JP2003023536A - Image correcting method and system thereof - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image correcting method and system, capable of removing any pattern deviation generated in a pattern image by reproducing the texture of an image itself at the part causing the pattern deviation. SOLUTION: At designation of a region to be corrected on an original image, and deciding each pixel in the designated region to be corrected; a mask in a prescribed shape is arranged on a reduced correction image region, corresponding to the region to be corrected (a), and masks in the same shape as that of the mask are arranged on the reduced original images (b-e). While the position of the mask on the reduced correction image region is fixed, the masks on the reduced original images are moved by one pixel each, the similarity of the region masked on the reduced correction image region and the regions masked on the reduced original images is calculated, the pixels, having the same position relation with the mask on the reduced original image (e) whose similarity, is decided to be the highest as a position relation between the mask on the reduced correction image region, and the pixels to be decided are assigned as pixels on the reduced correction image region.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for correcting an image of a specific portion such as a pattern habit generated on an image used for printing a building material.

[0002]

2. Description of the Related Art Conventionally, various pattern images have been used to decorate the surface of building materials and the like. Such a pattern image is obtained by correcting a photographed image and creating it or artificially creating it by a computer operation. By repeatedly arranging the pattern image thus created, an unintended pattern often called a pattern habit is generated, which may impair the aesthetic appearance of the pattern image.

When the above-mentioned pattern habit is generated, conventionally,
The operator directly performed retouching work using commercially available image processing software. In addition, some methods for automatically removing the pattern habit using computer software have been proposed.

[0004]

However, in the above-described conventional pattern habit removing method, when the operator performs retouching work, it is performed based on the subjectivity of the operator. Therefore, the texture of the entire image is corrected by the corrected portion. It may break. In addition, the present condition is that the method automatically performed by a computer has not been satisfactory.

In view of the above points, in the present invention, when a pattern habit is generated in a pattern image, the texture of the image itself is reproduced in a portion which causes the pattern habit, thereby creating the pattern habit. An object is to provide an image correction method and system for removing.

[0006]

In order to solve the above problems, in the present invention, an original image having a portion to be corrected is input, a correction target area is designated on the input original image, and the designated correction is performed. Each pixel of the target area is determined. To determine each pixel of the correction target area, a mask having a predetermined shape is arranged on the correction target area, and a mask having the same shape as the mask is arranged on the original image. While the position of the mask on the correction target area is fixed, the degree of similarity between the masked area on the correction target area and the masked area on the original image is obtained by moving the mask on the original image pixel by pixel. , The pixel whose positional relationship with the mask on the original image determined to have the highest degree of similarity is the same as the positional relationship between the mask and the pixel to be determined on the correction target area, Divided as pixels on the area It is those shed, portions masks placed on the correction object region protrudes from the correction target region, characterized in that to perform by using the pixel on the original image. According to the present invention, the pixel on the original image is assigned to the correction target area by obtaining the similarity between the specified correction target area and the area masked by the portion other than the correction area on the original image. , In particular, when the mask to be placed on the correction target area extends out of the correction target area, the pixels on the original image are used as they are, so that the correction target area is newly added with the characteristics of the original image. It is created, and the joint part of the correction target area becomes inconspicuous.

[0007]

DETAILED DESCRIPTION OF THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. FIG. 1 is a flowchart showing an outline of an image correction method according to the present invention. First, an original image in which a portion to be corrected such as a pattern habit is generated is displayed on the screen of the display device, and a correction target area in which the pattern habit is generated is designated (step S).
1). The state at this time is shown in FIG. Figure 2 (a)
In the above, sx and sy are the numbers of pixels in the horizontal and vertical directions of the correction target area, respectively.

Next, regarding the designated correction target area,
A resolution reduction rate, which is a rate at which the resolution is reduced, is set (step S2). The resolution reduction rate set here sets how many pixels of the original image are dropped to one pixel.
This is done mainly for the purpose of reducing the calculation amount by reducing the number of pixels.

Next, a reduced original image in which the resolution of the original image is reduced is created at the set resolution reduction rate (step S3). The reduced original image is created by giving an average value of horizontal r × vertical r pixels on the original image as a pixel value of one pixel of the reduced original image, where r is the resolution reduction rate. At this time, each pixel of the reduced original image is recorded together with the given average value, and which r × r pixels of the original image the pixel was obtained from are recorded. This is used later when restoring to the original size.

Then, the size of the reduced corrected image area obtained by reducing the image size (the number of pixels) of the area to be corrected specified in step S1 by using the resolution reduction rate set in step S2 is obtained. An area having a size obtained by adding two pixels to the upper, lower, left, and right sides of the reduced correction image area is secured in a working storage area such as an image memory (step S
4). Here, if the resolution reduction rate r is used, FIG.
The horizontal direction rx and the vertical direction ry of the reduced corrected image area shown in (a) are calculated by the following [Formula 1].

[Formula 1] rx = sx / r ry = sy / r

Based on this, by adding two pixels in the upper, lower, left and right shaded in FIG. 3B, horizontal (rx + 4) × vertical (ry + 4) pixels are secured in the work area of the image memory. Will be.

Next, the pixel value of the corresponding portion of the original image is given to the area added around the reduced corrected image area as shown in FIG. 3B (step S5). In particular,
Two pixels in the upper, lower, left, and right of the correction target area on the reduced original image as shown by hatching in FIG. 2B are used. That is, by duplicating the pixels in the shaded area in FIG. 2B, values are given to the pixels in the shaded area in FIG. 3B.

After the pixel values of the two pixels above, below, to the left and right of the reduced correction image area are determined, the reduced correction image is then created by determining the value of each pixel in the reduced correction image area ( Step S6). There are various conceivable orders for determining the pixel value in the reduced corrected image, but in the present embodiment, the pixel value is determined from the pixel at the upper left corner shown by the diagonal lines in FIG. To determine the upper left pixel, a mask as shown in the upper left is used. Specifically, a mask having the same shape as this mask is arranged at positions corresponding to all pixels other than the correction target region of the reduced original image, and the degree of similarity is obtained. The pixel at the position corresponding to the mask on the reduced original image determined to have the highest similarity is
The pixel is assigned as the upper left pixel shown in FIG. The mask on the reduced original image is first placed at the position shown in FIG. 4B, and the degree of similarity between the masked region and the masked region shown in FIG. 4A is obtained. Next, FIG.
As shown in (c), shift the mask by one pixel to the right,
The degree of similarity between the masked area and the masked area shown in FIG. 4A is calculated. Similarly, the process of obtaining the degree of similarity is performed for each pixel position. When the mask reaches the right end on the reduced original image, the mask is moved to the left end one line down, as shown in FIG. The degree of similarity between the masked area and the masked area shown in FIG. 4A is obtained. In this way, the similarity between the masked area and the masked area shown in FIG. 4A is obtained for the positions corresponding to all the pixels on the reduced original image. For example, when it is determined that the similarity to the masked area at the position shown in FIG. 4E is the highest, the pixels shaded with diagonal lines in FIG. It will be assigned as the upper left pixel shown in (a).

Here, not only the pixel value is assigned to the reduced corrected image area, but which pixel on the reduced original image is assigned is stored. This is because the reduced corrected image obtained by giving a pixel value to the reduced corrected image area is not the target image but must have the information necessary for restoring the original size. Since each pixel on the reduced original image records which pixel of the original image is obtained, it can be finally restored.

Here, in the present invention, a function for calculating the degree of similarity between the area masked with the reduced corrected image and the area masked with the reduced original image will be described. As this function, any function may be used as long as it can determine the similarity between the pixel existing at the mask position on the reduced corrected image area and the pixel existing at the mask position on the reduced original image. In the embodiment, the one shown in the following [Formula 2] is used.

[Formula 2] V = Σi (Rs ² + Gs ² + Bs ² ) where i = 1 to 12 (number of pixels)

In the above [Formula 2], V is an evaluation value for evaluating the degree of similarity, and Rs, Gs, and Bs are the pixels on the reduced corrected image area when the RGB color image is used, respectively. It is a difference for each color component of pixels on the reduced original image. Therefore, the smaller the value of the evaluation value V, the closer the value is, and it is determined that the degree of similarity is high. In the above example, i = 1 to 12, but this is
This is the case where a standard mask as shown in FIG. 4 is used, and it depends on the mask shape.

By the way, in order to obtain the mask similarity, a pixel must exist at the mask position, but as shown in FIG. 4A, the first pixel in the reduced corrected image area is determined. In that case, some pixel is required at the mask position. Here, based on the relationship between the actual correction target area and the surrounding pixels, the pixels in the shaded area in FIG. 2B are used. Therefore, the image obtained within the reduced corrected image size is affected by the surrounding pixels, and the image does not become completely different from the surroundings. That is, it is possible to prevent the correction target area from being in a state of being spliced to other parts of the original image.

When the pixel at the upper left corner as shown in FIG. 4A is determined, as shown in FIG. 5A, the mask on the reduced corrected image area is shifted to the right by one pixel and the same process is performed. To do. At this time, the pixels used to obtain the similarity of the masked area are 11 pixels outside the reduced corrected image area and the upper left pixel to which the pixel has already been assigned (see FIG. 5).
The value of (circle mark in (a)) is used.

Similarly, the pixels in the first row of the reduced corrected image size are determined. FIG. 5B shows the position of the mask when the rightmost pixel in the first row is determined. Figure 5
Also in (b), the values of the pixels indicated by the circles have already been determined. When the rightmost pixel in the first row is determined, the pixels in the second row are determined from the right edge instead of the left edge. At this time, as shown in FIG. 5C, the mask used has a shape that is left-right inverted from that used for determining the pixels in the first row. In addition, also in FIG.
The mark indicates a pixel whose value has already been determined. FIG. 5 (c)
When the mask as shown in FIG. 5 is used, the mask having the same shape as that shown in FIG. 5C is naturally used as the mask on the reduced original image for obtaining the similarity. Using such a mask, the pixels on the second row are sequentially determined from the right side.
Regarding the third line of the reduced correction image area, FIG.
It is determined from the pixel on the left end using the mask as shown in FIG. That is, the pixels are determined in the left-right opposite directions by using the mask that is horizontally inverted for each row. Although it is possible to determine all the rows from one direction at all times, in such a case, there may occur a phenomenon in which an unintended flow like a diagonal flow appears on the created image. . In order to eliminate such a problem, it is desirable to determine pixels in a zigzag manner as in the example of FIG.

For the first row to the third row from the bottom of the reduced corrected image area, the pixels are determined using a mask of 12 pixels as shown in FIGS. 4 and 5, but from the bottom. For two lines, masks having different shapes as shown in FIG. 6 are used. In FIG. 6, the shaded hatched pixels are the pixels to be determined, and the circled ones are the already determined pixels. To determine the pixels in the second row from the bottom,
A mask as shown in FIG. 6A is used. This mask has a shape in which, in addition to 12 pixels as shown in FIGS. 4 and 5, 5 pixels at a position two rows below the pixel to be determined are added. Also, to determine the pixel in the bottom row,
A mask as shown in FIG. 6B is used. This mask has a shape in which, in addition to the 12 pixels shown in FIGS. 4 and 5, 10 pixels in the lower two rows of the pixel to be determined are added. The determination of the degree of similarity can be performed by calculating the evaluation value as in [Equation 2] above. However,
In [Equation 2], the mask shown in FIG.
i = 1 to 17, and with the mask shown in FIG.
i = 1 to 22.

In determining the pixels for the lower two rows, the mask having the shape as shown in FIG. 6 is used to reflect the values of the pixels for the lower two rows on the reduced corrected image area. As described above, the two pixels on the upper, lower, left and right sides of the reduced corrected image area are duplicated from the corresponding positions of the reduced original image. Therefore, reflecting the values of these pixels means that the seams are not conspicuous when the corrected image is pasted in the correction target area later. That is, the corrected mark becomes inconspicuous, which greatly contributes to solving the problem of the present invention. Note that the upper and left and right sides of the reduced corrected image size sufficiently reflect the pixel values of the reduced original image, as can be seen from the examples shown in FIGS. 4 and 5.

When the reduced corrected image is obtained by determining each pixel in the reduced corrected image area, the corrected partial image whose resolution is returned to the original is created based on this (step S).
7). Since the reduced corrected image is created as described above, each pixel records which pixel of the reduced original image is assigned. Also, each pixel on the reduced original image is
It records which r × r pixels of the original image were obtained. That is, it is possible to know which r × r pixels on the original image each pixel on the reduced image corresponds to, so that each pixel on the reduced image corresponds to r × r pixels on the original image. Is assigned, a corrected partial image having the same resolution as the original image is created.

In the process of step S7, when the corrected partial image is created from the reduced corrected image, the corresponding pixel on the original image is obtained from the reduced corrected image through the reduced original image, and the pixel is directly displayed on the corrected partial image. I assigned it to
At this time, when the resolution reduction rate is small (r = 3 or so), there is not much problem, but when the resolution reduction rate is large, it is like a lump for each number of pixels corresponding to the resolution reduction rate. Will occur. For example, if the resolution reduction rate r = 40, then 40 × 40
The blocks consisting of pixels are arranged on the modified partial image. At this time, in the portion corresponding to the joint of 40 × 40 pixels, the continuity is not taken into consideration in the arrangement of the adjacent pixels, so that the joint becomes conspicuous. As a result, the entire modified partial image looks like an aggregate of blocks of 40 × 40 pixels.

Therefore, in the process of step S7, the above problem is solved by performing the following process. When the resolution reduction ratio r = 40, 40 × 40 pixels are allocated on the modified partial image, but 46 × 46 pixels on the original image are allocated on the modified partial image. In other words, the upper, lower, left, and right 3 pixels are extracted and allocated in excess. The extra extracted pixels are combined with other pixels assigned on the modified partial image. The synthesis rate at this time is set to a value that becomes smaller as it gets closer to the end of the block of 46 × 46 pixels.
Vary the composition rate using random numbers. The synthesis rate α is calculated by the following [Formula 3].

[Formula 3] α = DI (X) + RAND

In the above [Formula 3], DI (X) is
Is a function that increases with the distance from the edge of the block,
It takes a value from 0 to 1. Further, RAND is -0.5 to 0.
It is a random number that takes a value of 5. After the calculation, α is corrected to have a value of 0 to 1. In addition, here, 3
The extra pixels are extracted and assigned, but the number of extra pixels is also changed according to the reduction rate.

When the corrected partial image is obtained, the corrected partial image is pasted on the image of the correction target area portion of the original image (step S8). As a result, a new original image is obtained. That is, the correction target area of the original image has been corrected. As described above, the image in the designated area is corrected, but a more effective method for creating an image will be described below.

In the above example, in order to reduce the amount of calculation, the modified partial image is created after the reduced modified image having a reduced resolution is created, but in this case, the problem as shown in FIG. Occurs. FIG. 7A is an enlarged view of a part (three pixels) of the reduced corrected image, and FIG. 7B is a partially enlarged view of the corrected partial image. Note that FIG. 7 shows the resolution reduction rate r
= 3 is shown. When r = 3, one pixel on the reduced corrected image is equal to 9 (3 ×
3) It will be restored to pixels, but in this case, this 9
The pixel array is restored in the same state as the original image. Therefore, as shown in FIG. 7A, when the same pixel on the reduced corrected image is selected three times in succession, when restored, the pattern repeats in a narrow range on the corrected partial image. This is because a specific pixel is repeated, as can be seen from the pixel shaded in FIG.

Therefore, in the present invention, the above-mentioned problem does not occur by not selecting the same pixel as the neighboring pixels. FIG. 8 is a partially enlarged view of the reduced corrected image. Particularly, FIG. 8A shows a case where pixels are determined from left to right, and FIG. 8B shows a case where pixels are determined from right to left. When determining pixels from the left to the right, when determining the pixels indicated by hatching in FIG. 8A, the pixels should not be the same as the four pixels indicated by the circles. In determining the pixels shaded with diagonal lines, the similarity between the mask on the reduced corrected image and the mask on the reduced original image is obtained in the same manner as described in the above example, and the similarity is the highest. The pixel in the case of being higher is selected, but here, the pixel on the selected reduced original image is not made to be the pixel indicated by the diagonal line as it is,
It is checked whether it is not the same as the four pixels indicated by the mark. As a result, the same pixel as a neighboring pixel is not selected, and the above problem is solved. Further, when the pixels are determined from the right to the left, the positional relationship as shown in FIG. 8B, that is, the processing for preventing the overlapping with respect to the pixels symmetrical to the case of FIG. Done.

Next, a system configuration for executing the above image correction method will be described. FIG. 9 is a block diagram showing an embodiment of the image correction system according to the present invention. In FIG. 9, the original image input means 1 has a function of inputting an image having a portion to be corrected. As the original image input means 1, when the original image is already prepared as digital data, it can be realized by a reading device capable of reading an electronic recording medium on which the digital data is recorded, and the original image is a physical image such as paper. When prepared as a medium, it can be realized by an optical reading device such as a scanner.

The correction area designating means 2 executes the step S in FIG.
It is for executing the process 1 and can be realized by an input device such as a mouse or a keyboard. The parameter setting unit 3 sets various necessary parameters, and particularly has a function of setting the resolution reduction rate r in step S2 of FIG. The parameter setting means 3 is also realized by an input device such as a mouse or a keyboard, like the correction area designating means 2.

The mask shape storage means 4 stores the masks to be arranged on both the reduced corrected image and the reduced original image for determining the pixels of the corrected partial image as described in step S6 of FIG. It is for.

The arithmetic and control unit 5 plays a central role in the image correction system according to the present invention, and has a pixel determining unit 5a and a resolution converting unit 5b. The pixel determining means 5a has a function of executing the processing of steps S4 to S6 of FIG. 1, and secures an area for creating a reduced corrected image on the image memory 6 to determine pixels. The resolution converting means 5b has a function of executing the processing of steps S3 and S7 of FIG. 1, and creates a reduced original image from the original image according to the set resolution reduction rate and corrects it from the reduced corrected image. Create a partial image.
The arithmetic and control unit 5 is realized by a CPU and a memory of a computer, and the pixel determining unit 5a and the resolution converting unit 5b which are specific functions are realized by a dedicated software program installed in the computer.

The image memory 6 is a storage area for creating a reduced correction image, and is a size obtained by reducing the correction target area designated by the correction area designating means 2 by the resolution reduction rate set by the parameter setting means 3. The storage area is secured and the pixels calculated by the arithmetic and control unit 5 are stored.

The image display means 7 is for displaying an image handled in the present invention, and is particularly used for displaying an original image when designating a correction target area. Actually, it is realized by a display device such as a liquid crystal / CRT connected to a computer.

The preferred embodiment of the present invention has been described above, but the present invention is not limited to the above embodiment, and various modifications can be made. For example, in the above-described embodiment, in order to reduce the load of the arithmetic processing, the original image is reduced, the reduced corrected image of the reduced size is created, and then the resolution is restored to create the modified partial image. If the calculation processing load is not taken into consideration, a correction target area and two pixels in the upper, lower, left, and right directions are secured in the image memory, and pixels other than the correction target area of the original image are used to mask the same as described above. The corrected partial image may be directly created by obtaining the similarity between the regions.

[0039]

As described above, according to the present invention,
An original image having a portion to be corrected is input, a correction target area is specified on the input original image, and each pixel of the specified correction target area is determined. , A mask having a predetermined shape is placed on the correction target area, a mask having the same shape as the mask is placed on the original image, and the mask on the original image is set to 1 while the position of the mask on the correction target area is fixed. While moving pixel by pixel, the similarity between the masked area on the correction target area and the masked area on the original image is calculated, and the positional relationship with the mask on the original image that is determined to have the highest similarity. Is to allocate pixels having the same positional relationship as the positional relationship between the mask on the correction target area and the pixel to be determined, as the pixels on the correction target area. From the area The protruding portion is made by using the pixels on the original image, so that the correction target area is newly created with the characteristics of the original image, and the joint portion of the correction target area is also conspicuous. Has the effect of disappearing. In particular, when a pattern habit is generated in the pattern image, the characteristic of the original image itself is reproduced in the correction target area in which the pattern habit is generated, whereby a corrected image from which the pattern habit is removed can be obtained.

[Brief description of drawings]

FIG. 1 is a flowchart of an image correction method according to the present invention.

FIG. 2 is a diagram showing a relationship between an original image and a correction target area.

FIG. 3 is a diagram showing a work area secured on an image memory.

FIG. 4 is a diagram showing a relationship between a reduced corrected image, a reduced original image, and a mask arranged thereon.

FIG. 5 is a diagram showing a relationship between a reduced corrected image and a mask when the order of pixel determination is changed left and right alternately.

FIG. 6 is a diagram showing a relationship between a reduced corrected image and a mask when performing a seamless process on a lower side portion of a correction target area.

FIG. 7 is a diagram showing a pixel relationship when the same pixel is continuously selected on the reduced corrected image.

FIG. 8 is a diagram showing a positional relationship between a pixel to be determined and a pixel whose duplication is prohibited.

FIG. 9 is a configuration diagram showing an embodiment of an image correction system according to the present invention.

[Explanation of symbols]

1. Original image input means 2 ... Correction area designation means 3 ... Parameter setting means 4 ... Mask shape storage means 5 ... Arithmetic control device 5a ... Pixel determining means 5b ... Resolution conversion means 6 ... Image memory 7 ... Image display means

   ─────────────────────────────────────────────────── ─── Continued front page    F term (reference) 5B050 BA06 BA18 CA07 EA03 EA12                       EA18                 5C076 AA01 AA22 BA06 BB07                 5L096 BA18 CA24 EA03 EA14 FA69                       GA10 JA03

Claims (5)

[Claims] 1. A method comprising: specifying a correction target area on an original image; and determining each pixel of the specified correction target area, wherein each pixel of the correction target area is determined by A mask having a predetermined shape is arranged on the correction target area, a mask having the same shape as the mask is arranged on a portion other than the correction target area on the original image, and the position of the mask on the correction target area is fixed. As it is, the degree of similarity between the masked area on the correction target area and the masked area on the original image is calculated by moving the mask on the original image by one pixel at a time, and the similarity is the highest. A pixel having a positional relationship with the mask on the original image to be determined has a positional relationship that is the same as the positional relationship between the mask on the correction target region and the pixel to be determined,
The image correction method is to be assigned as pixels on the correction target area, and a portion of the mask arranged on the correction target area that extends outside the correction target area is performed by using pixels on the original image. . 2. When determining each pixel of the correction target area,
For the last predetermined number of rows of the correction target area, a mask having a shape different from the mask shape is used, and as a part of the pixels, pixels of a predetermined number of rows below the correction target area on the original image are used. The image correction method according to claim 1, wherein the image correction method is performed. 3. A step of designating an area to be corrected on the original image, a step of setting a resolution reduction rate, a step of creating a reduced original image having a reduced resolution in accordance with the set resolution reduction rate, A step of obtaining a reduced correction image by determining each pixel of the reduced correction image in which the resolution of the corrected target area is reduced according to the set resolution reduction rate; and information recorded in each pixel of the reduced correction image. And a step of creating a modified partial image that is an image of the modification target area portion of the original image, and determining each pixel in the step of obtaining the reduced modified image,
A mask having a predetermined shape is arranged on the reduced corrected image, a mask having the same shape as the mask is arranged on the reduced original image, and the position of the mask on the reduced corrected image is fixed, and the mask is placed on the reduced original image. While moving the mask of 1 pixel by 1 pixel,
The positional relationship between the masked area on the reduced corrected image and the masked area on the reduced original image is calculated, and the positional relationship between the mask on the reduced original image determined to have the highest similarity is Pixels that have the same positional relationship as the mask and the pixel to be determined on the reduced corrected image are
The portion to be allocated as a pixel on the reduced corrected image, and the portion of the mask arranged on the reduced corrected image that is outside the size of the reduced corrected image is performed by using the pixels on the reduced original image. The step of creating an image is based on the pixels on the reduced original image assigned to each pixel of the reduced corrected image,
The correction partial image is created by specifying a plurality of pixels on the corresponding original image and arranging the plurality of pixels on the original image corresponding to each pixel of the reduced correction image. And image correction method. 4. Image input means for inputting an original image to be corrected, correction area designating means for designating a correction target area on the input original image, and determination of each pixel on the correction target area, A mask having a predetermined shape is arranged on the correction target area, a mask having the same shape as the mask is arranged on the original image, and the mask on the original image is fixed while the position of the mask on the correction target area is fixed. While moving each pixel by 1 pixel, the similarity between the masked area on the correction target area and the masked area on the original image is obtained, and the mask on the original image is determined to have the highest similarity. Pixels having the same positional relationship as that of the mask and the pixel to be determined on the correction target area are assigned as pixels on the correction target area, Image modification system mask arranged in frequency is characterized by having when protruding from the correction object region, and the pixel determination unit is to utilize the corresponding pixel on the original image. 5. A method of arranging a mask of a predetermined shape on a correction target area, a step of arranging a mask of the same shape as the mask on the original image, and a position of the mask on the correction target area in a computer. While the mask is fixed, the mask on the original image is moved pixel by pixel, and the similarity between the masked area on the correction target area and the masked area on the original image is calculated. A pixel having a positional relationship with the mask on the original image that is determined to be high has the same positional relationship as the positional relationship between the mask on the correction target area and the pixel to be determined, as a pixel on the correction area. The step of allocating is performed to determine each pixel on the correction target area, and for the portion of the mask placed on the correction target area that extends beyond the correction target area, use the corresponding pixel on the original image. Program is intended to do more.