JP5652243B2 - Image processing apparatus and image processing method - Google Patents

Description

  The present invention relates to an image processing apparatus and an image processing method for performing resolution conversion on image data.

  2. Description of the Related Art Conventionally, there is a technique for performing resolution conversion for image data in which a color for each pixel is represented by a gradation value. For example, in a certain prior art, after calculating pixel positions of a plurality of pixels constituting an image after resolution conversion (hereinafter referred to as “new pixel position”), the pixel values of these pixels are converted to those before resolution conversion. Calculation is performed by interpolation based on the image.

  However, when the pixel value of each pixel is obtained by calculation, the load on the system is large and processing takes time.

  On the other hand, there is a method of calculating a pixel value at a new pixel position by using a weighted sum using the coefficients (filter coefficients) prepared in advance. In that method, when calculating the pixel value of a certain new pixel position where the pixel (old pixel) constituting the image before resolution conversion does not exist at that position, the old pixel existing in the vicinity of the new pixel position is calculated. There is a method of multiplying a plurality of pixel values by those weighting factors and calculating a pixel value at a new pixel position by a weighted sum thereof.

  However, the weight coefficient prepared in advance can be prepared only for a predetermined pixel position (for example, a position of 1/4, 2/4, 3/4 from the left in one pixel section of the old pixel). . For this reason, a weight suitable for an arbitrary magnification cannot be provided. Therefore, when the pixel value at the new pixel position is calculated using a weighting factor prepared in advance, in most cases, the quality of the generated image is lower than when calculating the pixel value of each pixel every time. Become.

  In order to solve this problem, when performing resolution conversion on certain image data, there is known an image processing for reducing a decrease in image quality while reducing a load by using a coefficient prepared in advance. . (See Patent Document 1)

JP 2010-114767 A

  According to the image processing apparatus described in Patent Document 1, when resolution conversion is performed on certain image data, a reduction in image quality is reduced while reducing a load by using a filter coefficient prepared in advance. Although it is possible, a large number of filter coefficients (4 times the number of divisions) are required as filter coefficients, and the load of the arithmetic processing is large.

  An object of the present invention is to provide an image processing apparatus and an image processing method capable of reducing the calculation processing load without requiring a large number of filter coefficients for conversion to image data having a different number of pixels. .

  An image processing apparatus according to the present invention includes an original image size storage unit that stores an image size of original image data, and an enlargement that stores an image size of enlarged image data generated by enlarging the image size of the original image data. A post-image size storage unit, a calculation unit that calculates an enlargement rate from the image size of the original image data and the image size of the post-enlargement image data, an enlargement rate storage unit that stores the enlargement rate calculated by the calculation, A fixed-point notation storage unit that stores a fixed-point notation of the reciprocal of the enlargement ratio, a rounded-up fixed-point notation storage unit that stores a fractional part obtained by rounding the fixed-point notation, and a pixel position after enlargement A fixed-point notation storage unit that stores a fixed-point notation that represents the pixel position of the corresponding original image, and a fixed-point notation data that represents the pixel position of the original image. Characterized in that it comprises an interpolation processor for interpolating process of adding or subtracting the bit cumulatively.

  The image processing apparatus of the present invention further includes a cycle setting register that sets the predetermined number of pixels, and a bit number storage register that stores the predetermined number of bits, and the interpolation processing unit includes the original image data More preferably, the predetermined number of bits are added to or subtracted from the lower bits of the horizontal and / or vertical addresses of the pixels for each of the predetermined number of periodic pixels.

  The image processing method of the present invention includes: obtaining an image size of the original image data; obtaining an image size of the enlarged image data; and an image size of the original image data and an image size of the enlarged image data. A step of calculating a magnification ratio, a step of obtaining a fixed-point notation of the reciprocal of the magnification ratio, a step of rounding the fixed-point notation to obtain a fractional part, and a fixed representing the pixel position of the original image A step of cumulatively adding or subtracting a predetermined number of bits to the decimal point notation data for each predetermined number of periodic pixels.

  The predetermined period is a period in which the decimal part data expressed in the fixed point is an integer.

  When performing conversion (resolution conversion) to image data having a different number of pixels, a large number of filter coefficients are not required, and the load of the arithmetic processing can be reduced.

1 is a block diagram illustrating a configuration of an image processing apparatus 10 according to an embodiment of the present invention. It is a flowchart which shows an example of the procedure of the resolution conversion of the image data by the image processing apparatus 10 which concerns on embodiment of this invention. It is a figure for demonstrating the principle of the resolution conversion at the time of the expansion conversion of the image data by the image processing apparatus 10 which concerns on embodiment of this invention, (a) shows an example of an original image, (b) is (a). FIG. 4C shows an enlarged image obtained by enlarging the original image in the horizontal direction, and FIG. 5C schematically shows interpolation processing for pixels of the enlarged image. It is a figure which shows the result of the resolution conversion when not performing the interpolation process by the image processing apparatus 10 which concerns on embodiment of this invention. It is a figure which shows the result of the resolution conversion at the time of performing the interpolation process by the image processing apparatus 10 which concerns on embodiment of this invention.

  FIG. 1 is a block diagram illustrating a configuration of an image processing apparatus 10 according to an embodiment of the present invention.

  The image processing apparatus 10 according to the present embodiment includes a calculation unit (CPU) 10a that performs various calculations associated with image data conversion, a RAM 10b, a ROM 10c, and an original image size storage unit that stores the image size of the original image data. 10d, an enlarged image size storage unit 10e that stores the image size of the enlarged image data, an enlargement rate storage unit 10f that stores an enlargement rate calculated from the original image size and the enlarged image size, and a reciprocal of the enlargement rate A fixed-point notation storage unit 10g for storing a fixed-point notation, a rounded fixed-point notation storage unit 10h for storing a fractional part obtained by rounding the fixed-point notation, a fractional part bit storage unit 10i, and an expanded fixed-point A notation storage unit 10j for calculating a fixed-point notation representing a pixel position of the original image corresponding to each pixel position after enlargement; It is connected to the bus with the face 10p.

  When the image processing apparatus 10 according to the present embodiment enlarges and converts image data from an original image to an enlarged image, for example, first, the image size of the original image data (original image data) is stored in the original image size storage unit 10d. Is done. Further, for example, the image size of the image data after enlargement conversion (image data after enlargement) set in advance by the user is stored in the image size storage unit 10e after enlargement. The calculation unit 10a calculates an enlargement ratio from the image size of the original image data stored in the original image size storage unit 1d and the image size of the enlarged image data stored in the enlarged image size storage unit 10e, and enlarges the image. Store in the rate storage unit 10f.

  For example, when the division number from 0 to 1 is 32, the arithmetic unit 10a derives 65536, which is the fractional part bit number representing “1” that is an integer, and stores the fractional part bit number. Stored in the unit 10i. The calculation unit 10a calculates a fixed-point notation of the reciprocal of the calculated enlargement ratio and stores it in the fixed-point notation storage unit 10g. Then, the arithmetic unit 10a rounds the calculated fixed-point notation to obtain a decimal part, and stores the decimal part in the rounded fixed-point notation storage unit 10h. Finally, the calculation unit 10a calculates a fixed-point notation representing the pixel position of the original image corresponding to each pixel position after enlargement, and stores the fixed-point notation storage part 10j after enlargement.

  In addition, the image processing apparatus 10 according to the present embodiment includes an interpolation processing unit 10k that performs an interpolation process for adding (or subtracting) a predetermined number of bits for each predetermined pixel period to the pixels of the enlarged image data; A pixel cycle / bit number setting unit 10r for setting the pixel cycle and the number of bits, a pixel cycle storage register 10m for storing the pixel cycle set by the pixel cycle / bit number setting unit 10r, and a pixel cycle / bit number A bit number storage register 10n for storing the number of bits set by the setting unit 10r. In the image processing apparatus 10 of this example, the pixel cycle / bit number setting unit 10r sets the pixel cycle and the bit number, but the setting method is not particularly limited. That is, the pixel period / bit number setting unit 10r may set the pixel period and the number of bits based on a predetermined program, or may be set based on an input by a user.

  FIG. 2 is a flowchart showing a processing procedure for converting an original image of 1024 × 768 pixels into an image of 1536 × 768 pixels by the image processing apparatus 10, that is, a processing procedure for enlarging and converting image data in the horizontal direction. is there. FIG. 3 is a diagram for explaining the principle of resolution conversion when the image processing apparatus 10 enlarges and converts image data based on the flow of FIG. 2. FIG. 3A is an example of an original image (1024 × (B) shows an enlarged image (1536 × 768 pixels) obtained by enlarging the original image of (a) in the horizontal direction, and (c) shows resolution conversion for the pixels of the enlarged image. An interpolation process performed at this time is schematically shown.

  In this flow, first, the arithmetic unit 10a first displays the image size (horizontal direction: 1024 pixels, vertical direction: 768 pixels) of the original image data (see FIG. 3A) and the enlarged image data (see FIG. 3B). ) (Horizontal direction: 1536 pixels, vertical direction: 768 pixels) are acquired and stored in the original image size storage unit 10d and the enlarged image size storage unit 10e, respectively (step S1).

  Next, the pixel cycle / bit number setting unit 10r sets the pixel cycle (= 3) to be interpolated and stores it in the pixel cycle storage register 10m, and further adds (subtracts) the bit number ( = 2) is set and stored in the bit number storage register 10n (step S2).

  Then, the arithmetic unit 10a stores 65536, which is the decimal part bit number representing the integer “1”, in the decimal part bit number storage unit 10i (step S3).

  Next, the calculation unit 10a calculates a horizontal enlargement ratio 1.5 from the image size of the original image data and the image size of the enlarged image data, and stores it in the enlargement ratio storage unit 10f (step S4).

  Next, the calculation unit 10a calculates a fixed point notation (= 43690.6667) of the reciprocal (= 0.6666666667) of the enlargement ratio (= 1.5) calculated by the calculation and stores it in the fixed point notation storage unit 10g. (Step S5).

  Further, the arithmetic unit 10a rounds the fixed-point notation (= 43690.6667) to obtain a decimal part (= 43690) and stores it in the rounded fixed-point notation storage unit 10h (step S6).

  As shown schematically in FIG. 3C, the interpolation processing unit 10k represents the pixel position of the original image when calculating a fixed-point notation representing the pixel position of the original image corresponding to each pixel position after enlargement. An interpolation process is executed in which the number of bits (= 2) stored in the bit number storage register 10n is cumulatively added to the fixed-point notation bits for each pixel period (= 3) stored in the pixel period storage register 10m. (Step S7).

  Then, the fixed point notation representing the pixel position of the original image corresponding to all the enlarged pixel positions is calculated and stored in the enlarged fixed point notation storage unit 10j (step S8). This flow is completed by the above.

  Next, characteristics of the interpolation processing will be described in comparison with the case where the resolution conversion of the image data is performed without performing the interpolation processing by the image processing apparatus 10.

  FIG. 4 is a table showing the result of resolution conversion when the interpolation processing by the image processing apparatus 10 is not performed. More specifically, FIG. 4 shows the pixel position (the pixel position after enlargement) in the enlarged image after enlargement conversion from the original image (1024 × 768 pixels) to the enlarged image (1536 × 768 pixels), the pixels of the original image Position (original pixel position), integer part of original image pixel position (original pixel position (integer part)), decimal part of original image pixel position (original pixel position (decimal part)), fixed decimal notation (decimal part 15 (Bit) is a table showing the relationship between each value of an actual original pixel position by bits and an ideal original pixel position by fixed point notation (decimal part 15 bits) for the expanded pixels (1 to 1536 pixels) (however, In the table of FIG. 4, among the pixels in the enlarged image, only the relationship between the above values is shown for 1 to 50 pixels, 1022 to 1026 pixels, and 1522 to 1536 pixels. Here, when dividing 0 to 1 into 32, the decimals between 0 and 1 and the fixed decimals corresponding to these decimals have the relationship shown on the left side of the table of FIG.

  Focusing on the relationship between the actual original pixel position in the fixed decimal notation and the ideal original pixel position in the fixed-point notation in the processing results shown in the table of FIG. 4, the pixel position of the enlarged image is 2 bits per 3 pixels. It can be seen that there is a shift each time and that the shifts are cumulative.

  Then, by accumulating the deviation, the original pixel position in the fixed decimal notation (actual value) becomes 43008 or less at the 1024th pixel after the enlargement, and from the relation between the decimal and the fixed decimal, it becomes 0.68625. What is supposed to be is 0.65625.

  That is, the relationship between the original pixel position (decimal part 15 bits) in the fixed decimal notation (actual value) and the original pixel position (decimal part 15 bits) in the fixed-point notation (ideal) is 2 per 3 pixels of the enlarged pixels. Deviations occur bit by bit, and the deviations are accumulated. By accumulating such shifts, an enlarged image obtained by enlarging the original image becomes an image in which a pixel shift occurs with respect to the original image.

  On the other hand, by performing the above-described interpolation processing by the image processing apparatus 10 according to the present embodiment, as shown in FIG.

  FIG. 5 is a table showing resolution conversion processing results when the above-described interpolation processing is performed by the image processing apparatus 10. More specifically, FIG. 5 shows the pixel position (pixel position after enlargement) in the enlarged image after enlargement conversion from the original image (1024 × 768 pixels) to the enlarged image (1536 × 768 pixels), and the pixels of the original image Position (original pixel position), integer part of original image pixel position (original pixel position (integer part)), decimal part of original image pixel position (original pixel position (decimal part)), fixed decimal notation (decimal part 15 Bit), the actual original pixel position in fixed decimal notation (15 bits in decimal part), the value obtained by adding 2 bits every 3 pixels to the actual original pixel position in fixed decimal notation (15 bits in decimal part), and the ideal in fixed point notation (15 bits in decimal part) FIG. 5 is a table showing the relationship between each value of the original pixel position for each pixel (1 to 1536 pixels) in the enlarged image (however, in the table of FIG. 5, 1 to 15 pixels among the pixels in the enlarged image, 1022-1026 pixels, And the relationship between the above values is shown only for 1531 to 1536 pixels.) Here, when dividing 0 to 1 into 32, the decimals between 0 and 1 and the fixed decimals corresponding to these decimals have the same relationship as in FIG.

  In the processing results shown in the table of FIG. 5, a value obtained by adding (adding) 2 to every three pixels in the original pixel position (15 bits of the decimal part) in the fixed decimal notation (actual value) and the fixed decimal notation (ideal) When the above interpolation process is not performed (FIG. 4), the relationship with the original pixel position (decimal part 15 bits) is shifted by 2 bits for each of the three enlarged pixels, but the deviation is eliminated by performing the above interpolation process.

  In the interpolation processing for obtaining the processing results shown in the table of FIG. 5, the original image (1024 × 768 pixels) is converted to the enlarged image (1536 × 768 pixels), so 2 bits are added for every three pixels. However, the number of bits to be added in the interpolation process is not limited to this. For example, if the original image (1024 × 768 pixels) is different from the enlarged image (1536 × 768 pixels), the pixel cycle and the number of bits stored in the pixel cycle storage register 10m and the bit number storage register 10n are also different.

  Further, the above-described image data conversion processing by the image processing apparatus 10 according to the present embodiment can be executed by a program stored in a computer.

DESCRIPTION OF SYMBOLS 10 ... Image processing apparatus, 10a ... Arithmetic unit (CPU), 10b ... RAM, 10c ... ROM, 10d ... Original image size storage unit, 10e ... Enlarged image size storage unit, 10f ... Enlargement rate storage unit, 10g ... Fixed point Notation storage unit, 10h: Rounded fixed point notation storage unit, 10i: Decimal part bit number storage unit, 10j: Expanded fixed point notation storage unit, 10k ... Interpolation processing unit, 10m ... Pixel cycle storage register, 10n ... Bit number storage Register, 10p ... Input / output interface, 10r ... Pixel cycle / bit number setting unit

Claims (6)

An original image size storage unit for storing the image size of the original image data;
An enlarged image size storage unit for storing the image size of the enlarged image data generated by enlarging the image size of the original image data;
An arithmetic unit that calculates a first fixed-point notation corresponding to the pixel position of the original image corresponding to each pixel position after enlargement based on the image size of the original image data and the image size of the enlarged image data;
A rounded fixed-point notation storage unit for storing a second fixed-point notation obtained by rounding the first fixed-point notation;
An interpolation processing unit that performs an interpolation process for cumulatively adding or subtracting a predetermined number of bits for each predetermined number of periodic pixels to the second fixed-point notation;
A setting unit that sets the predetermined number of bits according to the image size of the original image data and the image size of the enlarged image data;
An image processing apparatus comprising: A pixel cycle storage register for storing the predetermined cycle pixel number;
A bit number storage register for storing the predetermined number of bits;
Further comprising
The interpolation processing unit
2. The image processing apparatus according to claim 1, wherein the predetermined number of bits is added to or subtracted from a lower bit of a horizontal and / or vertical address of a pixel for each predetermined number of pixels in the original image data. .   The image processing apparatus according to claim 1, wherein the setting unit sets the predetermined number of periodic pixels according to an image size of the original image data and an image size of the enlarged image data. . Obtaining the image size of the original image data;
Obtaining an image size of the enlarged image data generated by enlarging the image size of the original image data;
Obtaining a first fixed-point notation corresponding to the pixel position of the original image corresponding to each pixel position after enlargement based on the image size of the original image data and the image size of the enlarged image data;
Rounding the first fixed-point notation to obtain a second fixed-point notation;
Setting a predetermined number of bits according to the image size of the original image data and the image size of the enlarged image data;
Wherein the second fixed-point notation, and the step of the predetermined number of bits to add or subtract cumulatively at every predetermined number cycles of pixels,
An image processing method comprising: The image processing method according to claim 4 , wherein the predetermined cycle is a cycle in which the decimal part data represented by the fixed point is an integer.   6. The image processing method according to claim 4, further comprising the step of setting the predetermined number of pixels according to an image size of the original image data and an image size of the enlarged image data.

Images