JP5251316B2 - Image correction apparatus, image correction method, and image correction program - Google Patents

Description

  The present invention relates to an image correction apparatus, an image correction method, and an image correction program.

  When automatically recognizing characters, barcodes, or the like printed on a form, for example, the form is photographed by a non-contact scanner installed so as to face the form, and image data is acquired. Then, characters, bar codes, and the like are recognized by processing the image data using a computer or the like.

  The non-contact scanner uses a charge coupled device (CCD) image sensor, a complementary metal oxide semiconductor (CMOS) image sensor, or the like as an image sensor. The number of images of the image sensor of the non-contact scanner is about 7 million to 10 million pixels. When A4 size paper is photographed using a non-contact scanner, the resolution of the photographed image is approximately 250 dpi to 350 dpi.

On the other hand, in order to reduce the printing space, a barcode with a narrow line width (element width) may be printed on a form.
In such a case, the image sensor of the non-contact scanner has insufficient resolution, and the captured image data may not be able to accurately reproduce the individual line widths of the barcode. If the individual line width of the barcode cannot be reproduced, the barcode cannot be recognized correctly.
Therefore, it is conceivable to combine a plurality of images taken while shifting the pixels to generate an image with a resolution higher than the actual resolution of the image sensor. Since each line width of the barcode is accurately reproduced in a high-resolution image, the barcode can be correctly recognized.

  As a method of generating a high-resolution image from a low-resolution image, a reference image and an image displaced obliquely with respect to the reference image (distance between pixels adjacent to each other in the horizontal and vertical directions) And an image obtained by displacing the position of the pixel by a half of the distance (see, for example, Patent Document 1).

  In this method, two images are first overlapped to generate an image in which pixels (hereinafter referred to as real pixels) are arranged in a checkered pattern. Next, a pixel in an empty region (hereinafter referred to as an imaginary pixel) in which there is no pixel between each real pixel is generated from the real pixels around the region.

  For example, the imaginary pixel can be generated by weighting and superimposing the real pixels adjacent to the top, bottom, left and right of the imaginary pixel. However, at the edge (where the luminance difference between two adjacent pixels is large), an imaginary pixel is generated using a pixel with a high luminance and a pixel with a low luminance, so blurring or noise may occur.

  Therefore, in the image correction method disclosed in Patent Document 1, first, the amount of change in the horizontal direction is detected from the difference between the pixel values of the left and right real pixels adjacent to the imaginary pixel, and the difference between the pixel values of the upper and lower real pixels is detected. The amount of change in the vertical direction is detected. When the horizontal change amount of the image is smaller than the vertical change amount, the average value of the two right and left real pixels adjacent to the imaginary pixel is set as the pixel value of the imaginary pixel. When the amount of change in the vertical direction of the image is smaller than the amount of change in the horizontal direction, the average value of the upper and lower real pixels adjacent to the imaginary pixel is used as the pixel value of the imaginary pixel. When the amount of change in both the horizontal direction and the vertical direction is the same, the average value of the four real pixels on the top, bottom, left, and right is set as the pixel value of the imaginary pixel.

Japanese Patent Laid-Open No. 10-108209

  However, the image correction method disclosed in Patent Document 1 may not be able to detect edges correctly. For example, when the line width (element width) of the photographed barcode is equal to the interval between the actual pixels, the difference between the pixel values of the left and right actual pixels and the difference between the pixel values of the upper and lower actual pixels are both zero. The amount of change in the vertical direction is the same. For this reason, the average value of four real pixels on the top, bottom, left, and right is the pixel value of the imaginary pixel. In other words, an edge cannot be detected, and an imaginary pixel is generated assuming that there is no edge.

  In view of the above situation, the present invention correctly detects an edge of an image in which a plurality of pixels are arranged in a checkered pattern, and determines a pixel in an empty region in which no pixel exists between pixels according to the detected edge. An object of the present invention is to provide an image correction apparatus, an image correction method, and an image correction program that can be generated appropriately.

The image correction apparatus of the present invention includes:
First photographing means for photographing a first image;
Taking a second image in which the pixel position is displaced by a distance half the distance between adjacent pixels in the direction of the pixel column and in a direction different from the pixel column with respect to the pixel position of the first image Second imaging means for
Based on the first image and the second image, each pixel column is composed of a plurality of pixels and empty areas where no pixels exist alternately at predetermined intervals. A checkered image that generates an image of a checkered arrangement in which pixels and empty regions included in odd-numbered pixel columns and pixels and empty regions included in even-numbered pixel columns are shifted in the direction of the pixel columns. Array image generation means ;
From the checkered array image, two adjacent pixels included in each of the odd-numbered pixel column and the even-numbered pixel column and two empty regions adjacent to both of the two adjacent pixels; Pixel block cutout means for cutting out a pixel block including a predetermined number of pixels and two empty areas around two adjacent pixels and two empty areas adjacent to both of the two adjacent pixels;
Edge information acquisition means for detecting an edge composed of two adjacent pixels having a difference in pixel value and obtaining edge information indicating a type of edge included in the pixel block;
For each piece of edge information included in a group of edge information indicating types of edges that may be included in the pixel block, both the correction method for correcting the two adjacent pixels and the two adjacent pixels Correction method storage means for storing a generation method for generating pixels corresponding to two adjacent empty regions;
The correction method and the generation method corresponding to the edge information obtained by the edge information acquisition unit are read from the correction method storage unit, and the two adjacent pixels are corrected according to the read correction method. Pixel correcting means for generating pixels corresponding to two empty regions adjacent to both of the two adjacent pixels according to the generated generation method;
It is characterized by providing.

Further, the image correction method of the present invention includes:
A first photographing step of photographing a first image;
Taking a second image in which the pixel position is displaced by a distance half the distance between adjacent pixels in the direction of the pixel column and in a direction different from the pixel column with respect to the pixel position of the first image A second imaging step to
Based on the first image and the second image, each pixel column is composed of a plurality of pixels and empty areas where no pixels exist alternately at predetermined intervals. A checkered image that generates an image of a checkered arrangement in which pixels and empty regions included in odd-numbered pixel columns and pixels and empty regions included in even-numbered pixel columns are shifted in the direction of the pixel columns. An array image generation process ;
From the checkered array image, two adjacent pixels included in each of the odd-numbered pixel column and the even-numbered pixel column and two empty regions adjacent to both of the two adjacent pixels; A pixel block cutting step of cutting out a pixel block including a predetermined number of pixels and an empty region around two adjacent pixels and two empty regions adjacent to both of the two adjacent pixels;
An edge information acquisition step of detecting an edge composed of two adjacent pixels having a difference in pixel value and obtaining edge information indicating a type of edge included in the pixel block;
For each piece of edge information included in a group of edge information indicating types of edges that may be included in the pixel block, both the correction method for correcting the two adjacent pixels and the two adjacent pixels The correction method and the generation corresponding to the edge information obtained in the edge information acquisition step from a correction method storage means storing a generation method for generating pixels corresponding to two adjacent empty regions Read out the method, correct the two adjacent pixels according to the read correction method, and correspond to two empty regions adjacent to both of the two adjacent pixels according to the read generation method A pixel correction step for generating pixels;
It is characterized by providing.

The image correction program of the present invention is stored in a computer.
A first shooting procedure for shooting a first image;
Taking a second image in which the pixel position is displaced by a distance half the distance between adjacent pixels in the direction of the pixel column and in a direction different from the pixel column with respect to the pixel position of the first image A second shooting procedure,
Based on the first image and the second image, each pixel column is composed of a plurality of pixels and empty areas where no pixels exist alternately at predetermined intervals. A checkered image that generates an image of a checkered arrangement in which pixels and empty regions included in odd-numbered pixel columns and pixels and empty regions included in even-numbered pixel columns are shifted in the direction of the pixel columns. A sequence image generation procedure ;
From the checkered array image, two adjacent pixels included in each of the odd-numbered pixel column and the even-numbered pixel column and two empty regions adjacent to both of the two adjacent pixels; A pixel block cutout procedure for cutting out a pixel block including a predetermined number of pixels and an empty region around two adjacent pixels and two empty regions adjacent to both of the two adjacent pixels;
An edge information acquisition procedure for detecting an edge composed of two adjacent pixels having a difference in pixel value and obtaining edge information indicating a type of edge included in the pixel block;
For each piece of edge information included in a group of edge information indicating types of edges that may be included in the pixel block, both the correction method for correcting the two adjacent pixels and the two adjacent pixels The correction method and the generation corresponding to the edge information obtained in the edge information acquisition procedure from the correction method storage means storing a generation method for generating pixels corresponding to two adjacent empty regions Read out the method, correct the two adjacent pixels according to the read correction method, and correspond to two empty regions adjacent to both of the two adjacent pixels according to the read generation method A pixel correction procedure for generating pixels;
Is executed.

  According to the present invention, for an image in which a plurality of pixels are arranged in a checkered pattern, an edge is correctly detected, and a pixel in an empty area in which there is no pixel between the pixels is appropriately generated according to the detected edge. can do.

  As shown in FIG. 1, the image correction apparatus 1 according to the embodiment of the present invention includes an image photographing unit 101, an image storage memory 121, an image superimposing unit 111, a pixel block cutout unit 112, and edge information acquisition. Unit 113, pixel correction unit 114, and optimum correction dictionary storage memory 122.

  The image capturing unit 101 uses a reference image (hereinafter referred to as a non-shifted image) and an image displaced in an oblique direction with respect to the reference image (distances between pixels adjacent to each other in the horizontal and vertical directions). An image in which the position of the pixel is displaced by a half distance; hereinafter referred to as a shifted image) is taken. The image capturing unit 101 is, for example, a non-contact scanner, and captures a form or the like.

  The image storage memory 121 stores two images (an image without shifting and an image with shifting) captured by the image capturing unit 101.

The image superimposing unit 111 reads out the non-shifted image and the shifted image from the image storage memory 121, and generates an image in which a plurality of real pixels are arranged in a checkered pattern (hereinafter referred to as a checkered array image).
Here, the actual pixel is, for example, a pixel that represents a luminance signal or a color difference signal. Alternatively, the pixel may represent any of an R (red) signal, a G (green) signal, and a B (blue) signal.

  The pixel block cutout unit 112 includes a predetermined number of real pixels and imaginary pixels from a checkered array image centering on an area including two adjacent real pixels and two imaginary pixels (hereinafter referred to as a correction target area). Cut out a pixel block. As will be described later, the pixel block cutout unit 112 sequentially cuts out pixel blocks around the correction target region from the upper left corner to the lower right corner of the checkered array image.

  As will be described later, the edge information acquisition unit 113 obtains edge information indicating the type of edge included in the pixel block extracted by the pixel block extraction unit 112.

The optimum correction dictionary storage memory 122 stores the optimum correction dictionary. As will be described later, the optimum correction dictionary holds a method for optimally correcting two real pixels included in the correction target region and a method for optimally generating two imaginary pixels for all edge information.
The pixel correction unit 114 determines an optimum correction method and generation method corresponding to the edge information generated by the edge information acquisition unit 113 with reference to the optimum correction dictionary stored in the optimum correction dictionary storage memory 122, and performs correction. Two real pixels in the target area are corrected to generate two imaginary pixels.

  The pixel block cutout unit 112 is an example of the pixel block cutout unit of the present invention, the edge information acquisition unit 113 is an example of the edge information acquisition unit of the present invention, and the optimum correction dictionary storage memory 122 is the correction method of the present invention. The pixel correction unit 114 is an example of a storage unit, the pixel correction unit 114 is an example of a pixel correction unit of the present invention, and the image shooting unit 101 is an example of a first shooting unit and a second shooting unit of the present invention. The matching unit 111 is an example of a checkered array image generation unit of the present invention.

  As shown in FIG. 2, the image correction apparatus 2 according to the embodiment of the present invention can be configured by an image photographing unit 101, a CPU 11, a memory 12, and a hard disk 13. The image capturing unit 101 is the same in FIGS.

A CPU (Central Processing Unit) 11 is a programmable processor.
The memory 12 includes memories such as a RAM (Random Access Memory), a ROM (Read Only Memory), and a flash memory. The memory 12 stores a program of the CPU 11 and various settings. The memory 12 is used as a work area.
The hard disk 13 stores a non-shifted image, a shifted image, a checkered array image, and an optimal correction dictionary.

  The image correction apparatus 2 implements an image superimposing unit 111, a pixel block cutout unit 112, an edge information acquisition unit 113, and a pixel correction unit 114 by executing a program with the CPU 12.

  Next, the image correction method according to the embodiment of the present invention will be described with reference to the image correction processing of FIG.

The image photographing unit 101 photographs a form using, for example, a non-contact scanner installed so as to face the form, and obtains a non-shifted image 231 and a shifted image 232 shown in FIG. S101).
The image photographing unit 101 stores the photographed image 231 without displacement and the image 232 with displacement in the image storage memory 121 (step S102).
The actual pixels of the non-shifted image 231 and the shifted image 232 are, for example, a 256-level multi-valued image, and the value of the actual pixel in the white background portion (the portion with high luminance) of the form is large.

  As shown in FIG. 4B, the image superimposing unit 111 reads a non-shifted image 231 and a shifted image 232 from the image storage memory 121, and superimposes these images to generate a checkered array image 233 ( Step S103). The inside of the correction target frame 234 is a correction target.

  As illustrated in FIG. 5A, the pixel block cutout unit 112 first cuts out the first pixel block 401 from the checkered array image 233 (step S104). The pixel block 401 is located in the upper left corner of the checkered array image 233, and is composed of 18 real pixels. The pixel block 401 includes a correction target area 402 including two real pixels and two imaginary pixels. The pixel block cutout unit 112 outputs the cut out pixel block 401 to the edge information acquisition unit 113.

Next, as illustrated in FIG. 5B, the pixel block cutout unit 112 cuts out the second pixel block 403 from the checkered array image 233 (step S104). The pixel block 403 includes a correction target area 404 composed of two real pixels and two imaginary pixels. The pixel block cutout unit 112 outputs the cut out pixel block 403 to the edge information acquisition unit 113.
Thereafter, the pixel block cutout unit 112 cuts out the pixel blocks sequentially from the left to the right of the checkered array image 233 and from the top to the bottom until the last pixel block at the lower right corner of the checkered array image 233, and the extracted pixel blocks are edged. The information is output to the information acquisition unit 113.

As illustrated in FIG. 6, the edge information acquisition unit 113 includes nine actual pixels A0 to A8 of the shifted image and nine actual pixels B0 to B8 of the shifted image in the pixel block 60. The edge information acquisition unit 113 obtains edge information based on the actual pixels A0 to A8 and the actual pixels B0 to B8 (step S105).
Then, the pixel correction unit 114 corrects the upper right real pixel B4 (612) and the lower left real pixel A4 (613) included in the correction target region 61 according to the edge information obtained by the edge information acquisition unit 113, and An upper left imaginary pixel 611 and a lower right imaginary pixel 614 are generated (step S106).

  Step S104 is an example of a pixel block cutting means, a pixel block cutting process, and a pixel block cutting procedure of the present invention. Step S105 is an example of an edge information acquisition means, an edge information acquisition process, and an edge information acquisition procedure of the present invention. Step S106 is an example of the pixel correction means, the pixel correction step, and the pixel correction procedure of the present invention.

  Next, the edge information acquisition unit 113 and the pixel correction unit 114 will be described in detail.

  The edge information acquisition unit 113 obtains edge information based on the following conditional expressions 1 to 9.

  Here, T1 and T2 are predetermined threshold values. | X | means the absolute value of X. The edge information acquisition unit 113 sets “1” when the conditional expressions 1 to 9 are satisfied, and “0” when the conditional expressions 1 to 9 are not satisfied. The edge information acquisition unit 113 obtains 9-bit data with the results determined by the conditional expressions 1 to 9 as the 9th to 1st bits, respectively. That is, the edge information acquisition unit 113 obtains the 9-bit data as edge information. There are 512 types of edge information.

Conditional expression 1 is satisfied when the difference between the values of the two actual pixels A4 (613) and B4 (612) included in the correction target region 61 is smaller than the threshold value T1. That is, it is satisfied when the real pixels A4 (613) and B4 (612) do not constitute an edge. In this case, the edge information acquisition unit 113 sets the ninth bit of the edge information to “1”. Conditional expression 1 is not satisfied when the difference between the values of the two real pixels A4 (613) and B4 (612) is equal to or greater than the threshold value T1. That is, when the real pixels A4 (613) and B4 (612) form an edge, the conditional expression 1 is not satisfied. In this case, the edge information acquisition unit 113 sets the ninth bit of the edge information to “0”.
Similarly, the edge information acquisition unit 113 obtains the 8th to 5th bits of the edge information according to the presence / absence of the edge in the vertical direction using the conditional expressions 2 to 5. Further, the edge information acquisition unit 113 obtains the fourth bit to the first bit of the edge information according to the presence / absence of the edge in the horizontal direction according to the conditional expressions 6 to 9.

  For example, if there is no edge, the edge information is 111111111 (511), and if there is an edge, the edge information is one of 0 to 510.

  The edge information acquisition unit 113 outputs the pixel block 60 and the obtained edge information to the pixel correction unit 114.

  The optimum correction dictionary storage memory 122 stores an optimum correction dictionary. As shown in FIG. 7, the optimum correction dictionary corresponds to the edge information and relates to the correction method of the real pixels A4 (613) and B4 (612) and the generation method of the imaginary pixels 611 and 614 in the correction target area 61. Holds information.

  The following formulas 1-1 to 1-4 show a method for generating the upper left imaginary pixel 611 in the correction target region 61. Formula 1-1 is selected when there is no edge around the imaginary pixel 611. Formulas 1-2 to 1-4 are selected according to the state of the edge when there is an edge around the imaginary pixel 611.

The following formulas 2-1 and 2-2 show how to correct the upper right real pixel B4 (612) in the correction target area 61. Formula 2-1 is selected when there is no edge around the actual pixel B4 (612). In Formula 2-1, smoothing processing is performed on the actual pixel B4 (612). Formula 2-2 is selected when there is an edge around the actual pixel B4 (612). Equation 2-2 outputs the actual pixel B4 (612) as it is.

  Equations 3-1 and 3-2 below show a method of correcting the lower left real pixel A4 (613) in the correction target region 61.

  The following expressions 4-1 to 2-4 indicate a method for generating the lower right imaginary pixel 614 in the correction target region 61.

  The pixel correction unit 114 refers to the optimal correction dictionary stored in the optimal correction dictionary storage memory 122, corrects the real pixel in the correction target area included in the pixel block, and generates a pixel corresponding to the imaginary pixel. The pixel correction unit 114 generates a corrected image (high-resolution image) in which noise near the edge where the luminance changes abruptly is reduced, and noise in a portion without the edge (background portion of the form) is also reduced.

  In the image correction apparatus 2, the CPU 11 sequentially executes the processes of the pixel block cutout unit 112, the edge information acquisition unit 113, and the pixel correction unit 114. Each time the processing of the pixel correction unit 114 ends, the CPU 11 determines whether the ended pixel block is the last pixel block at the lower right corner of the checkered array image 233. If it is not the last pixel block, the CPU 11 returns to step S104 and executes the process of the pixel block cutout unit 112. If it is the last pixel block, the CPU 11 ends the image correction process.

Next, a specific example will be described.
FIG. 8A and FIG. 8B are examples of images with edges in the vertical direction captured by the image capturing unit 101. FIG. 8A shows an image without shifting, and FIG. 8B shows an image with shifting. The image superimposing unit 111 superimposes these images to generate a checkered array image shown in FIG.

When the threshold values T1 = 16 and T2 = 32, the edge information acquisition unit 113 performs two kinds of edges of 011110000 (240) and 011110010 (242) according to Conditional Expression 1 to Conditional Expression 9 for the checkered array image of FIG. Ask for information.
The pixel correction unit 114 refers to, for example, the optimum correction dictionary in FIG. 7 for the edge information 011110000 (240), and sets the upper left imaginary pixel 611 to 1-2, the upper right real pixel 612 to 2-2, and the lower left actual pixel 613 to 3 -2 and the lower right imaginary pixel 614 are corrected or generated by the method of 4-2, respectively. The pixel correction unit 114 outputs the image shown in FIG. 10 as an image having an edge in the corrected vertical direction.

  FIG. 11A and FIG. 11B are examples of images with edges in the horizontal direction captured by the image capturing unit 101. FIG. 11A shows an image without shifting, and FIG. 11B shows an image with shifting. The image superimposing unit 111 superimposes these images to generate a checkered array image shown in FIG.

When the thresholds T1 = 16 and T2 = 32, the edge information acquisition unit 113 performs two kinds of edges of 000011111 (31) and 0000011111 (15) according to the conditional expressions 1 to 9 for the checkered array image of FIG. Ask for information.
For example, with respect to the edge information 000011111 (31), the pixel correction unit 114 refers to the optimal correction dictionary in FIG. 7, and sets the upper left imaginary pixel 611 to the 1-3 formula, the upper right real pixel 612 to the 2-2 formula, and the lower left real pixel 613 to the The formula 3-2 and the lower right imaginary pixel 614 are corrected or generated by the method of formula 4-3, respectively. The pixel correction unit 114 outputs the image shown in FIG. 13 as the corrected image.

  For example, when there is no edge as in the background portion of the form, the edge information is 111111111 (511). The pixel correction unit 114 refers to the optimum correction dictionary in FIG. 7, and the upper left imaginary pixel 611 is set to 1-1, the upper right real pixel 612 is set to 2-1, the lower left real pixel 613 is set to 3-1, and the lower right imaginary pixel. Each of 614 is corrected or generated by the method of Formula 4-1. These are corrections based on the average value of surrounding pixels, and the pixel correction unit 114 performs a smoothing process to remove paper surface noise and the like.

FIG. 14 shows an image correction apparatus 3 which is a modification of the embodiment of the present invention. 1 and 14 are denoted by the same reference numerals. The image correction apparatus 3 in FIG. 14 is different from the image correction apparatus 1 in FIG.
When the pixel position is displaced by half the distance between adjacent pixels in the horizontal direction and the vertical direction to increase the resolution, the vertical and horizontal resolution is fixed to be equivalent to twice the original image. The resolution conversion unit 115 changes the resolution of the output image in response to a user request.
The resolution conversion unit 115 inputs the image (corrected image) corrected by the pixel correction unit 114, calculates the magnification from the resolution of the corrected image and the requested resolution, and corrects it using a linear interpolation method or a cubic interpolation method. Reduce the image. Then, the resolution conversion unit 115 outputs an image having a resolution between the resolution of the original image and the resolution of the corrected image (equivalent to twice).

According to the present invention, for an image in which a plurality of real pixels are arranged in a checkered pattern (an image in a checkered pattern), a high-quality and high-resolution image can be obtained without detecting the edge correctly and generating noise near the edge. Can be generated. In particular, Conditional Expression 1 determines whether or not there is an edge composed of two real pixels included in the correction target area. For this reason, for example, even when the line width (element width) of the photographed barcode is equal to the interval between the actual pixels, the edge can be detected, and smoothing processing is performed despite the presence of the edge. Can be prevented.
In addition, since smoothing processing is performed on a portion where there is no edge (solid-filled portion) for an actual pixel, noise can be reduced even in a portion where there is no edge.

  Furthermore, according to the present invention, it is possible to appropriately change the information about the correction method of the real pixel and the generation method of the imaginary pixel in the correction target area defined by the optimal correction dictionary. By selecting an optimal correction method and generation method as needed, the corrected image can be improved in image quality.

  In the above embodiment, the edge information acquisition unit 113 obtains edge information based on the luminance signal. However, the edge information acquisition unit 113 performs color difference signals, R (red) signals, G (green) signals, and B (blue). ) Edge information may be obtained based on the signal.

  Although the embodiments of the present invention have been described above, various modifications and combinations necessary for design reasons and other factors are described in the inventions and embodiments of the invention described in the claims. It should be understood that the invention falls within the scope of the invention corresponding to the specific example.

It is a figure which shows an example of the image correction apparatus which concerns on embodiment of this invention. It is a figure which shows another example of the image correction apparatus which concerns on embodiment of this invention. It is a figure which shows an example of the flowchart of an image correction process. It is a figure which shows an example of the image of the checkered pattern produced | generated from the image without a shift, and the image with a shift. It is a figure which shows an example of the pixel block cut out from the image of a checkered arrangement | sequence. It is a figure which shows an example of the pixel block cut out from the image of a checkered arrangement | sequence. It is a figure which shows an example of the correction target area | region contained in the pixel block from which the edge information is calculated | required, and the pixel block. It is a figure which shows an example of the optimal correction dictionary. It is a figure which shows an example of the image without a shift | offset | difference which has an edge in a perpendicular direction, and a shift-present image. It is a figure which shows an example of the image of a checkered arrangement | sequence with an edge in a perpendicular direction. It is a figure which shows an example of the corrected image which has an edge in a perpendicular direction. It is a figure which shows an example of a non-shifting image with an edge in a horizontal direction, and an image with a shift. It is a figure which shows an example of the image of a checkered arrangement | sequence with an edge in a horizontal direction. It is a figure which shows an example of the corrected image which has an edge in a horizontal direction. It is a figure which shows an example of the image correction apparatus which is a modification of embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 2, 3 ... Image correction apparatus 101 ... Image photographing part 111 ... Image superposition part 112 ... Pixel block cutout part 113 ... Edge information acquisition part 114 ... Pixel correction part 121 ... Image storage memory 122 ... Optimal correction dictionary storage memory 231 ... Image without displacement 232 ... Image with displacement 233 ... Checkered array image 234 ... Correction target frame 401, 403 ... Pixel block 402, 404 ... Correction target region 60 ... Pixel block 61 ... Correction target region 611, 614 ... Imaginary pixels 612, 613 ... real pixels

Claims (5)

First photographing means for photographing a first image;
Taking a second image in which the pixel position is displaced by a distance half the distance between adjacent pixels in the direction of the pixel column and in a direction different from the pixel column with respect to the pixel position of the first image Second imaging means for
Based on the first image and the second image, each pixel column is composed of a plurality of pixels and empty areas where no pixels exist alternately at predetermined intervals. A checkered image that generates an image of a checkered arrangement in which pixels and empty regions included in odd-numbered pixel columns and pixels and empty regions included in even-numbered pixel columns are shifted in the direction of the pixel columns. Array image generation means ;
From the checkered array image, two adjacent pixels included in each of the odd-numbered pixel column and the even-numbered pixel column and two empty regions adjacent to both of the two adjacent pixels; Pixel block cutout means for cutting out a pixel block including a predetermined number of pixels and two empty areas around two adjacent pixels and two empty areas adjacent to both of the two adjacent pixels;
Edge information acquisition means for detecting an edge composed of two adjacent pixels having a difference in pixel value and obtaining edge information indicating a type of edge included in the pixel block;
For each piece of edge information included in a group of edge information indicating types of edges that may be included in the pixel block, both the correction method for correcting the two adjacent pixels and the two adjacent pixels Correction method storage means for storing a generation method for generating pixels corresponding to two adjacent empty regions;
The correction method and the generation method corresponding to the edge information obtained by the edge information acquisition unit are read from the correction method storage unit, and the two adjacent pixels are corrected according to the read correction method. Pixel correcting means for generating pixels corresponding to two empty regions adjacent to both of the two adjacent pixels according to the generated generation method;
An image correction apparatus comprising: The edge information acquisition means obtains edge information according to the presence or absence of an edge in the two adjacent pixels,
The correction method storage means stores the correction method and the generation method corresponding to edge information according to the presence or absence of an edge in the two adjacent pixels,
The pixel correction means corrects the two adjacent pixels according to the correction method and the generation method corresponding to edge information according to the presence or absence of an edge in the two adjacent pixels, and both the two adjacent pixels are corrected. Generate pixels corresponding to two empty regions adjacent to
The image correction apparatus according to claim 1. The edge information acquisition means obtains edge information corresponding to an edge existing in the direction of the pixel column and an edge existing in a direction different from the pixel column,
The correction method storage means stores the correction method and the generation method corresponding to edge information corresponding to an edge existing in the direction of the pixel column and an edge existing in a direction different from the pixel column,
The pixel correction unit corrects the two adjacent pixels according to the correction method and the generation method corresponding to edge information corresponding to an edge existing in the direction of the pixel column and an edge existing in a direction different from the pixel column. Generating pixels corresponding to two empty regions adjacent to both of the two adjacent pixels;
The image correction apparatus according to claim 1, wherein the image correction apparatus is an image correction apparatus. A first photographing step of photographing a first image;
Taking a second image in which the pixel position is displaced by a distance half the distance between adjacent pixels in the direction of the pixel column and in a direction different from the pixel column with respect to the pixel position of the first image A second imaging step to
Based on the first image and the second image, each pixel column is composed of a plurality of pixels and empty areas where no pixels exist alternately at predetermined intervals. A checkered image that generates an image of a checkered arrangement in which pixels and empty regions included in odd-numbered pixel columns and pixels and empty regions included in even-numbered pixel columns are shifted in the direction of the pixel columns. An array image generation process ;
From the checkered array image, two adjacent pixels included in each of the odd-numbered pixel column and the even-numbered pixel column and two empty regions adjacent to both of the two adjacent pixels; A pixel block cutting step of cutting out a pixel block including a predetermined number of pixels and an empty region around two adjacent pixels and two empty regions adjacent to both of the two adjacent pixels;
An edge information acquisition step of detecting an edge composed of two adjacent pixels having a difference in pixel value and obtaining edge information indicating a type of edge included in the pixel block;
For each piece of edge information included in a group of edge information indicating types of edges that may be included in the pixel block, both the correction method for correcting the two adjacent pixels and the two adjacent pixels The correction method and the generation corresponding to the edge information obtained in the edge information acquisition step from a correction method storage means storing a generation method for generating pixels corresponding to two adjacent empty regions Read out the method, correct the two adjacent pixels according to the read correction method, and correspond to two empty regions adjacent to both of the two adjacent pixels according to the read generation method A pixel correction step for generating pixels;
An image correction method comprising: On the computer,
A first shooting procedure for shooting a first image;
Taking a second image in which the pixel position is displaced by a distance half the distance between adjacent pixels in the direction of the pixel column and in a direction different from the pixel column with respect to the pixel position of the first image A second shooting procedure,
Based on the first image and the second image, each pixel column is composed of a plurality of pixels and empty areas where no pixels exist alternately at predetermined intervals. A checkered image that generates an image of a checkered arrangement in which pixels and empty regions included in odd-numbered pixel columns and pixels and empty regions included in even-numbered pixel columns are shifted in the direction of the pixel columns. A sequence image generation procedure ;
From the checkered array image, two adjacent pixels included in each of the odd-numbered pixel column and the even-numbered pixel column and two empty regions adjacent to both of the two adjacent pixels; A pixel block cutout procedure for cutting out a pixel block including a predetermined number of pixels and an empty region around two adjacent pixels and two empty regions adjacent to both of the two adjacent pixels;
An edge information acquisition procedure for detecting an edge composed of two adjacent pixels having a difference in pixel value and obtaining edge information indicating a type of edge included in the pixel block;
For each piece of edge information included in a group of edge information indicating types of edges that may be included in the pixel block, both the correction method for correcting the two adjacent pixels and the two adjacent pixels The correction method and the generation corresponding to the edge information obtained in the edge information acquisition procedure from the correction method storage means storing a generation method for generating pixels corresponding to two adjacent empty regions Read out the method, correct the two adjacent pixels according to the read correction method, and correspond to two empty regions adjacent to both of the two adjacent pixels according to the read generation method A pixel correction procedure for generating pixels;
An image correction program for executing

Images