JP4866340B2 - Image reduction apparatus and image reduction method - Google Patents

Description

  The present invention relates to an image reduction device, an image reduction method, and a computer program for controlling the image reduction device.

In compression based on JPEG (Joint Photographic Experts Group), orthogonal transformation processing is performed on image data. The orthogonal transform coefficient obtained by the orthogonal transform is recorded on the recording medium. In the case of reducing the compressed image obtained by the compression based on JPEG, there is an example in which the orthogonal transform coefficient used for the reduction is determined according to the reduction ratio and the reduction process is performed using the determined orthogonal transform coefficient ( Patent Document 1).
Japanese Unexamined Patent Publication No. 2006-60540

  Since compression based on JPEG is orthogonally transformed by an image block of 8 pixels × 8 pixels, the orthogonally transformed block obtained by orthogonally transforming the image block is also composed of 64 orthogonal transform coefficients of 8 × 8. When performing image reduction processing using orthogonal transform coefficients, if the number of pixels in the horizontal direction or the number of vertical pixels in the original image before compression is not a multiple of 8, the horizontal length or the vertical direction of the reduced image The length of is not an integer pixel. For this reason, the width or height of the edge pixel representing the edge of the reduced image obtained by the reduction process is theoretically a length indicated by a decimal number. Since the length indicated by the decimal cannot be expressed, rounding-up processing is performed, and the edge pixel is expressed by one pixel. In a theoretically reduced image, the edge portion is represented by one pixel even though it does not have a width or height of one pixel (non-integer pixel), so that the edge portion of the reduced image may be uncomfortable.

  An object of the present invention is to reduce a sense of incongruity that occurs at the edge portion of a reduced image.

  The image reduction device according to the present invention is a rectangular image block having a predetermined number of pixels in both the horizontal direction and the vertical direction, and each image block of a plurality of image blocks constituting one original image is orthogonally transformed. For each of the orthogonal transform blocks of a plurality of orthogonal transform blocks composed of a plurality of orthogonal transform coefficients, the horizontal and vertical directions are reduced to n / a predetermined number of pixels (where n is a natural number less than the predetermined number of pixels). An orthogonal transform coefficient determining means for determining an orthogonal transform coefficient to be used, and performing an inverse orthogonal transform process on a plurality of orthogonal transform blocks using the orthogonal transform coefficient determined by the orthogonal transform coefficient determining means, and a plurality of reduced images Reduced image block generating means for generating a block, and among the plurality of reduced image blocks generated by the reduced image block generating means, For a padded reduced image block obtained based on a padded image block to which padding data is added so that both of the horizontal and vertical directions have a predetermined number of pixels, a padded reduced image If the edge pixel corresponding to the edge of the reduced image among the pixels included in the block should be a non-integer pixel, when the non-integer value is less than 0.5, the level of the pixel of the reduced image adjacent to the edge pixel The edge pixel is deleted by adding the level of the edge pixel, and when the non-integer value is 0.5 or more, addition processing is performed to add the level of the pixel of the reduced image adjacent to the non-integer pixel to the edge pixel, and the edge of the reduced image When the edge pixel corresponding to is originally an integer pixel, the padding reduced image block for stopping the addition process is performed. And a fully reduced image block excluding a padded reduced image block among a plurality of reduced image blocks generated by the reduced image block generating means, and a padding reduction that has undergone addition processing by the padded reduced image block correcting means The image processing apparatus includes reduced image generation means for generating a reduced image using the image block and the padded reduced image block for which the addition processing by the padded reduced image block correcting means is stopped.

  The present invention also provides a reduced image generation method suitable for the image reduction device. That is, in this method, the orthogonal transform coefficient determining means is a rectangular image block having a predetermined number of pixels in both the horizontal direction and the vertical direction, and each image of a plurality of image blocks constituting one frame of the original image. N / predetermined number of pixels in the horizontal direction and the vertical direction for each orthogonal transform block composed of a plurality of orthogonal transform coefficients obtained by orthogonally transforming the block (where n is less than the predetermined number of pixels) The orthogonal transform coefficient to be used when the image is reduced to a natural number) is determined, and the reduced image block generation means performs inverse orthogonal transform processing on a plurality of orthogonal transform blocks using the orthogonal transform coefficient determined by the orthogonal transform coefficient determination means. And a plurality of reduced image blocks are generated, and the reduced image block correcting means generates a plurality of reduced image blocks generated by the reduced image block generating means. Among the plurality of image blocks, the reduced padding image block obtained based on the padding image block to which padding data is added so that both the horizontal direction and the vertical direction have a predetermined number of pixels. If the edge pixel corresponding to the edge of the reduced image among the pixels included in the padding reduced image block should be a non-integer pixel, the reduced image adjacent to the edge pixel when the non-integer value is less than 0.5 Adding the level of the edge pixel to the level of the pixel, deleting the edge pixel, and adding the level of the pixel of the reduced image adjacent to the non-integer pixel to the level of the edge pixel when the non-integer value is 0.5 or more If the edge pixel corresponding to the edge of the reduced image was originally an integer pixel, stop the above addition process The reduced image generating means performs the addition processing by the fully reduced image block excluding the padded reduced image block among the plurality of reduced image blocks generated by the reduced image block generating means and the padded reduced image block correcting means. A reduced image is generated using the padded reduced image block and the padded reduced image block for which the addition processing by the padded reduced image block correcting means has been stopped.

  The present invention also provides a computer program for controlling the image reduction device.

  According to the present invention, among the orthogonal transform coefficients constituting the orthogonal transform block, an orthogonal transform coefficient to be used for reduction is determined. An inverse orthogonal transform process is performed using the determined orthogonal transform coefficient to generate a plurality of reduced image blocks. For a padded reduced image block obtained based on a padded image block to which padding data has been added so that both the horizontal direction and the vertical direction have a predetermined number of pixels, among the pixels included in the reduced padding image block, It is determined whether the edge pixel corresponding to the edge of the image should have been a non-integer pixel, and if it should have been a non-integer pixel, if the non-integer value is less than 0.5, the reduced image adjacent to the edge pixel An addition process that adds the level of the edge pixel to the level of the pixel, deletes the edge pixel, and adds the level of the pixel of the reduced image adjacent to the non-integer pixel to the level of the edge pixel when the non-integer value is 0.5 or more. Done. If the pixel corresponding to the edge of the reduced image is originally an integer pixel, this addition process is stopped. A reduced image is generated using the completely reduced image block, the padded reduced image block that has undergone addition processing, and the padded reduced image block that has undergone addition processing.

  As described above, when the edge pixel corresponding to the edge of the reduced image should be a non-integer pixel, the edge pixel is set to the level of the pixel of the reduced image adjacent to the non-integer pixel when the non-integer value is less than 0.5. An edge pixel is deleted by adding a level. When the non-integer value is 0.5 or more, an addition process is performed to add the level of the pixel of the reduced image adjacent to the non-integer pixel to the level of the edge pixel. Even if the edge portion of the reduced image generated by the reduced image in theory has one pixel even though the edge portion does not have the width or height of one pixel, the level of one pixel of the edge portion is shown. Since the level is adjusted so as to be inconspicuous compared with the neighboring pixels, it is possible to reduce a sense of incongruity generated at the edge portion of the reduced image.

  The reduced block generation means performs an inverse orthogonal transform process on the modified orthogonal transform block obtained by orthogonally transforming the padding image block, and restores the padding image block restoration means for restoring the padding image block and the padding image block restoration means. A modified image block generating unit configured to generate a modified image block in which the padding data included in the padded image block is replaced with data obtained by copying the original image data included in the padded image block, and is generated by the modified image block generating unit; It is preferable to reduce the corrected image block according to the reduction ratio to generate a reduced image block.

  FIG. 1 shows an embodiment of the present invention and is a block diagram showing an electrical configuration of an image reduction apparatus.

  The overall operation of the image reduction apparatus 1 is controlled by the CPU 6.

  The image reduction device 1 includes a memory card interface 8. The memory card 9 stores JPEG compressed image data. The memory card interface 8 reads compressed image data stored in the memory card 9. The read compressed image data is expanded, and an image represented by the compressed image data is displayed on the display screen of the display device 2. In particular, in this embodiment, a reduced image of the image represented by the compressed image data is generated. The generated reduced image is also displayed on the display screen of the display device 2. Details of the reduced image generation method will be described later. Further, if the processing program for the image reduction device 1 is stored in the memory card 9, the program can be read and installed in the image reduction device 1. Of course, the processing program can be installed in the image reduction apparatus 1 by other methods.

  The image reduction device 1 includes an input device 7 such as a keyboard. The command output from the input device 7 is given to the CPU 6.

  The image reduction device 1 also includes a hard disk 5 and a hard disk drive 4 that reads compressed image data or the like stored in the hard disk 5 or writes compressed image data or the like to the hard disk 5. An image represented by compressed image data recorded on the hard disk 5 and a reduced image thereof can be displayed on the display screen of the display device 2.

  Furthermore, the image reduction device 1 also includes a memory 3 that temporarily stores image data and the like.

  FIG. 2 is a flowchart showing a processing procedure of image display.

  In the processing shown in FIG. 2, compressed image data based on JPEG is recorded in the memory card 9, and a reduced image of the image represented by the compressed image data is displayed. Not only compression based on JPEG, but also compression based on orthogonal transformation may be performed.

  Desired compressed image data is read from the compressed image data representing the images of many frames recorded on the memory card 9 (step 11). For example, when the data name (file name) of the compressed image data is designated, the compressed image data having the designated data name is read from the memory card 9.

  An image reduction process is performed on the read compressed image data (step 12). This image reduction process will be described later in detail. The reduced image generated by the image reduction process is displayed on the display screen of the display device 2 (step 13). The reduction ratio of the reduced image may be determined in advance or may be input from the input device 7.

  3A to 3D, FIGS. 4A to 4D, and FIGS. 5A to 5F show the principle of reduced image generation according to this embodiment.

  FIG. 3A shows the original image 15 before the compressed image is generated.

  When compression based on JPEG is performed on the original image 15, the original image 15 is divided into a large number of image blocks 16 composed of 64 pixels of 8 pixels (predetermined number of pixels) in both the horizontal and vertical directions. It is assumed that the rightmost block 20 of the original image 15 is not a block of 8 pixels × 8 pixels but a block 20 of 9 pixels in the horizontal direction and 8 pixels in the vertical direction.

  FIG. 3B shows one of the blocks 20 at the right end of the original image 19 shown in FIG.

  As described above, the block 20 is composed of 9 pixels in the horizontal direction and 8 pixels in the vertical direction. Among the pixels constituting the block 20, the pixels 22 and 32 that are painted black have a pixel level of 100 percent, and the white pixels 23 and 33 have a pixel level of 0 percent. The pixels constituting the block 20 can be divided into an image block 21 of 8 pixels × 8 pixels on the left side and a pixel group 30 composed of 8 pixels in one column on the right side.

  FIG. 3C shows an image block 21 of 8 pixels × 8 pixels on the left side among the pixels constituting the block 20 shown in FIG. In the image block 21 shown in FIG. 3C, all of the pixels 22 or 23 constituting the image block 21 represent the original image 15 shown in FIG.

  FIG. 3D shows an image block 31 in which a pixel group 30 composed of eight pixels in one column on the right side among the pixels constituting the block 20 shown in FIG. Is shown. Data called padding data (pixels) 34 is added to the right side of the pixel group 30 located on the leftmost side of the image block 31. In the compression based on JPEG, compression processing is performed in units of 64 pixel blocks of 8 pixels × 8 pixels, so the data added so that the image block becomes 64 pixels of 8 pixels × 8 pixels is padding data 34. is there. A block to which padding data is added is referred to as a padding image block.

  FIG. 4A shows a complete orthogonal transform block 25 obtained by orthogonal transform of the complete image block 21 shown in FIG. The complete orthogonal transform block 25 includes 8 × 8 64 orthogonal transform coefficients C1 to C64. The upper left is the orthogonal transformation coefficient of the low frequency component, and the lower right is the orthogonal transformation coefficient of the high frequency component. The orthogonal transform block 25 is encoded and recorded on the recording medium.

  FIG. 4B shows a padding orthogonal transformation block 35 obtained by orthogonal transformation of the padding image block 31 shown in FIG. Since the padding image block 31 is also composed of 64 pixels of 8 pixels × 8 pixels, the padding orthogonal transform block 35 also includes 8 × 8 64 orthogonal transform coefficients C65 to C128.

  The complete orthogonal transform block 25 and the padding orthogonal transform block 35 shown in FIGS. 4A and 4B are encoded and recorded on the memory card 9.

  When generating a reduced image of a JPEG-compressed original image, the complete orthogonal transform block 25 is selected from among the orthogonal transform coefficients C1 to C64 constituting the complete orthogonal transform block 25 as shown in FIG. An orthogonal transform coefficient is determined from the low frequency component according to the reduction ratio. For example, when generating a reduced image in which both the horizontal direction and the vertical direction are ½ of the original image, the complete orthogonal transform block 25 is composed of 64 orthogonal transform coefficients C1 to C64 of 8 × 8. Therefore, the orthogonal transform coefficients C1 to C4, C9 to C12, C17 to C20, and C25 to C28 included in the 4 × 4 block 25A in which both the horizontal direction and the vertical direction are ½ are generated as reduced images. It is determined as an orthogonal transform coefficient used for.

  The determined 4 × 4 16 orthogonal transform coefficients C1 to C4, C9 to C12, C17 to C20, and C25 to C28 are subjected to inverse orthogonal transform, whereby a fully reduced image block 26 shown in FIG. Is obtained. The fully reduced image block 26 includes white pixels 27, pixels 28 indicated by cross hatching, and pixels 29 indicated by vertical stripes. The white pixel 27 has a pixel level of 0 percent as in the white pixel 23 included in the complete image block 21 shown in FIG. 3C, and the pixel 28 indicated by cross-hatching is a 75 percent pixel. The pixel 29 indicated by vertical stripes is at a pixel level of 25%. The fully reduced image block 26 is considered to be composed of pixels obtained by reducing four adjacent pixels among the pixels 22 or 23 constituting the complete image block 21 shown in FIG. It will be understood that a fully reduced image block 26 shown in FIG. 4 is generated by generating a reduced image block from a complete orthogonal block 25 in which the image block 21 shown in C) is orthogonally encoded.

  For the padding image block 35 shown in FIG. 4, all orthogonal transform coefficients C65 to C128 constituting the padding image block 35 are determined as orthogonal transform coefficients used for the reduced image, and inverse orthogonal transform is performed. The padding image block 35 shown in FIG. 4C is subjected to inverse orthogonal transform, whereby the padding image block 31 shown in FIG. 4D is restored (same as the padding image block shown in FIG. 3D). Is done. In this embodiment, the padding data 34 constituting the padding image block 31 is replaced with a copy of the pixels 32 and 33 constituting the original image 15.

  FIG. 5B shows a modified padding image block 35 in which the padding data is replaced with pixels 32 and 33 constituting the original image 15.

  The first row of the modified padding image block 35 is a pixel 32 having a pixel level of 100 percent, and the padding data 34 is replaced so that the second to eighth rows are pixels 33 having a pixel level of 0 percent. It has been. The modified padding image block 35 is reduced.

  FIG. 5C shows a padding reduced image block 36 in which the modified padding image block 35 is reduced.

  The padding reduced image block 36 includes a pixel 37 having a pixel level of 50% and a pixel 38 having a pixel level of 0%. It will be understood that the padding reduced image block 36 shown in FIG. 5C is obtained by reducing the modified padding image block 35 shown in FIG. 5B.

  Of the original image block 31 shown in FIG. 3D, the pixels constituting the original image 15 are the pixel group 30 in the leftmost column. When the original image 15 is reduced to ½ in both the horizontal direction and the vertical direction, the width and height of the pixel group 30 in one column are halved. Therefore, a reduced block of 1/2 pixel in the horizontal direction and 4 pixels in the vertical direction must be extracted from the padded reduced image block 36, but non-integer pixels cannot be extracted. As shown, an adjustment pixel block 39 having a width rounded up from the padding reduced image block 36 is extracted.

  As shown in FIG. 5 (E), the fully reduced image block 26 shown in FIG. 5 (A) and the adjustment pixel block 39 shown in FIG. 5 (D) are combined to reduce the image block shown in FIG. 3 (B). An image block 40 is generated.

  The image block 20 shown in FIG. 3B has 9 pixels in the horizontal direction and 8 pixels in the vertical direction. When the image block 20 is reduced to 1/2 in both the horizontal direction and the vertical direction, FIG. As shown in F), it should be a reduced image block 41 of 4.5 pixels in the horizontal direction and 4 pixels in the vertical direction. However, since a reduced image block having a non-integer number of pixels cannot be generated, as described above, padding is performed. An adjusted image block 39 having a rounded-up pixel number is extracted from the reduced image block 36, and a reduced image block 40 having 5 pixels in the horizontal direction and 4 pixels in the vertical direction is generated as shown in FIG.

  As shown in the rightmost pixels 37A and 38A in FIG. 5F, the non-integer pixel originally having a width of 0.5 pixels is the edge pixel in order to correspond to the rightmost edge pixel in FIG. 5E (the edge of the reduced image). As shown in 37 and 38, since the pixels are one pixel wide, these edge pixels become conspicuous, and when a reduced image is viewed, a sense of incongruity may occur. For this reason, in this embodiment, the following processing is performed.

  In the above-described embodiment, the padding orthogonal transform block 35 is restored to the padding image block 31 to generate the padding reduced image block 36. However, the padding image block 31 is not restored and the padding orthogonal transform block 35 is restored in the same manner as the complete orthogonal transform block 25. An orthogonal transform coefficient corresponding to the reduction ratio may be determined to generate a padded reduced image block.

  FIG. 6 shows processing when the original width (non-integer value) of the edge pixels constituting the adjustment pixel block extracted from the padded reduced image block is less than 0.5 pixels.

  When the original width of the edge pixel IB constituting the adjustment pixel block 52 is less than 0.5, the edge pixel IB is added to the rightmost pixel IA of the fully reduced image block 50 adjacent to the adjustment pixel block 52. It is added at a rate according to the width. By adding the edge pixel IB to the pixel IA, the pixel IA of the fully reduced image block 50 becomes the pixel IC of the fully reduced image block 50A. The edge pixel IB is deleted. The reduced image portion consisting of 5 pixels in the horizontal direction and 4 pixels in the vertical direction composed of the fully reduced image block 50 and the pixel block 52 becomes a fully reduced image block 50A of 4 pixels in the horizontal direction and 4 pixels in the vertical direction, and the number of pixels in the horizontal direction Decrease. Similarly, the edge pixels of the pixel block 52 other than the edge pixel IB are added to the rightmost pixel of the fully reduced image block 50 and deleted.

  FIG. 7 shows a process when the original width of the edge pixel constituting the adjustment pixel block extracted from the padding reduced image block is 0.5 pixels or more.

  When the original width of the edge pixel IY constituting the adjustment pixel block 62 is 0.5 or more, the rightmost pixel IX of the fully reduced image block 60 adjacent to the adjustment pixel block 62 corresponds to the width of the edge pixel IY. It is added to the edge pixel IY at a rate. By adding the pixel IX to the edge pixel IY, the edge pixel IY of the image block 62 becomes the pixel IZ of the new image block 62A. The horizontal direction consisting of the fully reduced image block 60 and the image block 62A without changing the number of pixels of the reduced image portion of 5 pixels in the horizontal direction composed of the fully reduced image block 60 and the pixel block 62 and 4 pixels in the vertical direction. The size of 5 pixels and 4 pixels in the vertical direction is maintained.

  FIG. 8 is a flowchart showing the image reduction processing procedure (the processing procedure of step 12 in FIG. 2), and FIGS. 9 and 10 show how the original image is reduced. 9 and 10, the original image Io (w0, h0) has 17 pixels in the horizontal direction and 16 pixels in the vertical direction. In FIG. 9, a reduced image in which both the horizontal direction and the vertical direction are 1/4. FIG. 10 shows a state where a reduced image in which the horizontal direction and the vertical direction are both 3/4 is generated. When the number of pixels of an ideal reduced image in the calculation described later is not a non-integer but an integer, the above-described addition process is stopped.

  The JPEG-compressed original image Io is orthogonally transformed, but for the sake of clarity, the inverse transformation is performed from the orthogonal transformation coefficient to determine the orthogonal transformation coefficient used to generate the reduced image, and the reduction process is performed. Needless to say, the inverse orthogonal transformation or the like is performed as described above. First, a case where a 1/4 reduced image is generated will be described.

  When a reduced image is generated from the original image Io, the horizontal pixel w1 (5 pixels) rounded up so that the number of pixels becomes an integer when the number of pixels of the ideal reduced image in calculation is a non-integer , A rounded down reduced image Ir1 of the vertical pixel h1 (4 pixels) is generated (step 71). Further, the sizes of the horizontal pixel w0 (17 pixels) and the vertical pixel ho (16 pixels) of the original image Io are calculated (step 72).

  Subsequently, the horizontal pixel w (four pixels) and the vertical pixel h (four pixels) of the output reduced image Ir are calculated (step 73). This reduced image for output Ir is rounded off to the first decimal place when the number of pixels in the horizontal direction and the number of pixels in the vertical direction of the ideal reduced image Ir2 obtained by reducing the original image Io becomes a non-integer. Can be obtained.

  Further, the internal pixels constituting the rounded-down reduced image Ir1 are copied except for the pixels at the right end and the lower end to be the pixels of the output reduced image Ir (indicated by hatching 81 in FIG. 74).

  Subsequently, distances d1 and d2 from the right end and the lower end of the rounded down reduced image Ir to the right end and the lower end of the ideal reduced image Ir2 are calculated from Equations 1 and 2 (Step 75).

d1 = w1-w0 × n / 8 Equation 1
d2 = h1-h0 × n / 8 Equation 2
However, n is an integer from 1 to 7, n / 8 indicates the reduction ratio, and n is 2 in the example shown in FIG.

  In the example shown in FIG. 9, since the lower end of the rounded down reduced image Ir1 is the same as the lower end of the ideal reduced image Ir2, the distance from the lower end of the rounded down reduced image Ir1 to the lower end of the ideal reduced image Ir2 is zero. d1 is 3/4.

Subsequently, the pixel level of the rightmost pixel w of the output image Ir is calculated according to Equation 3 (step 76). However, y indicates that the y-coordinate values of the pixels w1, w1-1, and w are the same.
Ir (w, y) = d1 * Ir1 (w1-1, y) + (1-d1) * Ir1 (w1, y)
... Formula 3

Similarly, the pixel level of the pixel h at the lower end of the output image Ir is calculated according to Equation 4 (step 77). However, x indicates that the x-coordinate values of the pixels h1, h1-1, and h are the same.
Ir (x, h) = D2 × Ir1 (x, h1-1) + (1-d1) × Ir1 (x, h1)
... Formula 4

  The output reduced image Ir thus generated is output (displayed) (step 78). In the case of the ideal reduced image Ir2, a pixel having a width of 0.25 should exist at the right end, but the pixel having a width of 0.25 pixel is discarded. The pixel output reduced image Ir is smaller by one pixel in the horizontal direction than the rounded-down reduced image Ir1, but the edge pixels of the output reduced image Ir1 are level-adjusted, so that a sense of incongruity does not occur.

  Referring to FIG. 10, when a 3/4 reduced image of the original image Io is generated, the ideal reduced image Ir2 has a horizontal direction of 17 pixels × 3/4 = 12.75 pixels and a vertical direction of 16 pixels × 12 pixels. It becomes. The rounded down reduced image Ir1 and the output reduced image Ir both have 13 pixels in the horizontal direction and 12 pixels in the vertical direction. The rightmost pixel of the output image Ir2 is originally one pixel having a width of 0.75 pixels. However, as described above, since the level is adjusted, the uncomfortable feeling is reduced. The rounded-down reduced image Ir1 and the output reduced image Ir have the same size, but are shifted for easy understanding.

  In the above-described embodiment, only the right end of the reduced image is described, but it goes without saying that the same processing is performed on the lower end. Also, since the original image Io and the like define the right direction as the positive direction of the X coordinate system and the downward direction as the positive direction of the Y coordinate, and the origin of the XY coordinate system is aligned with the upper left vertex of the original image Io and so on, The processing described above is performed for the right end and the bottom end, but it goes without saying that the processing described above is performed for the right end and the top end of the reduced image according to the direction of the XY coordinate system by changing the direction of the XY coordinate system. Absent.

It is a block diagram which shows the electric constitution of an image reduction device. It is a flowchart which shows an image display process procedure. (A) to (D) show how a reduced image is generated. (A) to (D) show how a reduced image is generated. (A) to (F) show how a reduced image is generated. The pixel level adjustment process is shown. The pixel level adjustment process is shown. It is a flowchart which shows the process sequence of an image reduction process. It shows how a reduced image is generated. It shows how a reduced image is generated.

Explanation of symbols

1 Image reduction device 6 CPU
16, 20, 21 image block
26 Fully reduced image block
31 Padding image block
36 Padding reduced image block

Claims (3)

A rectangular image block having a predetermined number of pixels in both the horizontal direction and the vertical direction, and composed of a plurality of orthogonal transform coefficients obtained by orthogonally transforming each image block of a plurality of image blocks constituting one frame of the original image Orthogonal transform coefficient used when each of the orthogonal transform blocks is reduced to n / predetermined number of pixels (where n is a natural number less than the predetermined number of pixels) in the horizontal and vertical directions. Orthogonal transform coefficient determining means for determining
Reduced image block generating means for performing inverse orthogonal transform processing on a plurality of orthogonal transform blocks using the orthogonal transform coefficient determined by the orthogonal transform coefficient determining means and generating a plurality of reduced image blocks;
Of the plurality of reduced image blocks generated by the reduced image block generation means, padding data is added so that the predetermined number of pixels is obtained in both the horizontal direction and the vertical direction among the plurality of image blocks. For padded reduced image blocks obtained based on the padded image block, the edge pixels corresponding to the edges of the reduced image among the pixels included in the padded reduced image block should be non-integer pixels. When the numerical value is less than 0.5, the edge pixel is added to the level of the pixel of the reduced image adjacent to the edge pixel to delete the edge pixel, and when the non-integer value is 0.5 or more, the reduced image adjacent to the non-integer pixel The edge image corresponding to the edge of the reduced image is added by adding the level of the pixel to the edge pixel level. Is originally an integer pixel, the padding reduced image block correcting means for stopping the addition process, and the padding reduced image block excluding the padded reduced image block among the plurality of reduced image blocks generated by the reduced image block generating means A reduced image using a reduced image block, a padded reduced image block that has undergone addition processing by the padded reduced image block correcting means, and a padded reduced image block for which addition processing has been stopped by the padded reduced image block correcting means. Reduced image generation means for generating
An image reduction device comprising: The orthogonal transform coefficient determining means is a rectangular image block having a predetermined number of pixels in both the horizontal direction and the vertical direction, and each of the plurality of image blocks constituting one frame of the original image is orthogonally transformed. For each orthogonal transform block of a plurality of orthogonal transform blocks composed of a plurality of orthogonal transform coefficients, the horizontal and vertical directions are reduced to n / predetermined number of pixels (where n is a natural number less than the predetermined number of pixels). Determine the orthogonal transform coefficients to use
Reduced image block generation means performs inverse orthogonal transform processing on a plurality of orthogonal transform blocks using the orthogonal transform coefficient determined by the orthogonal transform coefficient determination means, and generates a plurality of reduced image blocks,
The reduced image block correcting unit is configured so that, among the plurality of reduced image blocks generated by the reduced image block generating unit, the predetermined number of pixels is obtained in both the horizontal direction and the vertical direction among the plurality of image blocks. For the padded reduced image block obtained based on the padded image block to which padding data is added, the edge pixel corresponding to the edge of the reduced image among the pixels included in the padded reduced image block should be a non-integer pixel. If the non-integer value is less than 0.5, the edge pixel level is added to the level of the pixel of the reduced image adjacent to the edge pixel and the edge pixel is deleted. If the non-integer value is 0.5 or more, the non-integer value Addition process that adds the pixel level of the reduced image adjacent to the pixel to the level of the edge pixel, The addition process is stopped if the edge pixels corresponding to Tsu di were those original integer pixel,
The reduced image generating means includes a fully reduced image block excluding a padded reduced image block among a plurality of reduced image blocks generated by the reduced image block generating means, and padding subjected to addition processing by the padded reduced image block correcting means Generating a reduced image using the reduced image block and the padded reduced image block in which the addition processing by the padded reduced image block correcting means is stopped;
Reduced image generation method. A computer program for controlling an image reduction device,
A rectangular image block having a predetermined number of pixels in both the horizontal direction and the vertical direction, and composed of a plurality of orthogonal transform coefficients obtained by orthogonally transforming each image block of a plurality of image blocks constituting one frame of the original image Orthogonal transform coefficient used when each of the orthogonal transform blocks is reduced to n / predetermined number of pixels (where n is a natural number less than the predetermined number of pixels) in the horizontal and vertical directions. Determine
A plurality of reduced image blocks are generated by causing a plurality of orthogonal transform blocks to perform inverse orthogonal transform processing using the determined orthogonal transform coefficients,
Among the plurality of generated reduced image blocks, obtained based on the padding image block to which padding data is added so that both the horizontal direction and the vertical direction have a predetermined number of pixels. If the edge pixel corresponding to the edge of the reduced image among the pixels included in the reduced padding image block should be a non-integer pixel, the non-integer value is less than 0.5. The edge pixel is deleted by adding the level of the edge pixel to the level of the pixel of the reduced image adjacent to the edge pixel. When the non-integer value is 0.5 or more, the level of the pixel of the reduced image adjacent to the non-integer pixel is set to the edge. If addition processing is added to the pixel level, the edge pixel corresponding to the edge of the reduced image was originally an integer pixel. The addition process is stopped, the
Among the plurality of generated reduced image blocks, a completely reduced image block excluding the padded reduced image block, a padded reduced image block subjected to the addition process, and a padded reduced image block from which the addition process has been stopped are used. And a computer program for controlling the image reduction device to generate a reduced image.

Images