JP4785437B2 - Image processing method and image processing apparatus - Google Patents

Description

  The present invention is an image processing apparatus characterized by optical means classified in the international classification G01B11 / 00. In particular, a subject to be photographed is photographed as image data having a plurality of gradations by a digital camera or the like, and measured from this image. This is related to processing for recognizing a target and obtaining a feature amount of the recognized target. For example, the present invention relates to a technique for detecting a fluorescent substance in a fluorescent microscope image and calculating a feature amount based on a luminance value in a pixel including the detected fluorescent substance.

  Conventionally, as a method for calculating a feature amount in a measurement target in image data, generally, an image recognition process is first performed to identify the position of the measurement target in the image data, and then measurement is performed from the image data at the recognized position. A technique for calculating the feature amount of the target is used.

  Various methods have been proposed for image recognition. Although they are performed for various purposes, as a technique, generally, a correlation value between a specific template image and a target image is obtained, and the degree of fitness between the template image and the target image is evaluated by evaluating the correlation value. The method of seeking is taken.

Hereinafter, a general recognition method used very widely will be described with reference to FIG. FIG. 15 shows an example of a general template matching procedure.
In FIG. 15, reference numeral 1501 denotes a template image, and a round image pattern indicated in the template image is searched from the target image 1502.

  There are many round patterns in the target image 1502 as indicated by 1503 and 1504. In template matching, a round image pattern shown in the template image 1501 is extracted from the target image 1502 by sequentially searching the target image 1502 using the template image 1501. First, a frame having the same size as the template image 1501 is assumed at the position of the upper left 1505 of the target image 1502, and the target image 1502 is cut out. Then, these images are overlapped, the pixel at the corresponding position is subjected to a difference calculation for each pixel, and the difference results of all the overlapped pixels are individually squared and then added. This value means the correlation between both images, and is a small value when they match, and a large value when they are different. If the obtained value is smaller than the threshold value, it is determined that a pattern has been detected, and if the obtained value is larger than the threshold value, it is determined that there is no pattern.

  When the determination is made at the position of the first pattern matching position 1505, the same determination is made by shifting the frame to the right by one pixel, and this is repeated, and the entire target image 1502 is repeated in the order of the arrow 1506. Do it sequentially.

  When the frame is at the position 1507, the pattern including the template image 1501 and the detection target 1503 match, and the sum of squares has a very small value, so that the presence of a round pattern can be determined.

  For such image recognition processing, a method of performing a search by binarizing a target image is disclosed. The template image is divided to create several partial template images. First, one partial template image is used to search for a matching position. If they match, a small region near the matching position is set, the remaining partial template images are matched, and if all the partial template images match, the presence of the search pattern is determined. By using such a method, recognition with high speed and high accuracy is realized as compared with a general method (for example, see Patent Document 1).

  Then, when obtaining brightness (average luminance value) as the feature quantity of the measurement object, the sum of the luminance values of the pixels included in the pattern area indicating the measurement object at the detected position is obtained, and the obtained sum is calculated in the pattern area. The luminance average value was calculated by dividing by the number of pixels.

  On the other hand, the method of calculating the feature amount of the measurement target in the image data is used in various fields. For example, there is a measurement of a fluorescent substance by a fluorescence microscope image to observe gene expression.

  Semiconductor nanocrystals having the property of emitting light of different wavelengths depending on the composition and size when irradiated with excitation light of a specific short wavelength are attached to the surface of a bead made of translucent acrylic. At this time, by attaching a plurality of semiconductor nanocrystals having different emission wavelengths at an arbitrary ratio, each bead is encoded with a unique luminance spectrum distribution, and a specific mRNA (messenger) is added to each encoded bead. A probe for binding to ribonucleic acid is attached, and the code and mRNA are made to correspond one-to-one. Thus, it was possible to analyze the type of expressed mRNA by observing and decoding the luminance spectrum distribution of beads emitted by irradiating excitation light (see, for example, Patent Document 2).

And the luminance spectrum distribution of the beads used for the analysis is photographed while switching the band pass filter with respect to the light emitting beads with a fluorescence microscope apparatus that performs measurement using a plurality of band pass filters having different pass bands. It was created by calculating the average luminance value of the bead area in the image taken through the vant pass filter. In addition, a well plate in which a plurality of beads are placed in one well to be imaged with a measurement target is prepared, and a plurality of bead-pass filters are photographed while switching a plurality of bandpass filters for one well of the well plate. Simultaneous analysis was possible. Specifically, for each well, first, an image for extracting the position of the beads is photographed, the position of each bead is extracted by the template matching on the image, and then a plurality of photographs photographed through each of the band pass filters. In each image, the average luminance value of the region corresponding to each extracted bead position was obtained, and the luminance spectrum of each bead was created and analyzed (for example, see Patent Document 3).
JP-A-62-232504 Special table 2004-500109 gazette Japanese Patent Laid-Open No. 10-309281

  However, when a plurality of beads are placed in one well, the beads are not evenly distributed and the beads are adjacent to each other. Since such adjacent beads are affected by the light emitted from each other, there is a problem that the luminance spectrum distribution cannot be created with high accuracy. To solve this problem, it is possible to obtain good analysis results by removing adjacent beads from the analysis target and using the luminance spectrum distribution of only non-adjacent beads, but statistical analysis is required. In this case, there is a problem that the absolute number of beads to be analyzed is reduced and the reliability of the analysis is impaired.

  Further, since the range in which adjacent beads are affected is a peripheral portion of the beads, a method of obtaining an average luminance value in a region excluding the peripheral portion of the extracted bead region is also conceivable. However, the emitting semiconductor nanocrystal is attached to the surface of the bead, and when such a bead is photographed, the peripheral part emits light more strongly than the central part of the bead. Therefore, removing the peripheral portion where the light emission characteristic appears most has a problem that the luminance spectrum distribution cannot be obtained accurately.

  The present invention solves the above-described conventional problems. In a field where it is necessary to accurately obtain the characteristic quantities of various measurement objects including the beads to which the semiconductor nanocrystals are attached, light from adjacent objects is obtained. It is an object of the present invention to provide an image processing method and an image processing apparatus capable of accurately obtaining a feature amount using a luminance value of a pixel including a measurement target even in a measurement target affected by the above.

In order to solve the conventional problem, the image processing method of the present invention sets one or a plurality of specific pattern areas for a digital image having multi-valued luminance values in each pixel, An image processing method for calculating a representative value in each pattern region using a luminance value of a pixel included in the pattern region,
A pattern region setting step for setting a region where a measurement object exists for the digital image as a pattern region, and a pattern region dividing step for dividing the pattern region set in the pattern region setting step into a plurality of pattern division regions In the pattern area set in the pattern area setting step, the measurement object existing in the pattern area other than the target pattern area affects the area influencing the luminance value of the digital image. An influence area setting step to set as an area, and among pixels included in the pattern area set in the pattern area setting process, only pixels that do not include the influence area set in the influence area setting step are configured. Pixels in the pattern division area divided in the pattern area division step It is obtained by comprising: a feature amount calculating step of calculating a representative value in the pattern area by using only.

  Furthermore, in the image processing method of the present invention, in the pattern region dividing step, the pattern region is divided using a mask pattern indicating a plurality of divided regions.

  Further, in the image processing method of the present invention, the mask pattern has a divided area indicating a pattern area, and pixels corresponding to the divided area are used as the pattern divided area.

  Furthermore, in the image processing method of the present invention, in the pattern region dividing step, the pattern region is divided into a plurality of pattern divided regions so that the areas of the divided regions are equal.

  Furthermore, in the image processing method of the present invention, in the pattern region dividing step, the pattern region is divided into a plurality of pattern divided regions by a straight line passing through the center of the pattern region.

  Furthermore, in the image processing method of the present invention, in the pattern region dividing step, a straight line passing through the center of the pattern region is divided into a plurality of pattern divided regions so that the central angles are equal. is there.

  Furthermore, in the image processing method of the present invention, in the pattern region dividing step, a specific region is excluded from a plurality of pattern divided regions obtained by dividing the pattern region.

  Further, in the image processing method of the present invention, the specific area excluding the pattern division area is set so that the areas are equal in each divided pattern division area.

  Furthermore, in the image processing method of the present invention, in the influence area setting step, the influence area is set as a neighboring area including a pattern area other than the target pattern area, and pixels constituting the pattern area are grouped. In the connected component, the number of pixels that are contours is increased outward by a predetermined number of pixels, and the increased connected component is set.

  Furthermore, in the image processing method of the present invention, the number of pixels for increasing the contour pixels outward is set based on the value obtained using the luminance value of each pixel of the connected component for which the contour is to be increased. It is characterized by doing.

  Furthermore, in the image processing method of the present invention, in the influence area setting step, an area that is influenced by objects other than the measurement object existing in the pattern area is added to the influence area due to the measurement object existing in the pattern area. It is characterized by that.

  Furthermore, in the image processing method of the present invention, in the feature amount calculation step, the representative value in the pattern area is an average luminance value obtained by averaging the luminance values of the pixels included in the pattern area, and only the pixels not included in the affected area are included. The cumulative value obtained by adding all the luminance values of the pixels in the pattern division area constituted by the above is obtained, and the representative value is obtained by dividing by the number of pixels obtained by adding the obtained cumulative value. It is.

  Furthermore, in the image processing method of the present invention, in the feature amount calculation step, the representative value in the pattern area is a luminance sum value obtained by summing the luminance values of the pixels included in the pattern area, and only the pixels not including the affected area are included. The pixel number of the second pattern division area constituted by the pixels including the influence area is obtained while obtaining the total number of the sum of the luminance values of the pixels in the first pattern division area constituted by the sum and the accumulated value. Then, the cumulative value of the second pattern division region is calculated based on the cumulative value of the first pattern division region, the number of pixels of the first pattern division region, and the number of pixels of the second pattern division region. By analogy, the representative value is obtained by adding together the accumulated value of the second pattern divided region estimated by the accumulated value of the first pattern divided region. .

  Further, in the image processing method of the present invention, the representative value obtained in the target pattern area using the information of the pattern division area divided in the pattern area division step and the influence area set in the influence area setting step. A determination step for determining whether or not is valid or invalid, this determination step is performed immediately before the feature amount calculation step, and a representative value in the pattern area determined to be invalid is not obtained.

  Furthermore, in the image processing method of the present invention, in the determination step, the determination is performed based on the number of pattern division regions or the number of pixels including pixels that include or do not include the affected region.

In order to solve the conventional problems, the image processing apparatus of the present invention sets one or a plurality of specific pattern areas for a digital image having multi-valued luminance values in each pixel, and the digital image In the image processing apparatus, a representative value in each pattern region is calculated using a luminance value of a pixel included in the pattern region,
A pattern region setting unit that sets a region where a measurement object exists for the digital image as a pattern region, and a pattern region dividing unit that divides the pattern region set in the pattern region setting unit into a plurality of pattern division regions In the pattern area set in the pattern area setting unit, the measurement object existing in the pattern area other than the target pattern area affects the area that affects the luminance value of the digital image. An influence area setting unit set as an area, and among pixels included in the pattern area set in the pattern area setting unit, only pixels that do not include the influence area set in the influence area setting unit are configured. A pattern is formed using only pixels in the pattern division area divided in the pattern area division unit. It is obtained by comprising: a feature amount calculating unit that calculates a representative value in the down area.

  Further, in the image processing apparatus of the present invention, the pattern area dividing unit divides the pattern area using a mask pattern indicating a plurality of divided areas.

  Furthermore, in the image processing apparatus of the present invention, the mask pattern has a divided area indicating a pattern area, and a pixel corresponding to the divided area is set as a pattern divided area.

  Furthermore, in the image processing apparatus of the present invention, the pattern region dividing unit divides the pattern region into a plurality of pattern divided regions so that the areas of the divided regions are equal.

  Furthermore, in the image processing apparatus of the present invention, the pattern area dividing unit divides the pattern area into a plurality of pattern divided areas by a straight line passing through the center of the pattern area.

  Furthermore, in the image processing apparatus according to the present invention, the pattern area dividing unit divides the pattern area into a plurality of pattern division areas so that the central angles are uniform along a straight line passing through the center of the pattern area. is there.

  Furthermore, in the image processing apparatus according to the present invention, the pattern area dividing unit excludes a specific area from a plurality of pattern division areas obtained by dividing the pattern area.

  Furthermore, in the image processing apparatus of the present invention, the specific area excluding the pattern division area is set so that the area is equal in each divided pattern division area.

  Furthermore, in the image processing apparatus of the present invention, the influence area setting unit sets the influence area as a vicinity area including a pattern area other than the target pattern area, and collects pixels constituting the pattern area as a group. In the connected component, the number of pixels that are contours is increased outward by a predetermined number of pixels, and the increased connected component is set.

  Furthermore, in the image processing apparatus according to the present invention, the number of pixels that increase the number of pixels that form the contour to the outside is set based on the magnitude of the value obtained using the luminance value of each pixel of the connected component for which the contour is to be increased. It is characterized by doing.

  Furthermore, in the image processing apparatus of the present invention, the influence area setting unit adds an area influenced by objects other than the measurement object existing in the pattern area to an influence area due to the measurement object existing in the pattern area. It is characterized by that.

  Furthermore, in the image processing apparatus of the present invention, in the feature amount calculation unit, the representative value in the pattern region is an average luminance value obtained by averaging the luminance values of the pixels included in the pattern region, and only the pixels not included in the affected region are included. The cumulative value obtained by adding all the luminance values of the pixels in the pattern division area constituted by the above is obtained, and the representative value is obtained by dividing by the number of pixels obtained by adding the obtained cumulative value. It is.

  Furthermore, in the image processing apparatus of the present invention, in the feature amount calculation unit, the representative value in the pattern area is set as a luminance sum total value of the luminance values of the pixels included in the pattern area, and only the pixels not including the affected area are included. The pixel number of the second pattern division area constituted by the pixels including the influence area is obtained while obtaining the total number of the sum of the luminance values of the pixels in the first pattern division area constituted by the sum and the accumulated value. Then, the cumulative value of the second pattern division region is calculated based on the cumulative value of the first pattern division region, the number of pixels of the first pattern division region, and the number of pixels of the second pattern division region. By analogy, the representative value is obtained by adding together the accumulated value of the second pattern divided region estimated by the accumulated value of the first pattern divided region.

  Further, in the image processing apparatus of the present invention, the representative value obtained in the target pattern area using the information of the pattern division area divided in the pattern area division section and the influence area set in the influence area setting section. Is provided with a determination unit that determines whether or not is valid and invalid, and a representative value in the pattern area determined to be invalid by the determination unit is not obtained.

  Furthermore, in the image processing apparatus of the present invention, the determination unit performs the determination based on the number of pattern division regions or the number of pixels including pixels that include or do not include the affected region.

  The image processing method and the image processing apparatus according to the present invention accurately obtain a feature amount using a luminance value in a pixel including a measurement target even in a measurement target affected by light from an adjacent target. Has the effect of being able to.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In addition, embodiment shown here is an example to the last, Comprising: It is not necessarily limited to this embodiment.
[Embodiment 1]
As Embodiment 1 of the present invention, when irradiated with excitation light having a specific short wavelength, a semiconductor nanocrystal having a property of emitting light of a different wavelength depending on the composition and size is attached at an arbitrary ratio of 10 μm in diameter. A microbead made of spherical and translucent acrylic (hereinafter referred to as a bead) is used as a measurement object, a plurality of beads are photographed with a fluorescence microscope, and the luminance value of the pixel including the beads is used in the photographed digital image. The case where the feature amount is obtained will be described as an example.

  FIG. 2 is a measurement image obtained by irradiating beads of a measurement object with excitation light and photographing the emission of the beads through a band pass filter. In the actual measurement image, several hundred beads are shown, but in the measurement image of this example, seven beads are shown for simplicity. In the measurement image shown in FIG. 2, the bead that emits a lot of light having a wavelength that passes through the bandpass filter used for photographing is brighter. Moreover, since the light-emitting semiconductor nanocrystal is attached to the surface of the bead, the periphery is brighter than the central portion.

  FIG. 1 is a flowchart showing processing performed by the image processing method according to Embodiment 1 of the present invention. The flowchart of FIG. 1 includes steps S101 to S104, where step S101 is a pattern region setting step, step S102 is a pattern region dividing step, step S103 is an affected region setting step, and step S104 including steps S1041 to S1044 is step S104. Each of the feature amount calculation steps is shown.

Hereinafter, image processing performed by the image processing method according to the first embodiment of the present invention on the measurement image shown in FIG. 2 will be described in detail with reference to FIGS. 3 to 10 according to flowcharts.
(Step S101 / Pattern area setting step)
First, in the measurement image, a pattern region is set which is a region where a bead exists and a feature amount is obtained. This is performed by performing a template matching process on a digital image (hereinafter referred to as a target image) for extracting a bead position photographed in advance and extracting the bead position. Here, the digital image to be subjected to the template matching process is set as a target image, not a measurement image, in order to increase the accuracy of bead extraction.

  FIG. 3 is an example of a target image obtained by photographing a shadow formed by applying light from the background of the beads shown in FIG. In the target image shown in FIG. 3, there is a very bright area at the center of the bead. This is because the bead is made of translucent acrylic. This is because the light is concentrated in the central part and the center becomes very bright.

  In this way, by photographing the shadow of the measurement object, even if the beads have different emission characteristics, all the beads have the same characteristics, and the extraction accuracy by the template matching process can be increased.

  Hereinafter, the pattern area setting method will be described more specifically. Prepare multi-value digital image data (pattern area image) initialized with 0 of the same size as the target image in advance, perform template matching processing on the target image, extract the bead position and extract the extracted beads A number is assigned in order from 1, and a process of writing the bead number to each pixel of the pattern area image corresponding to each pixel of the extracted bead area is performed.

  FIG. 4 is a pattern region image set based on the target image shown in FIG. The numerical values in the figure indicate the pixel values of each region. In this pattern region image, the pixel value of each bead region is 1 to 7 in the order in which the bead position is extracted, and the pixel value of the region other than the beads is 0.

  In order to refer to the pattern area, an area where the pixel value is not 0 in the pattern area image may be examined. Furthermore, each bead can be identified by examining the pixel value of the pattern area.

  The template matching process uses a template image obtained by cutting out only the bead region extracted from the target image, superimposes the target image and the template image, obtains a difference value in each overlapping pixel, and individually calculates the difference results of the entire template image. After squared, add. This added value indicates the correlation between both images, and is a small value when they are the same image, and a large value when they are different. Therefore, an addition value that can be determined to match in advance is used as a threshold value. When the obtained addition value is smaller than the threshold value, it is determined that a bead has been detected, and when it is larger than the threshold value, it can be determined that no bead exists. Then, the same determination is performed while shifting the template image pixel by pixel, and this determination process is repeated for the entire target image to extract the position of the beads.

The pattern area image indicating the pattern area set in this way is referred to in the following steps S102, S103, and S104.
(Step S102 / pattern area dividing step)
Next, a process of dividing each pattern area shown in the pattern area image set in step S101 into a plurality of pattern division areas using a mask pattern is performed.

  The mask pattern is a multi-value digital image data indicating a region for obtaining a feature amount in one bead and each divided region obtained by dividing this region into a plurality of regions. Pixel values assigned in order from 1 are set to 0, and other areas are set to 0.

  FIG. 5 is a diagram illustrating an example of a mask pattern. The numerical values in the figure indicate the pixel values of each region. As shown in FIG. 5, it is assumed that the mask pattern is a straight line passing through the center of the region for which the feature amount is obtained, and is divided so that the area of each region becomes equal when the region for obtaining the feature amount is divided. Further, in the mask pattern shown in FIG. 5, the area for obtaining the feature quantity is divided so that the central angles are equal by a straight line passing through the center of the area for obtaining the feature quantity.

  On the other hand, in the process of dividing the pattern area, multi-value digital image data (pattern divided area image) initialized with 0 of the same size as the pattern area image is prepared in advance, and the pattern in the pattern area image is prepared. The center of gravity of the area is obtained for each area, and the mask pattern is overlapped so that the center of gravity of the mask pattern coincides with the position on the pattern division area image corresponding to the obtained position of the center of gravity. The process of writing to the divided area image is repeated. As described above, by using the mask pattern, the pattern region can be divided only by referring to the mask pattern, so that processing can be performed at high speed.

  FIG. 6A is a pattern division region image created using the mask pattern shown in FIG. 5 with respect to the pattern region image shown in FIG. FIG. 6B is an enlarged view of the pattern division area formed by one pattern area shown in FIG. The numerical value in the figure indicates the pixel value of the region, and the pixel value of the pattern division region in the region for obtaining each feature amount is the same value as the pixel value of the corresponding mask pattern when superimposed.

  In order to refer to the pattern division area obtained by dividing the pattern area, the pixel value of the pattern division area image corresponding to the area in the area where the pixel value is not 0 in the pattern area image may be examined.

  In addition, when the pixel value of the pattern division region image corresponding to a non-zero pixel in the pattern region image is 0, this pixel may be excluded from the target for which the feature amount is to be obtained. This occurs when the shape of the pattern region is not correct, and the accuracy of the obtained feature amount can be improved by excluding the pixels other than the accurate pattern region from the target of obtaining the feature amount.

The pattern division area image indicating the pattern division area obtained by dividing the pattern area thus created is referred to in the following step S104.
(Step S103 / influence area setting step)
Next, based on the pattern area image set in step S101, the area where the beads existing in the pattern area other than the target pattern area have an influence on the luminance value of the measurement image is determined as the influence area. Set as. In other words, this affected area is a neighboring area that includes a pattern area other than the target pattern area, and the pixels that form the outline in the connected component that is a group of pixels that make up the pattern area are the predetermined number of pixels outside. Increase and set the connected component after the increase.

Specifically, a region where the light emission of the beads existing in the pattern region affects the luminance value of the measurement image is obtained.
Light emission from the beads is considered to spread uniformly in all directions because the emitting semiconductor nanocrystals are attached almost evenly to the surface of the beads. Therefore, the region where the light emission of the beads has an influence may be obtained by uniformly expanding the pattern region in all directions in each pattern region from which the beads are extracted.

  Therefore, multi-value digital image data (influence area image) initialized with 0 of the same size as the pattern area image is prepared in advance. First, a pattern area including one target bead from the pattern area image is prepared. A binary image in which the extracted pattern area is 1 and the other areas are 0 is created. The extraction of the target pattern area is done by assigning a number to the pixel value so that each bead can be identified when creating the pattern area image. Therefore, the beads existing in the target pattern area in the pattern area image This can be done by extracting pixels having the same value as the number of.

  Next, a general 8-neighbor expansion process is performed on the created binary image. Here, the pixels around the extracted pattern area are set as the target of the 8-neighbor expansion process, and the target pixel and one pixel indicating the pattern area in at least one of the eight adjacent pixels centered on the target pixel If there is, the target pixel is converted to 1 indicating the pattern area. By this processing, a binary image showing a pattern region that is expanded by one pixel in all directions with respect to the pattern region in the binary image before processing can be obtained. Then, the region where the pixel value is 1 in the binary image after this processing is performed a plurality of times is defined as the region in which the beads existing in the target pattern region have an influence, and this region corresponds to this region. The number of the bead existing in the target pattern area is written in each pixel of the affected area image. It should be noted that the number of times of performing the 8-neighbor expansion process may be set such that it is always included in the obtained area even when the area affected by the measurement object is the widest. Use to measure and set numerical values.

  Then, a series of processing from extraction of the target pattern area to writing to the influence area image is repeated for all the pattern areas, and the influence area indicating the area where the beads existing in all the pattern areas have an influence. Get an image.

  However, in this iterative process, when a bead number is written in the affected area image, a value indicating that the area in which another pattern area bead is already written is already written in the pixel to be written, that is, 0. Otherwise, a pixel value that does not exist in the pattern area image is written to that pixel. For example, a value obtained by adding 1 to the number of beads existing in the pattern area image is written.

  FIG. 7 is an influence area image created based on the pattern area image shown in FIG. The numerical value in the figure indicates the pixel value of that region. It should be noted that the pixel value of the region where the influential regions created from the respective pattern regions indicated by diagonal lines overlap is 8 which is obtained by adding 1 to the number of beads.

  When referring to the area where the light emission of the beads has an influence, a non-zero area in the influence area image may be examined. Furthermore, by examining the pixel value of the affected area image, it is possible to identify the affected area corresponding to each bead, and the light emission of the beads existing in the pattern area other than the target pattern area is affected. The affected area can be identified by examining values other than 0 in the affected area image or the pixel value of the target pattern area.

The affected area image created here is referred to in the following step S104.
Here, the number of times of performing the 8-neighbor expansion process on the pattern area is set to a preset value, and the expansion is performed with a uniform size for all the pattern areas. May be set for each pattern area according to the magnitude of the value obtained using the luminance value of the pixel included in the pattern area. The region affected by the light emission of the beads becomes wider as the light emission is stronger, and becomes narrower as the light emission is weaker. Therefore, the average luminance value of the pixels included in the pattern area in the measurement image is obtained, and the number of 8-neighbor expansion processes is set according to the obtained average luminance value, thereby accurately affecting the intensity of light emission. It is possible to set the area. This method of setting the number of 8-neighbor expansion processes has a table in which the number of 8-neighbor expansion processes for a certain average luminance value is recorded in a one-to-one manner, and measures numerical values in advance using a plurality of measurement objects. Create a table.

Further, the process performed in step S103 may be performed before the process performed in step S102, or a pattern area divided image may be generated after an affected area image is generated.
(Step S104 / feature amount calculation step)
Next, using the pattern area image created in step S101, the pattern division area image created in step S102, and the influence area image created in step S103, the luminance value of the pixel included in each pattern area From the above, the feature amount of each pattern area is calculated.

  In step S104, a pixel in one target pattern area is extracted from the pattern area image (step S1041), and among the extracted pixels, pixels that are affected by beads existing in the non-target pattern area. Pixels for obtaining an accurate feature amount excluding the pixel are further extracted (step S1042), and the feature amount is obtained using the luminance value of the pixel in the measurement image corresponding to the extracted pixel (step S1043). Then, this series of processing is performed in all pattern areas (step S1044), and the feature values of all beads are calculated.

  Hereinafter, the process of obtaining the average luminance value of the beads A shown in FIG. 2 in the case where the feature amount (representative value in the pattern region) of the pattern region is an average luminance value obtained by averaging the luminance values of the pixels included in the pattern region. An explanation will be given as an example.

First, the pattern area of the bead A is extracted. This can be done by examining the area where the pixel value is 2 in the pattern area image (see FIG. 4).
Next, each pixel value of the pattern division area image (see FIG. 6) and the influence area image (see FIG. 7) corresponding to the extracted pattern area is examined, and the beads present in the pattern area other than the target pattern area. Extract pixels in the pattern segmentation region that are composed only of pixels that do not include the affected region. The influence area in the bead A can be specified by examining an area in which the pixel value is not 0 or 2 in the influence area image.

FIG. 8 is a diagram showing a pattern division area and an influence area in the vicinity of the bead A in the bead A. The numerical values in the figure indicate the pixel values of each pattern division area obtained by dividing the pattern area.
In FIG. 8, this affected area is an area affected by light emission from the adjacent bead B. Therefore, it can be said that an accurate feature amount can be obtained by extracting only the pixels in the pattern region excluding the affected region and obtaining the feature amount. However, at this time, only the pixels in the pattern division area that does not include the influence area in each pattern division area are extracted. In FIG. 8, pixels in the pattern division areas having pixel values of 1, 2, 3, 4, 7, and 8 indicated by diagonal lines excluding the pattern division areas having pixel values of 5 and 6 are extracted.

  Then, an accumulated value obtained by adding all the luminance values of the pixels in the measurement image corresponding to the extracted pixel is obtained, and an average luminance value in the beads A is obtained by dividing the obtained accumulated value by the number of added pixels. it can.

  By performing the above processing, the average luminance value of the bead A excluding the influence of light emission from the bead B can be obtained. In addition, since the pixels used for calculation are selected in units of regions divided so that the area is equal to a straight line passing through the center of the pattern region, the peripheral portion is more than the central portion like a bead with a semiconductor nanocrystal attached. Even in a measurement object that does not have a uniform brightness value, such as a strong light emission, the ratio of the bright area to the dark area excludes the case where all pixels in the pattern area are used and the affected pixels This is the same when only the pixels in the pattern area are used, and an accurate average luminance value can be obtained.

  If all the pattern division regions include the affected region, there are no pixels for obtaining the feature amount, and the feature amount in the pattern region cannot be obtained. Thereby, it is possible to output only the accurately obtained feature amount without obtaining the feature amount in the pattern region where the entire pattern region is affected.

  Next, in the case where the feature amount of the pattern area (representative value in the pattern area) is a luminance total value obtained by summing up the luminance values of the pixels included in the pattern area, the luminance total value of the beads A shown in FIG. Will be described as an example.

  The processing in the case where the feature amount of the bead A is the luminance sum total value obtained by summing the luminance values of the pixels included in the bead A is similar to the processing for obtaining the luminance average value, and the pattern division region image corresponding to the extracted pattern region Each pixel value of the affected area image is examined, and pixels (1 and 2 indicated by hatching in FIG. 8) of a pattern divided area (hereinafter referred to as a first pattern divided area) composed only of pixels that do not include the affected area. Pixels of a pattern division region having pixel values of 3, 4, 7, and 8), and pixels of a pattern division region (hereinafter referred to as a second pattern division region) composed of pixels including the influence region (In FIG. 8, the pixels of the pattern division region having pixel values of 5 and 6) are also extracted.

Then, an accumulated value obtained by adding all the luminance values of the pixels in the measurement image corresponding to the pixels of the first pattern division area is obtained, and the obtained cumulative value of the first pattern division area and the pixels of the first pattern division area are obtained. The cumulative value of the second pattern division region is estimated by using the number and the number of pixels of the second pattern division region. Since the ratio of the bright area and the dark area in the first pattern division area and the second pattern division area is the same in the second pattern division area,
(Cumulative value of the second pattern division region) = (cumulative value of the first pattern division region) / (number of pixels in the first pattern division region) × (number of pixels in the second pattern division region)
And you can ask for it.

Then, the accumulated value of the first pattern divided area and the accumulated value of the second pattern divided area are added, and the added value is set as the luminance total value of the beads A.
Similar to the process of obtaining the luminance average value, if there is no first pattern division area, the feature amount of this pattern area is not obtained.

With the above processing, the luminance sum total value of the bead A excluding the influence of the light emission from the bead B can be accurately obtained.
As described above, according to the image processing method according to the first embodiment of the present invention, the pattern area that is the area for obtaining the feature quantity of the measurement target is set from the target image, and this pattern area is further divided into a plurality of pattern division areas. An area that is affected by a measurement target that exists in a pattern area other than the target pattern area is set as an affected area, and pixels that do not include the affected area among the pixels included in the pattern area Pixels that include the measurement object with high accuracy even in the measurement object that is affected by the light from the adjacent measurement object The feature amount using the luminance value at can be obtained.

  The mask pattern used in the image processing method according to the first embodiment is obtained by dividing the pattern area into eight as shown in FIG. 5, but there is no limitation on the number of divisions, and it may be set according to the application. . FIG. 9 shows an example of a mask pattern. The numerical values in the figure indicate the pixel values of each region. For example, as shown in FIG. 9A, a mask pattern that increases the number of divisions and divides the pattern region into 16 may be used. In this way, if the number of divisions is large, it is possible to suppress waste of pixels to be removed when removing a pattern division region composed of pixels including the affected region. Therefore, this leads to an increase in the number of effective pixels, and the feature amount can be obtained with higher accuracy.

  However, if the number of divisions is increased too much, if there is only one pattern division area that does not include the influence area, the effective number of pixels for obtaining the feature amount becomes extremely small, and it is easily affected by noise. The accuracy of the obtained feature amount is deteriorated. Therefore, the number of divisions needs to be set so as to be a region having an area where a feature amount can be obtained with high accuracy even with only pixels in one pattern division region. This may be set according to the accuracy required for the feature quantity to be output.

  Furthermore, in the mask pattern, a specific area may be removed from the area for which the feature amount is obtained, and the pixel value of the removed area may be set to 0. By using such a mask pattern, a feature amount can be obtained using pixels excluding a specific region even in a pattern region. FIG. 9B shows a mask pattern in which a specific area of the mask pattern shown in FIG. Further, in this mask pattern, the specific area is set to 0 so that the area is equal from each area into which the pattern area is divided.

  Then, when the feature amount of the bead is obtained using the mask pattern shown in FIG. 9B, the feature amount is obtained using the luminance value of the pixel in the bright region excluding the dark region in the central portion of the bead. . This is because the feature amount of the bead is obtained in a region excluding a dark region that is susceptible to noise due to the influence of dark current due to the characteristics of the CCD, and can be obtained with higher accuracy.

  Further, in the first embodiment, the case where the measurement object is a bead to which a semiconductor nanocrystal is attached and the luminance value of the pattern region is characteristic has been described. However, the mask pattern shown in FIGS. 5 and 9 is used. Thus, it goes without saying that the feature quantity can be obtained with high accuracy in the same manner even if the feature quantity is obtained for the measurement target of a sphere having a uniform luminance value. Further, in the case of a measurement target having a light emission characteristic different from that of the present bead, a mask pattern for dividing the pattern region may be set according to the light emission characteristic.

  The measurement object of the present invention is not limited to a sphere, and a mask pattern that matches the shape of the measurement object may be prepared. FIG. 10 shows an example of a mask pattern when the measurement target is a square. The numerical value in the figure indicates the pixel value of each region. In the pattern area dividing step, when creating a pattern divided area image, the center of gravity of the area is obtained for each pattern area in the pattern area image, and the mask pattern is placed at the position on the pattern divided area image corresponding to the obtained center of gravity position. In other words, the pixel values of the mask pattern may be written into the pattern division region image for each of the overlapped pixels so that the directions of the center of gravity coincide with each other.

  In addition, when multiple types of measurement objects exist in one measurement image, as many mask patterns as the number of types of measurement objects are prepared, and the types of measurement objects existing in each pattern area are prepared. The pattern area may be divided using the corresponding mask pattern.

  Furthermore, according to the image processing method according to the first embodiment, the influence area is set to the area where the measurement object existing in the pattern area has an influence. You may add an area that is affected by. For example, by adding a region where illumination is reflected or a region influenced by light emission from an object other than the measurement object, the feature amount can be obtained with higher accuracy. As a specific addition method, when creating an influence area image in the influence area setting step, the pixel value corresponding to the influence area to be added in the influence area image may be a value obtained by adding 1 to the number of beads. .

  Next, an image processing apparatus that performs the image processing method according to Embodiment 1 of the present invention will be described. FIG. 11 is a block diagram illustrating a schematic configuration of a main part of the image processing apparatus 100 that performs the image processing method according to the first embodiment of the present invention.

  The image processing apparatus 100 of FIG. 11 is formed by a computer, and includes a setting data storage unit 103, a pattern region setting unit 104, a pattern region dividing unit 105, an affected region setting unit 106, and a feature amount calculating unit 107. Composed.

  The target image 101 input to the image processing apparatus 100 is image data obtained by photographing a measurement target bead with a CCD area sensor and digitizing it, and is an image in which the measurement target bead is reliably captured. Using this target image 101, a pattern area is set.

  The measurement image 102 input to the image processing apparatus 100 is image data obtained by irradiating a measurement target bead with excitation light and photographing the emission from the bead with respect to the excitation light with a CCD area sensor through a bandpass filter. is there. The feature value of each bead is obtained using the luminance value of the pixel of the measurement image 102.

  The setting data storage unit 103 stores setting data such as a template image and a mask pattern necessary for processing performed in the image processing apparatus 100 in advance. The stored setting data is stored in a pattern area as necessary. The data is output to the setting unit 104 and the pattern area dividing unit 105.

  The pattern area setting unit 104 uses the target image 101 and the template image output from the setting data storage unit 103 to extract the position of the measurement target bead existing in the target image 101 and obtain the feature amount of each extracted bead. The area is set as a pattern area, and a pattern area image indicating the pattern area is output to the pattern area dividing unit 105, the affected area setting unit 106, and the feature amount calculating unit 107.

  The pattern area dividing unit 105 divides each pattern area into a plurality of pattern divided areas by using the pattern area image output from the pattern area setting unit 104 and the mask pattern output from the setting data storage unit 103. A pattern division area image indicating each pattern division area is output to the feature amount calculation unit 107.

  Based on the pattern region image output from the pattern region setting unit 104, the influence region setting unit 106 creates an influence region image indicating a region in which the light emission of beads existing in each pattern region has an influence, and a feature amount It outputs to the calculation part 107.

  The feature amount calculation unit 107 includes the measurement image 102, the pattern region image output from the pattern region setting unit 104, the pattern division region image output from the pattern region division unit 105, and the influence region output from the influence region setting unit 106. Using the image, the representative value of each pattern area is obtained from the luminance value of the pixel included in each pattern area, and the obtained representative value is output as the feature value 108.

Using the image processing apparatus 100 configured as described above, the pattern area setting unit 104 performs the process performed in step S101, the pattern area dividing unit 105 performs the process performed in step S102, and the affected area setting unit 106 performs the process in step S103. By performing the processing performed in step S104 in the feature amount calculation unit 107, the feature amount of each pattern region in the input measurement image 102 can be obtained with high accuracy.
[Embodiment 2]
FIG. 12 is a flowchart showing processing performed by the image processing method according to Embodiment 2 of the present invention. The flowchart of FIG. 12 includes steps S201 to S205, where step S201 is a pattern region setting step, step S202 is a pattern region dividing step, step S203 is an influence region setting step, and step S204 including steps S2041 to S2045 is step S204. Step S205 consisting of steps S2051 to S2054 indicates the feature amount calculation step. The difference from the processing of the first embodiment is that a determination process is newly added, and whether the feature amount to be obtained is valid or invalid based on the pattern division area and the influence area in the target pattern area. This is the point that the determination is made and the pattern area is excluded when it is invalid and the feature amount is not obtained.

Hereinafter, image processing performed by the image processing method according to the second embodiment of the present invention on the measurement image shown in FIG. 2 will be described in detail according to a flowchart.
(Step S201 / Pattern area setting step)
First, in the measurement image, a pattern region is set which is a region where a bead exists and a feature amount is obtained. The pattern region setting method is the same as that described in step S101 in FIG.

The pattern area image indicating the pattern area set here is referred to in the following step S202, step S203, and step S204.
(Step S202 / pattern area dividing step)
Next, a process of dividing each pattern region shown in the pattern region image set in step S201 into a plurality of pattern division regions using a mask pattern is performed.

The pattern region dividing method is the same as that described in step S102 in FIG.
The pattern division area image indicating the pattern division area obtained by dividing the pattern area created here is referred to in steps S204 and S205 below.
(Step S203 / influence area setting step)
Next, based on the pattern region image set in step S201, a region where the light emission of the beads existing in the pattern region has an influence on the luminance value of the measurement image is obtained.

Note that the method for obtaining the region affected by the light emission of the beads existing in the pattern region is the same as that described in step S103 in FIG.
The affected area image created here is referred to in the following step S204 and step S205.
(Step S204 / determination step)
Next, based on the pattern region image set in step S201, the pattern division region image created in step S202, and the influence region image created in step S203, It is determined whether it is valid or invalid according to the accuracy of the feature quantity required in the pattern area. If it is invalid, the pattern area is removed.

  More specifically, referring to the pattern area image, the pattern division area image, and the influence area image, the pattern division area and the influence area of one target pattern area are extracted (step S2041), and the extracted pattern division area and The target pattern area is determined based on the information on the affected area (steps S2042 and S2043). If it is determined to be invalid, the value of each pixel of the pattern area image corresponding to the target pattern area is changed to 0 in order to remove the pattern area (step S2044). As a result, there is no pixel value having a bead number existing in the pattern area determined to be invalid on the pattern area image indicating the pattern area, and the feature amount in the pattern area determined to be invalid is not obtained. Can do. This series of processing is performed on all pattern areas (step S2045), and a pattern area image excluding information on invalid pattern areas is obtained.

Hereinafter, the determination process in step S204 will be described in detail.
FIG. 13 is a diagram showing a pattern division area and an influence area in the vicinity of the bead E in the bead E shown in FIG. The numerical value in the figure indicates the pixel value of the pattern divided region image.

  In the case of the bead E, when the feature amount is obtained using only the pixels of the pattern division area that does not include the influence area in each pattern division area, only the pixel values of the pattern division area having the pixel values of 1 and 2 indicated by diagonal lines are used. It will be.

  As described above, when the number of pixels used for obtaining the feature amount is small, the number of effective pixels is small, so that it is easily affected by noise, and is obtained in the best case where there is no adjacent bead and all the pixels in the pattern area are used. It is considered that the difference is larger than the feature amount, and it is generally better not to output the feature amount whose accuracy is lowered.

  Therefore, the number of pattern division areas composed of pixels including the influence area is counted, and when the number is larger than a predetermined threshold, it is determined that the pattern area is invalid. The predetermined threshold value to be compared with the number of divided areas of the pattern divided area can be set to any value from 0 to the number of divided pattern areas according to the accuracy required for the output feature quantity. Good. When high accuracy is required, the predetermined value is increased, and in the opposite case, the predetermined value may be set small.

  Further, by performing such a determination process, even if the area of the pattern division region is reduced in order to eliminate waste of pixels to be removed from the pattern region, the feature amount can be accurately adjusted by adjusting the threshold value to be set. It is possible to ensure the minimum number of pixels that can be obtained.

  Note that the determination criterion value may be the number of pixels included in the pattern division region including the pixels including the affected region. In step S2042, the number of pixels is counted. In step S2043, the number of pixels is greater than a predetermined threshold value. If it is larger, this pattern area may be determined to be invalid.

  Further, the validity / invalidity determination may be performed by using the number of pattern division areas configured by pixels not including the affected area or the number of pixels included in the pattern division area configured by the pixels not including the affected area. When the number or the number of pixels counted in step S2042 is smaller than a predetermined threshold set for each value in step S2043, this pattern area may be determined to be invalid.

The determination is made as described above, and the pattern area image from which the pattern area determined to be invalid is removed is referred to in step S205.
(Step S205 / feature amount calculation step)
Next, using the pattern area image created in step S204, the pattern division area image created in step S202, and the influence area image created in step S203, the luminance values of the pixels included in each pattern area are The feature amount of the pattern area is calculated.

Note that the feature amount calculation method for each pattern region is the same as that described in step S104 of FIG.
As described above, according to the image processing method according to the second embodiment of the present invention, in each pattern area for which the feature amount is to be determined, whether the feature amount to be calculated in the target pattern area is valid or invalid is determined. Thus, the feature amount of the pattern area determined to be invalid is not obtained, so that the analysis can be performed using only the feature amount obtained with higher accuracy.

Next, an image processing apparatus that implements the image processing method according to Embodiment 2 of the present invention will be described.
FIG. 14 is a block diagram illustrating a schematic configuration of a main part of an image processing apparatus 200 that performs the image processing method according to the second embodiment of the present invention.

  14 is formed by a computer, and includes a setting data storage unit 103, a pattern region setting unit 104, a pattern region dividing unit 105, an affected region setting unit 106, a feature amount calculating unit 107, and a determination. The unit 209 is configured. The same components as those described in the first embodiment are denoted by the same reference numerals, and description thereof is omitted.

  A difference from the configuration of the first embodiment is that a determination unit 209 is newly added, and a pattern region image indicating a pattern region set by the pattern region setting unit 104 is converted into a pattern region dividing unit 105, an affected region setting unit 106, and a determination. Output to the unit 209 and output to the feature amount calculation unit 107 and the determination unit 209 the pattern division region image indicating each pattern division region obtained by dividing the pattern region into a plurality of patterns by the pattern region division unit 105 and set by the influence region setting unit 106 The influence area image indicating the affected area is output to the feature amount calculation unit 107 and the determination unit 209.

  The determination unit 209 is based on the pattern region image output from the pattern region setting unit 104, the pattern division region image output from the pattern region division unit 105, and the influence region image output from the influence region setting unit 106. In each pattern area for which a feature amount is to be obtained, whether the feature amount is accurate or not is determined according to the accuracy of the feature amount to be obtained in the target pattern area. If invalid, the pattern area is excluded, and all pattern areas are excluded. The pattern area image after the determination is output to the feature amount calculation unit 107.

  Using the image processing apparatus 200 configured as described above, the pattern area setting unit 104 performs the process performed in step S201, the pattern area dividing unit 105 performs the process performed in step S202, and the affected area setting unit 106 performs the process in step S203. By performing the processing performed in step S204 in the determination unit 209 and the processing performed in step S205 in the feature amount calculation unit 107, the processing is further performed in the feature amount of each pattern region obtained for the input measurement image 102. It is possible to perform analysis using only the feature quantity obtained with high accuracy.

  The image processing method according to the present invention can accurately obtain a feature amount using luminance information of an object from a digital image obtained by photographing the object, and measures various feature amounts of a measurement object in the object image. Useful for processing equipment.

It is a flowchart explaining the image processing method in Embodiment 1 of this invention. It is a figure which shows an example of the measurement image of the image processing method. It is a figure which shows an example of the target image of the image processing method. It is a figure which shows the pattern area image set in the pattern area setting process of the image processing method. It is a figure which shows an example of the mask pattern in the image processing method. It is a figure which shows the pattern division area image produced at the pattern area division process of the image processing method. It is a figure which shows the influence area image set in the influence area setting process of the image processing method. It is a figure which shows the pattern division | segmentation area | region and influence area | region of the bead A vicinity in the image processing method. It is a figure which shows another example of the mask pattern in the image processing method. It is a figure which shows an example of a mask pattern in case the measuring object is a square in the image processing method. It is a block diagram of the image processing apparatus which implements the image processing method. It is a flowchart explaining the image processing method in Embodiment 2 of this invention. It is a figure which shows the pattern division area | region and influence area | region of the bead E vicinity in the image processing method. It is a block diagram of the image processing apparatus which implements the image processing method. It is a figure which shows the procedure of a general template matching.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100,200 Image processing apparatus 101 Target image 102 Measurement image 103 Setting data storage part 104 Pattern area setting part 105 Pattern area division part 106 Influence area setting part 107 Feature amount calculation part 108 Feature quantity 209 Determination part 1501 Template image 1502 Target image 1503 , 1504 Detection object 1505 First pattern matching position 1506 Pattern matching scanning path 1507 Position where pattern is detected

Claims (30)

One or a plurality of specific pattern areas are set for a digital image having multivalued luminance values in each pixel, and each pattern area is set using the luminance values of the pixels included in the pattern area in the digital image. An image processing method for calculating a representative value in
A pattern region setting step of setting a region where a measurement object exists for the digital image as a pattern region;
A pattern region dividing step of dividing the pattern region set in the pattern region setting step into a plurality of pattern divided regions;
In the pattern area set in the pattern area setting step, a measurement object existing in a pattern area other than the target pattern area has an influence on the luminance value of the digital image as an influence area. Influence area setting process to be set;
Of the pixels included in the pattern area set in the pattern area setting step, the pattern divided in the pattern area dividing step is composed only of pixels not including the affected area set in the affected area setting step. And a feature amount calculating step of calculating a representative value in the pattern area using only pixels in the divided area. The image processing method according to claim 1, wherein in the pattern area dividing step, the pattern area is divided using a mask pattern indicating a plurality of divided areas. The image processing method according to claim 2, wherein the mask pattern has a divided area indicating a pattern area, and pixels corresponding to the divided area are used as the pattern divided area. The image processing method according to claim 1, wherein in the pattern region dividing step, the pattern region is divided into a plurality of pattern divided regions so that the areas of the divided regions become equal. The image processing method according to claim 1, wherein in the pattern region dividing step, the pattern region is divided into a plurality of pattern divided regions by a straight line passing through a center of the pattern region. 6. The image processing method according to claim 5, wherein in the pattern region dividing step, the image is divided into a plurality of pattern divided regions so that the central angles are equal along a straight line passing through the center of the pattern region. The image processing method according to claim 1, wherein in the pattern region dividing step, a specific region is excluded from a plurality of pattern divided regions obtained by dividing the pattern region. 8. The image processing method according to claim 7, wherein the specific area excluding the pattern division area is set so that the area is equal in each divided pattern division area. In the influence area setting step, the influence area is set as a neighboring area including a pattern area other than the target pattern area, and pixels that form a contour in a connected component in which the pixels constituting the pattern area are grouped are arranged outside. The image processing method according to claim 1, wherein the connected component after the increase by a predetermined number of pixels is set. 10. The number of pixels for increasing the number of pixels to be the contour to the outside is set based on the magnitude of the value obtained using the luminance value of each pixel of the connected component whose contour is to be increased. Image processing method. The area of influence other than the measurement object existing in the pattern area is added to the influence area of the measurement object existing in the pattern area in the influence area setting step. Image processing method. In the feature amount calculating step, the representative value in the pattern region is an average luminance value obtained by averaging the luminance values of the pixels included in the pattern region,
A cumulative value obtained by adding all the luminance values of the pixels in the pattern division area composed only of pixels not included in the affected area is obtained, and the representative value is divided by the number of pixels obtained by adding the obtained cumulative values. The image processing method according to claim 1, wherein the image processing method is obtained by: In the feature amount calculating step, the representative value in the pattern area is a luminance sum total value obtained by summing the luminance values of the pixels included in the pattern area,
The number of pixels obtained by adding together the cumulative value obtained by adding all the luminance values of the pixels in the first pattern division area constituted by only the pixels not including the influence area and the pixel including the influence area is obtained. Obtain the number of pixels of the second pattern division region,
Next, the cumulative value of the second pattern division region is estimated based on the cumulative value of the first pattern division region, the number of pixels of the first pattern division region, and the number of pixels of the second pattern division region,
The image processing method according to claim 1, wherein the representative value is obtained by adding together the accumulated value of the second pattern divided area analogized with the accumulated value of the first pattern divided area. Determining whether the representative value obtained in the target pattern area is valid or invalid by using the information of the pattern division area divided in the pattern area division step and the influence area set in the influence area setting step With a process,
The image processing method according to claim 1, wherein the determination step is performed immediately before the feature amount calculation step, and the representative value in the pattern area determined to be invalid is not obtained. 15. The image processing method according to claim 14, wherein in the determination step, the determination is performed based on the number of pattern division regions or the number of pixels configured with pixels including or not including the affected region. One or a plurality of specific pattern areas are set for a digital image having multivalued luminance values in each pixel, and each pattern area is set using the luminance values of the pixels included in the pattern area in the digital image. An image processing apparatus for calculating a representative value in
A pattern region setting unit that sets a region where a measurement object exists for the digital image as a pattern region;
A pattern region dividing unit for dividing the pattern region set in the pattern region setting unit into a plurality of pattern divided regions;
In the pattern area set in the pattern area setting unit, the measurement object existing in the pattern area other than the target pattern area has an influence on the luminance value of the digital image as an influence area. An influence area setting section to be set;
The pattern divided in the pattern area dividing unit, which is composed only of pixels included in the pattern area set in the pattern area setting unit and not including the affected area set in the affected area setting unit. An image processing apparatus comprising: a feature amount calculation unit that calculates a representative value in a pattern region using only pixels in a divided region. The image processing apparatus according to claim 16, wherein the pattern area dividing unit divides the pattern area using a mask pattern indicating a plurality of divided areas. The image processing apparatus according to claim 17, wherein the mask pattern has a divided area indicating a pattern area, and pixels corresponding to the divided area are used as the pattern divided area. The image processing apparatus according to claim 16, wherein the pattern region dividing unit divides the pattern region into a plurality of pattern division regions so that the areas of the divided regions are equal. The image processing apparatus according to claim 16, wherein the pattern region dividing unit divides the pattern region into a plurality of pattern division regions by a straight line passing through a center of the pattern region. 21. The image processing apparatus according to claim 20, wherein the pattern region dividing unit divides the pattern region into a plurality of pattern division regions so that the central angles are uniform along a straight line passing through the center of the pattern region. The image processing apparatus according to claim 16, wherein the pattern area dividing unit excludes a specific area from a plurality of pattern division areas obtained by dividing the pattern area. 23. The image processing apparatus according to claim 22, wherein the specific area excluding the pattern division area is set to have the same area in each divided pattern division area. In the influence area setting unit, the influence area is set as a vicinity area including a pattern area other than the target pattern area, and pixels that are outlines in a connected component including a group of pixels constituting the pattern area are arranged outside. The image processing apparatus according to claim 16, wherein the connected component after being increased by a predetermined number of pixels is set. 25. The number of pixels that increase the number of pixels that form the contour to the outside is set based on the magnitude of a value obtained by using the luminance value of each pixel of the connected component for which the contour is to be increased. Image processing apparatus. The said influence area setting part adds the area | region which things other than the measurement target object which exists in the said pattern area affects to the influence area by the measurement target object which exists in a pattern area | region. Image processing device. In the feature amount calculation unit, the representative value in the pattern region is an average luminance value obtained by averaging the luminance values of the pixels included in the pattern region,
A cumulative value obtained by adding all the luminance values of the pixels in the pattern division area composed only of pixels not included in the affected area is obtained, and the representative value is divided by the number of pixels obtained by adding the obtained cumulative values. The image processing device according to claim 16, wherein the image processing device is obtained by: In the feature amount calculation unit, the representative value in the pattern region is a luminance sum value obtained by summing the luminance values of the pixels included in the pattern region,
The number of pixels obtained by adding together the cumulative value obtained by adding all the luminance values of the pixels in the first pattern division area constituted by only the pixels not including the influence area and the pixel including the influence area is obtained. Obtain the number of pixels of the second pattern division region,
Next, the cumulative value of the second pattern division region is estimated based on the cumulative value of the first pattern division region, the number of pixels of the first pattern division region, and the number of pixels of the second pattern division region,
The image processing apparatus according to claim 16, wherein the representative value is obtained by adding the accumulated value of the second pattern divided area analogized with the accumulated value of the first pattern divided area. Judgment for determining whether the representative value obtained in the target pattern area is valid or invalid by using the information of the pattern division area divided in the pattern area division unit and the influence area set in the influence area setting unit Part
The image processing apparatus according to claim 16, wherein the representative value in the pattern area determined to be invalid by the determination unit is not obtained. 30. The image processing apparatus according to claim 29, wherein the determination unit performs the determination based on the number of pattern division regions or the number of pixels including pixels that include or do not include an affected region.

Images