diagram illustrating a noise removal filter according to the seventh embodiment. Further, in describing the noise removal filter according to the seventh embodiment, the same member number and step number are used for the same member and process as those in the noise removal filter S1 according to the first embodiment. Detailed description is omitted.

  In the first to sixth embodiments described above, the filter coefficient C or the noise threshold H is set by, for example, visual evaluation of a user who actually viewed a digital image, but will be described below. In the seventh embodiment and the eighth to tenth embodiments described later, the filter coefficient C or the noise threshold H is automatically set based on the contents of the digital image.

  First, in a seventh embodiment described below and an eighth embodiment described later, a filter that multiplies the difference Δ corresponding to each pixel Px based on the state of the dark current noise included in the digital image. Coefficient C or noise threshold H is automatically set.

  That is, in general, dark current noise is generated without depending on one frame as a digital image. Therefore, if the digital image G of at least two frames is acquired and the difference is calculated, the portion of the digital image G including dark current noise (hereinafter, this portion is referred to as “dark current noise portion”). Can be detected. Specifically, for example, it is assumed that the digital image G1 for one frame illustrated in FIG. 11A and the digital image G2 for one frame illustrated in FIG. In FIGS. 11A and 11B, it is assumed that dark current noise is generated in the white spot portion. When the difference between the digital image G1 and the digital image G2 of these two frames is calculated, as shown in FIG. 11C, the dark current noise portion BN1 including dark current noise in the digital image G1 and the dark current in the digital image G2 A dark current noise portion BN2 including noise can be detected. And in 7th Embodiment and 8th Embodiment mentioned later, the filter coefficient C is changed and set according to the detection condition of this dark current noise part BN. In the seventh embodiment and the eighth embodiment described below, these dark current noises are detected and the filter coefficient C is switched, so that the filter coefficient C is automatically set appropriately and an effective noise removal process is performed. I do.

  As shown in FIG. 12A, the noise removal filter S7 according to the seventh embodiment detects dark current noise in the digital image G with respect to the noise removal filter S1 according to the first embodiment. A detection unit 30 is provided. The output signal Sout output from the calculation unit 4 is fed back to the dark current noise detection unit 30 in order to evaluate the occurrence of dark current noise in the digital image G (after noise removal processing) included in the output signal Sout. Yes.

  In this configuration, in addition to the noise threshold setting operation similar to that of the noise removal filter S1 according to the first embodiment, the user can count the number of dark current noise portions BN allowed in one digital image G as described later. (Hereinafter, the number of dark current noise portions BN is referred to as “allowable number”) is performed on the operation unit 7. An operation signal Son corresponding to this operation is output from the operation unit 7 to the dark current noise detection unit 30. Thereby, the dark current noise detection unit 30 generates a table designation signal Snc for designating the filter coefficient table T based on the operation signal Son, and outputs the table designation signal Snc to the filter coefficient setting unit 61 (FIG. 12B, step S30). ). Note that the operation signal Soc output from the operation unit 7 to the filter coefficient setting unit 61 in the noise removal filter S1 according to the first embodiment is not output from the noise removal filter S7 according to the seventh embodiment.

  Thus, the filter coefficient setting unit 61 selects and sets the filter coefficient table T indicated by the table specifying signal Snc from the filter coefficient table T stored in the table memory 60, and the set filter coefficient table The filter coefficient C included in T is output to the calculation unit 4 for each pixel Px as the filter coefficient signal Sc (FIG. 12B, step S30).

  Thereby, the calculation unit 4 executes the multiplication / addition process corresponding to the above (5) and the addition process corresponding to the above (6) based on the output filter coefficient signal Sc (step (b) in FIG. 12). S4).

  Here, based on the memory signal Sm output from the memory unit 1, the dark current noise detection unit 30 first selects the filter coefficient C having the weakest noise removal strength (that is, the weight of the filter coefficient C in the center pixel is large). The table specifying signal Snc is set so as to specify the filter coefficient table T to be included, and is output to the filter coefficient setting unit 61. When the output signal Sout generated as a result of the noise removal processing using the designated filter coefficient table T is fed back from the calculation unit 4, the dark current noise detection unit 30 is included in the output signal Sout. When the dark current noise portion BN is detected again and the detected number is larger than the allowable number specified by the operation signal Son, the filter coefficient table T including the filter coefficient C having a strong one-step noise removal strength is specified. Thus, the table designation signal Snc is reset and output to the filter coefficient setting unit 61 again. The above processing is automatically repeated until the number of dark current noise portions BN included in the output signal Sout fed back from the calculation unit 4 becomes equal to or less than the allowable number. When the number of dark current noise portions BN included in the output signal Sout fed back from the calculation unit 4 becomes less than the allowable number, the filter coefficient table T at that time is an appropriate filter coefficient table for the filter target region F at that time. Assuming T, the subsequent noise removal processing for the digital image G including the filter target region F is executed for each filter target region F using the filter coefficient table T. The setting process of the filter coefficient table T may be performed for each digital image G or may be performed for each digital image G for several frames.

  As described above, according to the operation of the noise removal filter S7 according to the seventh embodiment, in addition to the effect of the operation of the noise removal filter S1 according to the first embodiment, the filter coefficient C is a dark current in the digital image G. Since the filter coefficient C is automatically set based on the content of the digital image G and the number of dark current noise portions BN included in the digital image G, the filter coefficient C is associated with the number of noise portions BN. Therefore, more effective noise removal processing can be performed.

(IX) Eighth Embodiment Next, an eighth embodiment which is another embodiment according to the present invention will be described with reference to FIG. FIG. 13 is a diagram illustrating a noise removal filter according to the eighth embodiment. In the description of the noise removal filter according to the eighth embodiment, the same member numbers and step numbers are used for the same members and processes as those in the noise removal filter S7 according to the seventh embodiment. Detailed description is omitted.

  In the noise removal filter S7 according to the seventh embodiment, the filter coefficient C corresponding to each pixel Px is switched according to the number of dark current noise portions BN included in the digital image G. In addition, in the noise removal filter according to the eighth embodiment, the noise threshold used for the noise determination processing in the threshold determination unit 3 is also automatically switched according to the number of dark current noise portions BN.

  That is, as shown in FIG. 13, the noise removal filter S8 according to the eighth embodiment is different from the noise removal filter S7 according to the seventh embodiment in that the threshold setting includes a noise threshold setting unit 5 and a threshold memory 51. Part 50 is provided. At this time, in the threshold memory 51, as in the case of the noise removal filter S4 according to the fourth embodiment shown in FIG. 7, a plurality of noise threshold values H1 to Hn (n is a natural number), which are different noise threshold values, are nonvolatile. Is remembered. The table designation signal Snc in the noise removal filter S8 according to the eighth embodiment is output not only to the filter coefficient setting unit 61 but also to the noise threshold setting unit 5. Thereby, the noise threshold value setting unit 5 sets the noise threshold value H corresponding to the table specifying signal Snc.

  In this configuration, the user performs an allowable number designation operation on the operation unit 7 as in the noise removal filter S7 according to the seventh embodiment. Then, the dark current noise detection unit 30 performs the same processing as the dark current noise detection unit 30 according to the seventh embodiment, generates the table designation signal Snc, and sends it to the noise threshold setting unit 5 and the filter coefficient setting unit 61. Output each.

  Thereby, the noise threshold value setting unit 5 sets one of the noise threshold values H corresponding to the table designation signal Snc from the noise threshold values H stored in the threshold value memory 51 based on the table designation signal Snc. The set noise threshold H is output to the threshold determination unit 3 for each digital image G as a noise threshold signal Sh. Specifically, when the mth noise threshold value Hm is preset as the noise threshold value to be associated with the specification of the filter coefficient table T including the filter coefficient C having the weakest noise removal strength, for example, the noise threshold value setting unit 5 first The mth noise threshold value Hm is output to the threshold value determination unit 3 as a noise threshold value signal Sh. On the other hand, similarly to the noise removal filter S7 according to the seventh embodiment, the filter coefficient setting unit 61 selects one of the filters from the filter coefficient table T stored in the table memory 60 based on the table designation signal Snc. The coefficient table T is set, and the filter coefficient C included in the set filter coefficient table T is output to the calculation unit 4 for each pixel Px as the filter coefficient signal Sc.

  Thus, the threshold determination unit 3 executes the noise determination process corresponding to the above (4) for each filter target region F based on the output noise threshold signal Sh. Similarly, the arithmetic unit 4 executes the multiplication / addition process corresponding to (5) above and the addition process corresponding to (6) above based on the output filter coefficient signal Sc.

  As described above, according to the operation of the noise removal filter S8 according to the eighth embodiment, in addition to the effect of the operation of the noise removal filter S7 according to the seventh embodiment, the contents of the digital image G and the dark current noise Since the filter coefficient C and the noise threshold H are set based on the number of the partial BN, more effective noise removal processing can be performed in consideration of the characteristics of the digital image G.

(X) Ninth Embodiment Next, a ninth embodiment, which is another embodiment according to the present invention, will be described with reference to FIG. FIG. 14 is a diagram illustrating a noise removal filter according to the ninth embodiment. In describing the noise removal filter according to the ninth embodiment, the same member numbers and step numbers are used for the same members and processes as those in the noise removal filter S1 according to the first embodiment. Detailed description is omitted. Furthermore, in the ninth embodiment described below and the tenth embodiment described later, as another example of a configuration in which the filter coefficient C or the noise threshold H is automatically set, a standard (for example, a digital image G should comply with) Based on the so-called file format), the filter coefficient C or the noise threshold H is automatically set.

  In other words, the noise included in the digital image G may correspond to a standard such as a file format of the digital image G. Therefore, in the ninth embodiment described below and the tenth embodiment described later, noise generated according to the standard is statistically analyzed in advance, and a filter coefficient table to be applied to the noise. T (or noise threshold H) is determined in advance. Further, it is possible to specify the standard as the digital image G by the user's operation or automatic recognition using the header information of the file itself, and by switching the filter coefficient C or the like in accordance with the specified standard, the effect can be further improved. Noise removal processing is automatically performed.

  As illustrated in FIG. 14, the noise removal filter S9 according to the ninth embodiment includes a parameter selection unit 40 with respect to the noise removal filter S1 according to the first embodiment.

  In this configuration, in addition to the noise threshold setting operation similar to that of the noise removal filter S1 according to the first embodiment, the user performs an operation for specifying the standard of the digital image G input as the image signal Sin on the operation unit 7. Run. An operation signal Sop corresponding to this operation is output from the operation unit 7 to the parameter selection unit 40. Here, the parameter selection unit 40 stores in advance information indicating a filter coefficient table T to be applied to noise generated in accordance with the above-described standard such as the file format. The parameter selection unit 40 generates an identification signal Spc for identifying the filter coefficient table T to be applied to the standard indicated by the operation signal Sop, and outputs the identification signal Spc to the filter coefficient setting unit 61. Each filter coefficient table T stored in the table memory 60 of the noise removal filter S9 according to the ninth embodiment is a filter coefficient table T corresponding to any of the above standards. In addition, the operation signal Soc output from the operation unit 7 to the filter coefficient setting unit 61 in the noise removal filter S1 according to the first embodiment is not output from the noise removal filter S9 according to the ninth embodiment.

  Thereby, the filter coefficient setting unit 61 sets the filter coefficient table T identified by the identification signal Spc from the filter coefficient table T stored in the table memory 60, and is included in the set filter coefficient table T. The filtered filter coefficient C is output to the calculation unit 4 for each pixel Px as a filter coefficient signal Sc.

  Thereby, the calculation unit 4 executes the multiplication / addition processing corresponding to (5) above and the addition processing corresponding to (6) above based on the output filter coefficient signal Sc.

  As described above, according to the operation of the noise removal filter S9 according to the ninth embodiment, in addition to the effect of the operation of the noise removal filter S1 according to the first embodiment, the filter coefficient table T includes the digital image G. Since the filter coefficient table T is set on the basis of the standard that complies with the standard, more effective noise removal processing can be performed.

(XI) Tenth Embodiment Finally, a tenth embodiment which is another embodiment according to the present invention will be described with reference to FIG. FIG. 15 is a diagram illustrating a noise removal filter according to the tenth embodiment. Further, in describing the noise removal filter according to the tenth embodiment, the same member numbers and step numbers are used for the same members and processes as those in the noise removal filter S9 according to the ninth embodiment. Detailed description is omitted.

  In the noise removal filter S9 according to the ninth embodiment, the filter coefficient C corresponding to each pixel Px is switched according to the standard that the digital image G should comply with. In addition, in the noise removal filter according to the tenth embodiment, the noise threshold used for the noise determination process in the threshold determination unit 3 is also switched according to the standard.

  That is, as shown in FIG. 15, the noise removal filter S10 according to the tenth embodiment is different from the noise removal filter S9 according to the ninth embodiment in that the threshold setting includes a noise threshold setting unit 5 and a threshold memory 51. Part 50 is provided. At this time, as in the case of the noise removal filter S4 according to the fourth embodiment shown in FIG. 7, a plurality of noise threshold values H1 to Hn (n is a natural number) having different noise threshold values H are stored in the threshold value memory 51. Each noise threshold value H corresponding to the standard is stored in a nonvolatile manner. The identification signal Spc in the noise removal filter S10 according to the tenth embodiment is output not only to the filter coefficient setting unit 61 but also to the noise threshold setting unit 5.

  In this configuration, the user performs a standard designation operation on the operation unit 7 similar to that of the noise removal filter S9 according to the ninth embodiment. Then, the parameter selection unit 40 performs the same processing as the parameter selection unit 40 according to the ninth embodiment, generates the identification signal Spc, and outputs it to the noise threshold setting unit 5 and the filter coefficient setting unit 61.

  Thereby, the noise threshold value setting unit 5 sets one of the noise threshold values H corresponding to the standard indicated by the identification signal Spc from the noise threshold values H stored in the threshold value memory 51, and the set noise threshold value H is output to the threshold determination unit 3 for each digital image G as a noise threshold signal Sh. On the other hand, similarly to the noise removal filter S9 according to the ninth embodiment, the filter coefficient setting unit 61 selects any one of the filter coefficient tables T stored in the table memory 60 corresponding to the standard indicated by the identification signal Spc. The filter coefficient table T is set, and the filter coefficient C included in the set filter coefficient table T is output to the calculation unit 4 as the filter coefficient signal Sc for each pixel Px.

  Thus, the threshold determination unit 3 executes the noise determination process corresponding to the above (4) for each filter target region F based on the output noise threshold signal Sh. Similarly, the arithmetic unit 4 executes the multiplication / addition process corresponding to (5) above and the addition process corresponding to (6) above based on the output filter coefficient signal Sc.

  As described above, according to the operation of the noise removal filter S10 according to the tenth embodiment, in addition to the effect of the operation of the noise removal filter S9 according to the ninth embodiment, the filter coefficient table T and the noise threshold H are Since the digital image G is associated with a standard to which the digital image G complies, and the filter coefficient table T and the noise threshold H are set based on the standard, more effective noise removal processing can be performed.

  The noise removal filter S7 according to the seventh embodiment to the noise removal filter S10 according to the tenth embodiment is an embodiment in the case of automatically selecting the filter coefficient table T and the noise threshold H. Various noise analysis results can be applied to the present invention. Moreover, since evaluation with respect to noise differs depending on a user, an automatic selection function such as a filter coefficient table T in the noise removal filter S7 according to the seventh embodiment to the noise removal filter S10 according to the tenth embodiment, and the first embodiment. By combining the noise removal filter S1 to the manual selection function such as the filter coefficient table T in the noise removal filter S6 according to the sixth embodiment, a more versatile and effective noise removal process can be executed. it can.

  Further, the noise removal filter S according to each embodiment described above is an information recording apparatus that records the digital image G on a recording medium, and after the encoding process on the digital image G is completed and before the recording process on the recording medium. It can be configured to be inserted in stages. Examples of the recording medium include an optical disk and a hard disk. Further, the noise removal filter S is configured to be prepared for a stage before or after the decoding process for the digital image G in an information reproducing apparatus for reproducing the digital image G from a recording medium on which the digital image G has been recorded. It is also possible to do.

  Furthermore, a program corresponding to the functions of the memory unit 1, the difference calculation unit 2, the threshold value determination unit 3, the calculation unit 4, the noise threshold value setting unit 5, the filter coefficient setting unit 61, and the like according to each embodiment, for example, an optical disc or the like. A memory unit according to each embodiment is recorded on a recording medium or acquired and recorded from a network such as the Internet, and read out and executed by a general-purpose microcomputer. 1, a difference calculation unit 2, a threshold determination unit 3, a calculation unit 4, a noise threshold setting unit 5, a filter coefficient setting unit 61, and the like.

  As described above, the present invention can be used in the field of noise removal devices, and particularly when applied to the field of noise removal devices that remove noise contained in digital images, a particularly remarkable effect can be obtained. .

DESCRIPTION OF SYMBOLS 1 Memory part 2 Difference calculation part 3 Threshold determination part 4 Calculation part 5 Noise threshold value setting part 6 Filter setting part 10 Dark part determination part 11 Dark part threshold value setting part 20 Area determination part 30 Dark current noise detection part 40 Parameter selection part 50 Threshold setting part 51 Threshold memory 60 Table memory 61 Filter coefficient setting unit F Filter target area Px11, Px12, Px13, Px14, Px15, Px21, Px22, Px23, Px24, Px25, Px31, Px32, Px33, Px34, Px35, Px41, Px42, Px43 , Px44, Px45, Px51, Px52, Px53, Px54, Px55 Pixels S1, S2, S3, S4, S5, S6, S7, S8, S9, S10 Noise reduction filter T1 First filter coefficient table T2 Second filter coefficient table T3 Third filter coefficient table T4 Fourth filter coefficient table Tn Nth filter coefficient table Sin Image signal Sm Memory Signal Sdif Difference signal Soc, Sot, Sod, Son, Soa, Sop Operation signal Sc Filter coefficient signal Sth Threshold signal Spx Pixel signal Sout Output signal Sdc Dark part signal G, G1, G2 Digital image D Dark part H1 First noise threshold H2 Second Noise threshold H3 Third noise threshold H4 Fourth noise threshold Hn nth noise threshold Sh noise threshold signal Ga, Gb, Gc, Gd, Ge, Gf, Gg, Gh Image area Ssc area signal BN1, BN2 Dark current noise part Snc table Designation signal Spc identification signal

Claims (6)

Each of at least one or more other pixels around one pixel in a digital still image formed by a plurality of pixels, and a detection means for detecting a difference value of a signal between the one pixel,
A comparing means for comparing each detected difference value with a threshold value;
A specifying unit for specifying the other pixel to be subjected to a noise removal process based on a comparison result by the comparing unit;
Dark part detection means for detecting a region of the digital still image containing dark current noise as a dark part of the digital still image;
When the detected dark portion includes the one pixel, a filter coefficient to be multiplied with the detected difference value of each of the specified other pixels is detected in the digital still image. Setting means for setting each of the other pixels based on the position of the dark part that has been performed;
Multiplying means for multiplying the detected difference value by the set filter coefficient for each of the identified other pixels;
Adding means for adding the result of multiplication by the multiplication means to the signal of the one pixel and outputting it;
The dark portion is formed by detection processing by the detection unit, comparison processing by the comparison unit, specification processing by the specification unit, multiplication processing by the multiplication unit using the set filter coefficient, and addition processing by the addition unit. Each of the pixels to be used as the one pixel, and a control unit that repeats the pixels forming the dark part;
A noise removal apparatus comprising: In the noise removal apparatus of Claim 1,
A filter coefficient table including the filter coefficient corresponding to each of the other pixels included in the dark portion in one digital still image, and each of the regions included in a region other than the dark portion in one digital still image. A storage means for storing at least a filter coefficient table including the filter coefficient corresponding to each of the other pixels;
The setting means switches the stored filter coefficient table based on the detected dark part and outputs to the multiplication means.
The noise removing apparatus , wherein the multiplication means performs the multiplication process using the filter coefficient included in the output filter coefficient table . In the noise removal apparatus according to claim 1 or 2,
The other pixel is the pixel in a symmetrical position with respect to the one pixel in the digital still image;
The noise removal apparatus according to claim 1, wherein the filter coefficients are the same in the other pixels at the symmetrical positions . In a noise removal method executed in a noise removal apparatus for removing noise included in a digital still image formed by a plurality of pixels,
A detection step of detecting a difference value of a signal between each of at least one other pixel around one pixel in the digital still image and the one pixel;
A comparison step of comparing each detected difference value with a threshold value;
Based on the comparison result in the comparison step, a specifying step for specifying the other pixel to be subjected to noise removal processing;
A dark portion detection step of detecting a region of the digital still image including dark current noise as a dark portion of the digital still image;
When the detected dark portion includes the one pixel, a filter coefficient to be multiplied with the detected difference value of each of the specified other pixels is detected in the digital still image. A setting step for setting each of the other pixels based on the position of the dark part,
For each of the identified other pixels, a multiplication step of multiplying the detected difference value by the set filter coefficient;
An addition step of adding the multiplication result in the multiplication step to the signal of the one pixel and outputting the signal;
The dark part is formed by the detection process in the detection process, the comparison process in the comparison process, the specific process in the specific process, the multiplication process in the multiplication process using the set filter coefficient, and the addition process in the addition process. Each of the pixels to be used as the one pixel, and a control step for repeating each of the pixels forming the dark part;
The noise removal method characterized by including . A computer included in a noise removal device that removes noise included in a digital still image formed by a plurality of pixels ,
Detecting means for detecting a difference value of a signal between each of at least one other pixel around one pixel in the digital still image and the one pixel;
Comparison means for comparing each detected difference value with a threshold value,
Identifying means for identifying the other pixel to be subjected to noise removal processing based on a comparison result in the computer functioning as the comparing means ;
Dark part detecting means for detecting a region of the digital still image containing dark current noise as a dark part of the digital still image;
When the detected dark portion includes the one pixel, a filter coefficient to be multiplied with the detected difference value of each of the specified other pixels is detected in the digital still image. Setting means for setting each of the other pixels based on the position of the dark portion
Multiplying means for multiplying the detected difference value by the set filter coefficient for each of the identified other pixels;
Adding means for adding the result of multiplication by the computer functioning as the multiplying means to the signal of the one pixel and outputting the signal ; and
Detection processing by the computer functioning as the detection means, comparison processing by the computer functioning as the comparison means, specification processing by the computer functioning as the specification means, function as the multiplication means using the set filter coefficient Control means for repeating the multiplication process by the computer and the addition process by the computer functioning as the addition means, with each of the pixels forming the dark part as the one pixel, for each of the pixels forming the dark part ,
A noise removal program characterized by functioning as 6. A recording medium in which the noise elimination program according to claim 5 is recorded so as to be readable by the computer .