JP2006323870A - Image subtractive color processing device, image encoding device, image decoding device, image subtractive color processing method, image encoding method and image decoding method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image subtractive color processing device capable of subtracting the number of colors of an original image with the image prevented from being deteriorated. <P>SOLUTION: The image subtractive color processing device, which outputs an input digital image data with the subtracted number of gradations, includes a dividing section 602 which divides the input image into small areas with a predetermined process, and adaptive subtractive color processing sections 616-620 which sets the number of gradations for each small area to appropriately subtract colors of each small area based on image data included in the small area. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image color reduction device for compressing data by reducing the number of gradations of an image, an image encoding device, and an image decoding device for decoding such a compressed image, and in particular, after color reduction. The present invention relates to an image color reduction device, an image encoding device, and an image decoding device that can prevent image degradation.

  When performing data communication and data storage, it is desirable to reduce the amount of data in order to save the storage area. In particular, since image data has a large amount of data, various techniques for reducing the amount of data have been devised. As an image encoding method for the purpose of reducing the amount of data, there is an image encoding method called an entropy encoding method. Entropy coding methods such as arithmetic coding and Huffman coding are known.

  However, before entropy coding is performed, the amount of image data itself is often reduced by image processing. As a typical example, there is a method of reducing the color of a multi-value image to a sufficient number of gradations according to the purpose of use. The simplest method is to reduce the entire image to a common number of tones. However, image attributes are generally not uniform throughout. For example, even a simple text image is divided into a monochrome background area and a character area that requires at least two or more pixel values. Therefore, if the entire color is uniformly reduced, for example, a character area may be reduced more than necessary, and the image may be deteriorated.

  In order to solve these problems, a technique has been developed in which an image is divided into small areas such as blocks, and the color is reduced to a different number of gradations for each small area. An example of such a technique is disclosed in Japanese Patent Laid-Open No. 6-350986 (Patent Document 1). In the technique disclosed in this publication, a block is taken as each small region, and the number of gradations is determined for each block based on the approximation error and statistics within the block.

Another example is disclosed in Japanese Patent Laid-Open No. 5-56282 (Patent Document 2). This technique also determines the number of gradations for each block. However, in this case, the resolution of the block having a large number of gradations is reduced. By doing so, the data amount of each block becomes constant.
JP-A-6-350986 JP-A-5-56282

  The method as described above for setting the number of gradations for each small area of the image can reduce the amount of data while preventing deterioration of the image, rather than reducing the entire color uniformly. On the other hand, the above method has the following problems.

  The first problem is that the visual connection for each small area is not considered. Such a problem will be described with reference to FIGS. 59 and 60. FIG. In the following description, “small area” refers to a block having an image.

FIG. 59 shows an example of the input image. It is assumed that this input image has 256 gradations as a whole. This image includes a star region 101 having a pixel value of “150”, an arrow region 102 and a star region 103 having a pixel value of “175”, and a background region having a pixel value of “255”. Yes.

  Now, for each small area, either 4 gradations or 8 gradations are selected as the number of gradations after color reduction. In the case of four gradations, for example, the pixel value after color reduction is one of 0, 85, 170, and 255. This is four gradations obtained by equally dividing the space having the gradation number 0 to 255. Similarly, in the case of 8 gradations, for example, the pixel value takes any of 0, 36, 73, 109, 146, 182, 219, and 255. After the number of gradations after color reduction is determined, the conversion from the input image is performed by taking a value closest to the input image among the pixel values after color reduction possible with the number of gradations.

  Under such logic, it seems reasonable that the arrow region 102 shown in FIG. 59 is represented by the pixel value “170” in four gradations. The same applies to the star area 103. However, for the star region 101 (pixel value 150), the closest pixel value in the four gradations is 85 or 170, so the error becomes large regardless of which one is selected. Therefore, according to the existing logic, it may be determined that 8 gradations are necessary for the small area including the star area 101 (divided square area shown in FIG. 59). An example of the processing result in such a case is shown in FIG.

  Referring to FIG. 60, for star region 101, 146, which is 8 gradations and relatively close to original pixel value 150, is selected. Therefore, there is no particular problem with the star area 101. However, in the arrow area 102, the part in the same small area as the small area where the star area 101 exists is represented by a value closest to the original pixel value 175 in 8 gradations, that is, a pixel value of 182. ing. Therefore, this portion has a pixel value different from the pixel value “170” of the other portion of the arrow region 102. Thus, as shown in FIG. 60, a step appears in the arrow region 102 along the boundary between the lower left small region and the other small regions. This level difference was not present in the original picture. Needless to say, such a phenomenon is not preferable.

  Such a phenomenon occurs because, in the prior art, the determination of the optimum number of subtractive color gradations for each small area is performed almost in the small area.

  Another problem with the prior art is that the prior art does not take into account graphical features in determining the optimum number of tones for each block. Even if the error in the entire block is the same, the deterioration of the image may be particularly noticeable due to the graphic feature. For example, the error is not noticeable if the area is relatively flat. On the other hand, straight lines, particularly vertical or horizontal lines, are prone to looseness if they are subjected to excessive color reduction, and viewers are more likely to notice such image degradation. However, it cannot be said that the conventional technology fully considers such problems.

  SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide an image color reduction device capable of performing color reduction processing of an original image while preventing image deterioration, and image coding capable of efficiently image coding by using such an image color reduction device. Is to provide a device.

  Another object of the present invention is to provide an image decoding apparatus capable of decoding image data that has been subjected to color reduction processing for each small region and further encoded.

An image color reduction apparatus according to an aspect of the present invention is an image color reduction apparatus that outputs the input digital image data by reducing the number of gradations, and divides the input image into small regions by a predetermined method. And dividing means for each small area, and adaptive color reduction means for setting the number of gradations for appropriately reducing the color of the small area based on the image data included in the small area.

  Preferably, the adaptive color reduction means is a predetermined shape in the input image based on an extraction means for extracting a predetermined shape from the input image, for example, a straight line, and an extraction result of the predetermined shape by the extraction means. Means for setting the number of gradations of each small region so that the number of gradations of the region where the shape is present is equal to or greater than the number of gradations of the region where the predetermined shape is not present.

  More preferably, the adaptive color reduction means includes: a distribution calculation means for calculating an index representing a distribution of either background pixels or non-background pixels in at least a part of an image under different color reduction methods; Suitability determining means for determining suitability of the color reduction method based on the comparison result of the distribution calculated by the calculating means.

  More preferably, the distribution calculating unit is configured to extract a predetermined shape from the input image, and to extract background pixels and non-background pixels in the vicinity of the predetermined shape under different color reduction methods. A pixel distribution detecting unit for detecting any one of the distributions, and the suitability determining unit detects the first subtractive gradation number and the second subtractive gradation number detected by the pixel distribution detecting unit. For comparing the distribution of the background pixels and the non-background pixels and determining whether or not the second subtractive color gradation number is appropriate based on whether the comparison result satisfies a predetermined condition or not. including.

  The predetermined condition may be that a difference between the number of background pixels under the second number of subtractive color gradations and the number of background pixels under the first number of subtractive color gradations is a certain level or more. The difference between the number of non-background pixels under the first number of subtractive color gradations and the number of non-background pixels under the second number of subtractive color gradations may be a certain value or more.

  The adaptive color reduction means includes a labeling means for labeling an input image, and a small area labeling means for giving a common label to a small area including a common label based on an output of the labeling means. A configuration in which a common number of gradations is set in a small region to which a common label is assigned by the labeling unit may be used.

  The adaptive color reduction means, when two pixels adjacent across the boundary of the small area of the input image have a common pixel value, is a small color that gives a common label to the two small areas to which the two pixels belong. It is also possible to employ a configuration in which a common number of gradations is set in a small area that includes area labeling means and is given a common label by the small area labeling means.

  An image encoding device according to another aspect of the present invention provides an image color reduction device according to any one of claims 1 to 9 and an image color reduction for reducing the number of gradations of input digital image data. Image encoding means for performing entropy encoding of the image reduced in color by the means.

  An image encoding apparatus according to still another aspect of the present invention is an image encoding apparatus that divides an input image that may have a different number of gradations for each small area into bit planes and compresses the input image, and is more adjacent in an upper plane. In order to increase the continuity between pixels, an image conversion means for performing a different conversion on each pixel value according to the number of gradations of the small area to be encoded and dividing it into bit planes, and a conversion by the image conversion means And image encoding means for performing predetermined entropy encoding for each bit plane.

An image encoding apparatus according to another aspect of the present invention is an image encoding apparatus that divides an input image, which may have a different number of gradations for each small area, into bit planes, and compresses the input image, wherein all of the small areas are stored. In response to the determination means for determining whether or not the logarithms are equal and the determination means determines that the gradation numbers of all the small areas are equal, the entire input image is regarded as one small area. In response to determining that the number of gradations of all the small regions is not equal by the determination unit and the first image encoding unit for encoding the image with the number of small regions being 1, Second image encoding means for encoding each small region of the input image.

  According to still another aspect of the present invention, an image decoding apparatus includes: input means for inputting image data encoded by the image encoding apparatus according to claim 10; and image data input by the input means. And decoding means for performing decoding corresponding to the entropy encoding according to claim 10.

  An image color reduction device according to an aspect of the present invention provides an input unit for inputting image data encoded by the image encoding device according to claim 11, and an image data input by the input unit. A decoding means for performing decoding corresponding to the entropy encoding according to claim 11 and an image for performing inverse conversion corresponding to the image conversion means according to claim 11 for the image data decoded by the decoding means. Reverse conversion means.

  An image color reduction device according to an aspect of the present invention is an image decoding device for decoding image data encoded by the image encoding device according to claim 12, encoded by image encoding means. A determination unit for determining the number of small areas included in the image data; and a decoding unit for dividing and decoding the image data into the number of small areas determined by the determination unit.

  As described above, the image color reduction device according to the present invention determines the number of gradations after color reduction in a small area in consideration of the connection between the small areas. For this reason, it is possible to perform a color reduction process for an image in which a step due to a difference in each small area that has a large visual effect is unlikely to occur without limiting the arrangement of pixel values of the number of color gradations.

  The image color reduction apparatus according to the present invention performs control to increase the number of subtractive color gradations in a small area when there is a line area in the small area. Therefore, it is possible to perform image color reduction processing so that deterioration of the line area such as rattling is not noticeable.

  The image color reduction apparatus according to the present invention actually detects deterioration due to excessive color reduction around the vertical and horizontal lines of characters included in a small area, and performs color reduction with an appropriate number of gradations according to the result. . Since unreasonable color reduction is not performed, it is possible to perform image color reduction processing so that line portions included in characters are not particularly deteriorated.

  In addition, the image compression apparatus according to the present invention is configured to increase the compression effect of the bit plane after color reduction in consideration of small areas having different numbers of gradations. Therefore, more efficient image compression can be performed.

  Further, the above image decoding apparatus according to the present invention is configured to skip decoding of a part of information when the number of gradations of the small area is equal. Therefore, it is possible to perform image compression with higher encoding efficiency than in the past.

[First Embodiment]
Hereinafter, the image color reduction apparatus according to the first embodiment of the present invention will be described. In the following description, not only the description of the first embodiment, but a numerical example such as a specific data structure and the number of gradations of an image is given for easy understanding. However, these are merely illustrative examples and should not be construed as limiting the technical scope of the present invention.

  Further, in this specification, the number of gradations of a small area or image means the number of types of pixel values allowed in the small area or image, and how many types of pixels are actually included in the small area or image. It does not mean that the value appears. For example, when pixel values of 0 to 7 are allowed in a certain small area, even if only three types of values actually appear as pixel values, the small area is considered as an 8-value area.

  Referring to FIG. 1, this image color reduction device is stored in an image input device 301 such as a scanner, an input image buffer 302 for storing image data input by the image input device 301, and an input image buffer 302. An adaptive color reducer 303 for performing an adaptive color reduction process for reducing the number of gradations appropriate for each small area to an appropriate minimum color number, and the color-reduced image data output from the adaptive color reducer 303. The color-reduced image buffer 304 for storing, the output image creator 305 for converting the image data stored in the color-reduced image buffer 304 into an image of 256 gradations, and the output, and the output of the output image creator 305 An output image buffer 306 for storing images of 256 gradations, and a disk for displaying or printing the data stored in the output image buffer 306. And an image output device 307 consisting of lay or printer.

  Here, a scanner is assumed as the image input device 301. However, the image input device is not limited to the scanner, and may be any other image input device such as a digital camera or a storage device that stores image data received through communication.

  The input image buffer 302 stores the number of pixels in the horizontal direction, the number of pixels in the vertical direction, and the number of gradations in addition to the data of each pixel. FIG. 2 shows the data structure of the image data stored in the input image buffer 302. Referring to FIG. 2, the image data includes a horizontal pixel number (LX) 501, a vertical pixel number (LY) 502, a gradation number 503, and image data 504. In this example, it is assumed that the horizontal pixel number 501, vertical pixel number 502, and gradation number 503 are all 4-byte integer data. In the following description, unless otherwise specified, an integer refers to a 4-byte integer. The image data 504 includes pixels as many as the number of horizontal pixels 501 × the number of vertical pixels 502, and 1 byte (8 bits) is assigned to each pixel. Assume that the image is a 256 grayscale image.

  In the image data stored in the input image buffer 302, the pixel value of the background area has a value close to 255, and the pixel value of the non-background area has a value close to 0. Note that the subtractive color image buffer 304 and the output image buffer 306 also have the same data structure as the input image buffer 302.

  The output image creator 305 refers to the small area gradation number buffer 617 (FIG. 3) and the reference pixel value table 622 (FIG. 3), and reduces the image of the reduced color image buffer 304 for each small area to the 2560th floor. It has a function of converting it into a key and outputting it to the output image buffer 306.

Referring to FIG. 3, the adaptive color reducer 303 shown in FIG. 1 includes a controller 601 for controlling various components described below to perform an adaptive color reduction process. The adaptive color reducer 303 further includes a small area information generator 602, a small area number register 603, a small area information buffer 604, an image binarizer 605, a binary image buffer 606, and a line extractor 607. The line number register 608, the line information buffer 609, the character rectangle information extractor 610, the character rectangle number register 611, the character rectangle information buffer 612, the first counter 613, the character rectangle color reducer 614, and the character rectangle Tone number buffer 615, small area color reducer 616, small area gradation number buffer 617, line area gradation number register 618, small area gradation number verifier 619, small area simple color reducer 620, A pixel value converter 621 and a reference pixel value table 622 are included.

  The controller 601 controls other components of the adaptive color reducer 303 and controls data communication between the other components and between each component and the outside of the adaptive color reducer 303. is there. The controller 601 is always involved in the operation of each component of the adaptive color reducer 303 described below. However, for the sake of simplification of description, the controller 601 may not be referred to in places where it seems that there is no problem in understanding.

  Each of the character rectangular gradation number buffer 615 and the small area gradation number buffer 617 is an integer array. The i-th elements of these two buffers 615 and 617 are written as RLB [i] and ALB [i], respectively.

  The small area information generator 602 generates information for dividing the image data (hereinafter referred to as “original image”) stored in the input image buffer 302, and stores the small area image in the small area number register 603. The small area information buffer 604 has a function of storing information related to each small area including the position of each small area.

  There are various methods for dividing an image into small regions. Here, a method is used in which an image is divided into small square blocks each having 8 pixels horizontally and 8 pixels vertically.

  A configuration example of the small area information buffer 604 is shown in FIG. Referring to FIG. 5, in small area information buffer 604, a set of x and y coordinates of the upper left corner of each small area (block) and x and y coordinates of the lower right corner corresponds to the number of small areas. Stored. In general, the horizontal and vertical widths of an image are not necessarily a multiple of 8. Therefore, a half-sized block may be formed in a part of the image. In this embodiment, the remainder in the x direction is absorbed by creating blocks with a width smaller than 8 at the left and right edges of the image. The remainder in the y direction is absorbed by making a block with a vertical width smaller than 8 at the lowermost end of the image.

  The image binarizer 605 has a function of generating a binary image from the image data stored in the input image buffer 302 and storing it in the binary image buffer 606. The image binarizer 605 binarizes the original image data using a binarization threshold T described below. That is, the image binarizer 605 sets 0 (zero) as the pixel value of the corresponding binarized image for pixels (pixels in the non-background area) whose pixel value is greater than or equal to 0 and less than the threshold T in the original image data. Assign. The other pixels (background region pixels) are assigned 255 as the pixel value of the corresponding binarized image.

  As a method for obtaining the threshold value T, a method called “Otsu's method” is used here. This “Otsu's method” is described in “Image Analysis Handbook”, supervised by Takagi and Shimoda, University of Tokyo Press, pages 502-504.

  As described above, the binarized image is generated from the image data stored in the image data buffer 302 in order to facilitate the distinction between the background region and the non-background region in the subsequent processing.

  The binary image buffer 606 has a function of storing image data binarized as 1 byte per pixel.

The line extractor 607 has a function of referring to the binary image buffer 606 to extract horizontal and vertical straight lines from the non-background area and storing the information in the line information buffer 609. As a method for extracting information on a straight line as described above, a known technique such as that described in Japanese Patent No. 3100825 can be used. In particular, the line extractor 607 according to this embodiment extracts horizontal lines and vertical lines. The reason why the straight lines to be extracted are limited to the horizontal lines and the vertical lines is that, as described in the section of the prior art, special processing is performed on the pixels forming the straight lines in the vertical direction or the horizontal direction. This is because it is considered that the effect is great in preventing.

  A data structure stored in the line information buffer 609 is shown in FIG. Referring to FIG. 5, line information buffer 509 stores the upper left corner X coordinate and Y coordinate and lower right corner X coordinate and Y coordinate of the circumscribed rectangle of the straight line extracted by line extractor 607. The reason why the straight line can be approximated by a circumscribed rectangle is that the straight line extracted by the line extractor 607 is a straight line in the horizontal direction and the vertical direction.

  The character rectangle information extractor 610 refers to the binary image stored in the binary image buffer 606, extracts the rectangle information circumscribing the character element composed of the pixels belonging to the non-background area, and extracts the extracted rectangle The number has a function of storing the number in the character rectangle number register 611 and the coordinate information in the character rectangle information buffer 612, respectively. In addition, what is called a character element here is typically each connection area | region which the pixel which belongs to a non-background area | region makes. A technique for extracting a rectangle circumscribing such a character element from an image is disclosed in, for example, Japanese Patent No. 3058389.

  The data structure of the character rectangle information buffer 612 is shown in FIG. Referring to FIG. 6, the character rectangle information also includes the upper left corner X coordinate and Y coordinate and the lower right corner X coordinate and Y coordinate of the circumscribed rectangle of each character element.

  The character rectangle color reducer 614 determines the optimum gradation number of each rectangle for the character rectangle extracted by the character rectangle information extractor 610 and stored in the character rectangle information buffer 612, and uses the information as the character rectangle gradation number. It has a function of storing in the buffer 615. Details thereof will be described later.

  The small area color reducer 616 has a function of determining the optimum number of gradations of each small area at this point and storing the information in the small area gradation number buffer 617. Details of this will also be described later.

  The small area gradation number verifier 619 has a function of verifying whether the contents of the small area gradation number buffer 617 are consistent and rewriting the contents of the small area gradation number buffer 617. Details thereof will be described later.

  The small area simple color reducer 620 refers to the input image buffer 302 and the small area gradation number buffer 617, reduces the color of each pixel belonging to each small area, and outputs it to the corresponding small area of the reduced color image buffer 304. Have Details thereof will be described later.

The reference pixel value table 622 is a table in which pixel values corresponding to the number of gradations can be referred to.
Hereinafter, a control structure of processing in which the controller 601 performs color reduction processing using each component shown in FIG. 3 will be described with reference to FIG. The details of the processing contents in each block will be described later, and the overall flow of processing performed by the controller 601 is as shown in FIG. This processing is all performed under the control of the controller 601, but in the following description, the control of the controller 601 may not be mentioned.

  First, the small area information generator 602 generates information for dividing the original image stored in the input image buffer 302, and stores the number of small areas in the small area number register 603. Information about each small area including the position of each small area is stored in the small area information buffer 600 (S701).

  Next, the image binarizer 605 generates a binary image from the image data stored in the input image buffer 302 and stores it in the binary image buffer 606 (S702). Subsequently, the line extractor 607 refers to the binary image buffer 606, extracts horizontal and vertical straight lines from the non-background area, and stores the contents in the line information buffer 609 (S703).

  Subsequently, the character rectangle information extractor 610 refers to the binary image buffer 606, extracts the information of the rectangle circumscribing the character element composed of the pixels belonging to the non-background area, and determines the number of extracted rectangles as the number of character rectangles. The register 611 and the coordinate information are respectively stored in the character rectangle information buffer 612 (S704). The character rectangle color reducer 614 determines the optimum gradation number of each rectangle and stores the information in the character rectangle gradation number buffer 615 (S705). Details of this will be described later. The small area color reducer 616 determines the optimum number of gradations of each small area at this time, and stores the information in the small area gradation number buffer 617 (S705A). Details of this will be described later.

  The controller 601 rewrites the contents of the small area gradation number buffer 617 with reference to the gradation number information for each rectangle stored in the character rectangular gradation number buffer 615 (S706). Details of this will be described later. Further, the controller 601 refers to the line information stored in the line information buffer 609 and rewrites the contents of the small area gradation number buffer 617 (S707). Details of this will also be described later.

  Subsequently, the small area gradation number verifier 619 verifies whether the contents of the small area gradation number buffer 617 are consistent, and rewrites the contents of the small area gradation number buffer 617 according to the verification result (S708). ). Details thereof will be described later. Further, the small area simple color reducer 620 reduces the color of each pixel belonging to each small area while referring to the input image buffer 302 and the small area gradation number buffer 617, and outputs it to the corresponding small area of the reduced color image buffer 304. (S709). Details will be described later. When the process of S709 ends, the color reduction process ends.

  The control structure of the process performed in S705 shown in FIG. 4 will be described with reference to FIG. The controller 601 initializes the first counter 613 with 0 (S1061). The value of the counter 613 is hereinafter referred to as “i”.

  The character rectangle color reducer 614 refers to the input image buffer 302 and the character rectangle information buffer 612 to determine the optimum number of gradations for the area included in the i-th rectangle. The determined information is stored in the i-th element of the character rectangular gradation number buffer 615 (S1062). The operation of the character rectangular color reducer 614 will be described later.

  Subsequently, the controller 601 increments the value i of the first counter 613 (S1603). The contents i of the first counter 613 and the contents of the character rectangle number register 611 are compared (S1604), and if they are equal, the process of S704 is terminated. If i is not equal to the number of rectangles, the control returns to S1602.

  Next, details of the processing of S705A shown in FIG. 4 will be described with reference to FIG. The controller 601 initializes the value i of the first counter 613 with 0 (S5201). Hereinafter, the value of the first counter 613 is referred to as i.

  Subsequently, the small area color reducer 616 refers to the input image buffer 302 and the small area information buffer 604, and for the i-th small area, the subtractive color gradation such that the original image and the error after color reduction fall within a certain range. The number is determined from the number of possible subtractive color gradations. The small area color reducer 616 outputs the number of subtractive color gradations to the small area gradation number buffer 617 (S5202). The small area gradation number buffer 617 is an integer array corresponding to the small area on a one-to-one basis. The operation of the small area color reducer 616 will be described later.

  Subsequently, the controller 601 increments the value i of the first counter 613 (S5203). The controller 601 compares the value of the first counter 613 with the value of the small area number register 603 (S5204). If they are equal, the process ends. If they are not equal, the control returns to S5202.

  Details of the processing performed in S706 shown in FIG. 4 will be described with reference to FIG. First, the controller 601 initializes the value i of the first counter 613 with 0 (S5301). Subsequently, the controller 601 initializes the value of the second counter 616 with 0 (S5302). Hereinafter, the value of the second counter 616 is referred to as j.

  Subsequently, the controller 601 refers to the character rectangle information buffer 612 and the small area information buffer 604, and determines whether or not the i-th rectangle overlaps the j-th small area (S5303). If they overlap, the control proceeds to S5304, and if they do not overlap, the control proceeds to S5306. Each of the small area information and the rectangular information is uniquely determined by the upper left coordinates and the lower right coordinates. Therefore, the determination as in S5303 can be made by comparing the coordinates of the i-th rectangle and the small area.

  Further, the controller 601 compares the value of the i-th element of the character rectangular gradation number buffer 615 with the value of the j-th element of the small area gradation number buffer 617 (S5304). If the former is larger, control proceeds to S5305; otherwise, control proceeds to S5306.

  In step S5305, the value of the i-th element of the character rectangle gradation number buffer 615 is substituted for the j-th element of the small area gradation number buffer 617.

  Subsequently, in S5306, the controller 601 increments the value of the second counter 616 by 1 (S5306). Then, the value is compared with the value of the small area number register 603 (S5307). If they are equal, control proceeds to S5308, otherwise control returns to S5303.

  In step S5308, the controller 601 increments the value of the first counter 613 by one. The incremented value of the first counter 613 is compared with the value of the character rectangle number register 611 (S5309). If they are equal, this process ends. If they are not equal, control returns to S5302.

  Next, the contents of the process of S707 in FIG. 4 will be described with reference to FIG. Prior to this processing, it is assumed that the number of subtractive color gradations to be used for the line area is determined in advance and is already stored in the line area gradation number register 618.

  The value stored in the line area gradation number register 618 is not particularly limited. However, the purpose of extracting lines from the image is to assign a larger number of gradations than other regions so that deterioration such as rattling is not noticeable during color reduction. Therefore, it is desirable to take a larger value as the number of gradations after color reduction.

  In the apparatus of this embodiment, it is assumed that “8” is stored in the line area gradation number register 618. This is the same as the maximum number of gradations after color reduction stored in the second color reduction table 1103 and the color reduction table 1603.

  Referring to FIG. 10, the controller 601 initializes the value i of the first counter 613 with 0 (S5401). Subsequently, the value j of the second counter 616 is initialized with 0 (S5402).

  The controller 601 refers to the line information buffer 609 and the small area information buffer 604 and determines whether or not the i-th rectangle overlaps the j-th small area (S5403). If they overlap, the control proceeds to S5404, and if they do not overlap, the control proceeds to S5406.

  Information on each line is uniquely determined by the coordinates of the upper left corner and the lower right corner of the circumscribed rectangle. Therefore, the determination in S5403 can be easily performed by comparing the coordinates of the i-th rectangle and the j-th line.

  If the i-th line and the j-th small area overlap, in step S5404 the controller 601 determines the value of the line-area gradation number register 618 and the value of the j-th element of the small-area gradation number buffer 617. Compare (S5405). If the former is larger, control proceeds to S5405, otherwise control proceeds to S5406.

  If the value of the line area gradation number register 618 is larger than the value of the jth element of the small area gradation number buffer 617, the line area gradation number register 618 is set to the jth element of the small area gradation number buffer 617 in step S5405. The value of the area gradation number register 618 is substituted (S5405).

  Subsequently, the controller 601 increments the value j of the second counter 616 (S5406), and compares the value j of the second counter 616 with the value of the small area number register 613 (S5407). If the two are equal, control proceeds to S5408, otherwise control returns to S5403.

  In step S5408, the controller 601 increments the value i of the first counter 613. Further, in S5409, the value i of the first counter 613 is compared with the value of the line number register 608 (S5409). If they are equal, the process ends. If they are not equal, control returns to S5402.

  Next, details of the processing in S709 shown in FIG. 4 will be described with reference to FIG. The controller 601 initializes the value i of the first counter 613 with 0 (S5501).

  The small area simple color reducer 620 reduces the color of each pixel belonging to the i-th small area while referring to the input image buffer 302 and the small area gradation number buffer 617, and outputs it to the corresponding pixel in the reduced color image buffer 304. (S5502). The pixel value converter 621 refers to the reference pixel value table, refers to the value of the i th element of the small area gradation number buffer 617 corresponding to the i th small area, and stores it in the reference pixel value table 622. The value of each pixel in the small area is subtracted from the closest reference pixel value.

  Subsequently, the controller 601 increments the value i of the first counter 613 (S5503), and compares the value i of the first counter 613 with the value of the small area number register 603 (S5504). If they are equal, the process ends, otherwise control returns to S5502.

  FIG. 12 shows the configuration of the character rectangular color reducer 614 shown in FIG. Referring to FIG. 12, the character rectangular color reducer 614 includes a controller 1101, a first color reduction table 1102, a second color reduction table 1103, a possible gradation number register 1104, a line extractor 1105, and a line number register 1106. A line information buffer 1107, a first counter 1108, a second counter 1109, a cumulative histogram generator 1110, a cumulative histogram buffer 1111, and a temporary line tone number buffer 1112.

  The controller 1101 controls other components of the character rectangular color reducer 614, and controls data communication between the other components and between each component and the outside of the character rectangular color reducer 614. It is. The controller 1101 is always involved in the operation of the other components of the character rectangle color reducer 614. However, for simplification of explanation, in the following description, the involvement of the controller 1101 may not be mentioned to the extent that may be understood.

  The first color reduction table 1102 is an integer array in which the number of gradations that can be output as the output of the character rectangular color reducer 614 is stored in advance. The output of the character rectangular color reducer 614 is required to be 1 or more and less than the number of gradations of the original image. As already described, the number of gradations of the original image is 256 in the present embodiment. Therefore, the output of the character rectangular color reducer 614 is 1 or more and 256 or less. However, from the viewpoint of reducing the data capacity, it is desirable that the value be as small as possible within a range where sufficient image quality can be maintained.

  In the present embodiment, the output of the character rectangle color reducer 614 is assumed to be four types of 1, 2, 4, and 8. These values are stored in the first color reduction table 1102 in this order. That is, the first color reduction table 1102 has four elements. Hereinafter, the j-th element of the first color reduction table 1102 may be represented as S [j]. If the elements of the first color reduction table 1102 are expressed using this notation, S [0] = 1, S [1] = 2, S [2] = 4, and S [3] = 8.

  The number of gradations that can be output from the character rectangle color reducer 614 is determined in advance and stored in the possible gradation number register 1104. In this embodiment, the value stored in the possible gradation number register 1104 is 4. In the following description, in order to generalize the description, the value stored in the possible gradation number register 1104 will be described as v. This embodiment is a case where v = 4.

  The second color reduction table 1100 is a two-dimensional array. In the (j, k) element, when the color is reduced to j gradation, the pixel value of the minimum original image that is reduced to the kth (where j> k) gradation counted from 0 is stored in advance. It shall be. Hereinafter, the (j, k) element of the second color reduction table 1103 is written as T [j] [k].

  The algorithm for determining the pixel value after color reduction from the pixel value of the original image is as follows. As a premise, a reference pixel value (hereinafter referred to as “reference pixel value”) is determined as follows. First, among pixel values 0 to 255 constituting 256 gradations, 0 and 255 at both ends are set as reference pixel values, respectively. The remaining (number of gradations after subtractive gradation−2) reference pixel values are arranged between 0 and 255 so that the intervals are equal, including 0 and 255 at both ends. Each reference pixel value is numbered as 0, 1, 2,... In order from the smallest reference pixel value. Then, the distribution of reference pixel values is determined. The distributions of the reference pixel values corresponding to the number of gradations of 2, 4, and 8 are shown in FIGS. 13, 14, and 15, respectively.

  Referring to FIG. 13, in the case of two gradations, the reference pixel value is composed of 0 and 255 as can be seen from the above description. Depending on whether the pixel value of the original image is close to 0 or 255, the pixel after color reduction takes a value of 0 or 1.

  Referring to FIG. 14, in the case of 4 gradations, 85 and 170 are the reference pixel values in addition to 0 and 255. Depending on whether the pixel of the original image is closest to 0, 85, 170, or 255, the pixel after color reduction takes any value from 0 to 3.

In the case of 8 gradations, as shown in FIG. 15, in addition to 0 and 255, 36, 73, 109
146, 182, and 219 are also reference pixel values. As in the case of FIG. 13 and FIG. 14, one of 0 to 7 is assigned to the pixel value after color reduction depending on which of the reference pixel values the pixel value of the original image is closest to.

  Depending on the number of gradations after color reduction, the number of gradations of the original image, and the pixel value, pixels that have the same absolute value of the difference from the upper and lower reference pixel values (positioned exactly in the middle of the upper and lower reference pixel values) May appear). In such a case, in this embodiment, the lower reference pixel value is always adopted. Of course, any pixel value determination method in this case may be adopted as long as an agreement is made in a range in which there is no contradiction. For example, a method of always adopting the higher reference pixel value of the two reference pixel values may be employed.

  Under this algorithm, for j smaller than the number of gradations of the original image, when the color is reduced to j gradations, the minimum pixel value is such that the pixel value after color reduction is counted from 0 and becomes the kth pixel value. It must exist. However, the minimum pixel value here is not a value from an actual input image, but a calculated value when all possible pixel values are assumed as an input. For example, when the image is reduced to two gradations, a pixel having a value between 0 and 127 is determined to be close to the 0th reference pixel value of 0. Therefore, T [2] [0], that is, the (2,0) th element of the second color reduction table 1103 is 0. A pixel having a value of 128 or more and 255 or less is determined to be close to the first reference pixel value 255. Therefore, T [2] [1] in the second color reduction table 1103 is 128.

  Similarly, when considering the case of decreasing to four gradations, pixel values of 0 to 42 are set to the 0th reference pixel value 0, pixel values of 43 to 127 are set to the first reference pixel value 85, and 128 to 212. The following pixels are determined to be closest to the second reference pixel value 170, and the pixels from 213 to 255 are determined to be closest to the third reference pixel value 255, respectively. Therefore, T [4] [0] = 0, T [4] [1] = 43, T [4] [2] = 128, and T [4] [3] = 213.

  If the same calculation is performed for 8 gradations, the elements of T [4] [0] to T [4] [7] are 0, 18, 55, 91, 128, 164, 200, and 237, respectively. Of course, the numerical values given here are merely examples, and the present invention is not limited thereto.

  From the above description, the second color reduction table 1103 is a two-dimensional array, but it can be seen that there are many unnecessary elements. That is, a T [j] [k] element corresponding to j that is not an element of the first color reduction table 1102 is not necessary. Also, T [j] [k] corresponding to j and k is not necessary. In this embodiment, it is assumed that 0 is stored in the element T [j] of the second color reduction table 1103 corresponding to such an unnecessary combination of subscripts.

  With reference to FIG. 16, a control structure of processing in which the controller 1101 of the character rectangular color reducer 614 controls each component will be described. In the following description, the controller 1101 is always involved, but the controller 1101 may not be referred to for the sake of simplicity.

First, the line extractor 1105 refers to the binary image buffer 606, extracts a straight line in the vertical direction or the horizontal direction from the rectangular region currently being processed, and stores the information in the line information buffer 1107. (S1201). Note that the functions of the line extractor 607 shown in FIG. 3 and the line extractor 1105 shown in FIG. 12 are similar to each other, so that they can be shared by the same module. However, what is intended to be taken out by the line extractor 1105 is not a line in a general sense, but a horizontal bar or vertical bar in characters as shown in FIGS. 17 and 18. It is. Therefore, the parameters when the line extractor 1105 operates (such as the minimum length to be regarded as a line) are generally different from the parameters when the line extractor 607 shown in FIG. 3 operates. . In the present embodiment, the line extractor 1105 uses the method used in the line extractor 607 of FIG. 3, that is, the technique disclosed in Japanese Patent No. 3100825.

  Referring to FIG. 16 again, the controller 1101 initializes the value i of the first counter 1108 to 0 (S1202). Hereinafter, i means the value of the first counter 1108.

  The cumulative histogram generator 1110 generates a cumulative histogram of pixel values of the band-like regions on both sides of the i-th line and stores it in the cumulative histogram buffer 1111 (S1203). Here, the cumulative histogram is a total of the number of pixels less than the pixel value for each pixel value in the band-like regions on both sides of the i-th line. The cumulative histogram buffer 1111 is an integer array that has a one-to-one correspondence with possible pixel values in the original image. That is, since the original image has 256 gradations in this example, the cumulative histogram buffer 1111 is also composed of 256 elements. Hereinafter, the m-th element of the cumulative histogram buffer 1111 may be expressed as H [m].

  In short, H [m] is the number of pixels having a pixel value less than m in the processing region. Therefore, H [0] is always 0 regardless of the pixel value distribution in that region. If an image area is designated, one set of all the pixels contained therein is determined, and it is clear that such a cumulative histogram can be created. One skilled in the art can readily understand and implement the configuration and operation of the cumulative histogram generator 1110.

  The cumulative histogram will be described with reference to FIGS. Referring to FIG. 19, a band-like region 1502 is assumed on the upper side of horizontal line 1501 extracted from the input character, and a band-like region 1503 is assumed on the lower side. Here, the band-like region is a region obtained by removing the line from the region in which the line is expanded two dots up and down (two dots left and right in the case of a vertical line).

  FIG. 20 shows an example after the same input image as FIG. 19 is converted into eight values. In this example, a non-background region 1504 having a pixel value other than 7 is obtained, which is thicker than the horizontal line 1501 shown in FIG. This is a reference pixel having a pixel value such that the binarization threshold value used when extracting a straight line becomes a non-background region with 8 values, and becomes a pixel value other than 7 when the color is reduced to 8 values. The reason is that it is usually much lower than the value 237, and that at the time when a straight line is extracted from the non-background region obtained in such a manner, whiskers and the like are generally removed.

  FIG. 21 shows an example in which the same input image is binarized. In the example shown in FIG. 21, the non-background region 1505 having a pixel value other than 3 is thinner than the non-background region 1504 shown in FIG. In the belt-like areas 1502 and 1503 shown in FIG. 19, when the number of subtractive color gradations is changed from 8 (FIG. 20) to 4 (FIG. 21) in this way, it belongs to the non-background area at the time of 8-value color reduction, and the 4-value color reduction. Sometimes, when there are many pixels that become the background area, there is not much deterioration with 8 gradations of color reduction, but with 4 gradations, it is considered that the deterioration is severe.

  Referring to FIG. 16 again, the controller 1101 initializes the value j of the second counter 1109 to a value of v−1 (S1204). In the present example, j = 4-1 = 3.

  Subsequently, the controller 1101 refers to the first subtractive color table 1102, the second subtractive color table 1103, the possible gradation number register 1104, the second counter 1109, and the cumulative histogram buffer 1111 to refer to H [T [S [j]] [S The value of [j] -1]] is taken out and compared with H [T [S [v]] [S [v] -1]] (S1205). If there is no significant difference between the two, the control proceeds to S1206, otherwise the value j of the second counter 1109 is incremented (S1208), and the control proceeds to S1209.

The physical meaning of the comparison performed here will be described. T [S [v]] [S [v] -1
] Indicates the minimum value before color reduction of the pixel value that becomes the background area under the maximum gradation number S [v] after color reduction. This is because S [v] represents the maximum number of subtractive color gradations, and S [v] -1 represents the maximum pixel value after subtractive color under such subtractive color. Similarly, T [S [j]] [S [j] −1] is the minimum value before color reduction of the pixel value that becomes the background region under the jth subtractive color gradation number S [j]. Indicates.

  In general, since H [m] is the number of pixel values having a pixel value less than m, H [T [S [v]] [S [v] -1]] is reduced under the maximum number of gradations. Here, the number of pixels that do not become the background region is shown. Similarly, H [T [S [j]] [S [j] −1]] indicates the number of pixels that do not become the background area under the jth subtractive color number.

  The large difference between the two numbers means that for a certain j such that j <v, although it was not a background area in the color reduction to S [v] gradation, This means that there are many pixels that will become the background area in color reduction. An object for which such a determination is made is a band area on both sides of a straight line in the vertical and horizontal directions in the character. Therefore, this means that the color reduction to the S [j] gradation has a significant deterioration of the lines constituting the character compared to the color reduction to the S [v] gradation. That is, rattling or thinning due to color reduction is conspicuous. In such a case, in step S1205, it is determined that no further color reduction can be performed for the area, and further color reduction is stopped.

  Referring to FIG. 16 again, in step S1206, the controller 1101 decrements the value j of the second counter 1109, and compares j with 0 (S1207). If j is less than 0, control proceeds to S1208, otherwise control returns to S1205.

  When the value j of the second counter 1109 is smaller than 0, the controller 1101 increments the value j of the second counter 1109 (S1208), and further the i-th element (hereinafter, TLB [i]) of the temporary line gradation number buffer 1112. ) Is substituted for S [j] (S1209).

  Subsequently, the controller 1101 increments the value i of the first counter 1108 (S1210), and compares the value i of the first counter 1108 with the value of the line number register 1106 (S1211). If they are equal, control proceeds to S1212, otherwise control returns to S1203.

  When the value i of the first counter 1108 is equal to the value of the line number register 1106, the next process is a process for extracting the minimum element of the temporary line gradation number buffer 1112. For this purpose, the controller 1101 substitutes the value of TLB [0] for the value j of the second counter 1109 (S1212). The controller 1101 assigns 1 to the value i of the first counter 1108 (S1213). Subsequently, the controller 1101 compares j with TLB [i] (that is, S [i]) (S1214). If j is larger, control proceeds to S1215, otherwise control proceeds to S1216. If j is larger, TLB [i] is substituted for the value j of the second counter 1109 (S1215).

  Subsequently, the controller 1101 increments the value i of the first counter 1108 (S1216), and compares the value i of the first counter 1108 with the value of the line number register 1106 (S1217). If they are equal, control proceeds to S1218, otherwise control returns to S1214.

If the value i of the first counter 1108 is equal to the value of the line number register 1106, the controller 1101 outputs the value of the second counter 1109 as the output of the character rectangular color reducer 614 (
S1218), the process is terminated.

  Next, the small area color reducer 616 that performs the gradation number determination processing for the small area i performed in S5202 shown in FIG. 8 will be described. Referring to FIG. 22, the small area color reducer 616 includes a controller 1701, a possible gradation number register 1702, a color reduction table 1703, a first counter 1704, a pixel value converter 1705, and a temporary color reduction image buffer 1706. A reference pixel value table 1707.

  The controller 1701 controls the other components of the small area color reducer 616 and controls data communication between the other components or between each component and the outside of the small area color reducer 616. is there. The controller 1701 is always involved in the operation of the other components of the small area color reducer 616. However, for the sake of simplification of explanation, the operation of the controller 1701 may not be mentioned below as long as it is understandable.

  The color reduction table 1703 has the same structure and contents as the second color reduction table 1103 shown in FIG. The possible gradation number register 1702 is for storing in advance the number of gradations possible as an output of the small area color reducer 616. The number of gradations is determined in advance. In this example, the value stored in the possible gradation number register 1702 is the same value “4” as that stored in the possible gradation number register 1104 shown in FIG. In the following description, v is used as the value of the possible gradation number register 1702 in order to generalize the case where a value other than 4 is stored in the possible gradation number register 1104.

  The pixel value converter 1705 has a function of converting pixel values from 256 gradations to S [i] gradations based on the reference pixel values stored in the reference pixel value table 1707. The temporary color-reduced image buffer 1706 is for storing the image after color reduction by the pixel value converter 1705.

  The reference pixel value table 1707 is a two-dimensional array similar to the second color reduction table 1103 shown in FIG. However, the value is a reference pixel value. Hereinafter, the (ij) component of the two-dimensional table of the reference pixel value table 1707 is written as U [i] [j].

  A control structure of processing performed by the controller 1701 will be described with reference to FIG. First, the controller 1701 initializes the value i of the first counter 1704 to v−1 (S1801). Hereinafter, i means the value of the first counter 1704. Subsequently, the controller 1701 uses the pixel value converter 1705 to reduce the processing area specified by the i and small area information buffer 604 of the original image stored in the input image buffer 302 to S [i] gradation. The temporary color-reduced image buffer 1706 is stored (S1802).

  As a conversion method from 256 gradations to S [i] gradations by the pixel value converter 1705, here, the same technique as described in the explanation regarding the second color reduction table 1103 shown in FIG. 12 is used. That is, the reference pixel values are evenly arranged in the original color space, and the pixel value after color reduction is determined depending on which reference pixel value the pixel value of the original image is close to. A value stored in the reference pixel value table 1707 is used as the reference pixel value for this purpose.

  Under the condition that the original image has 256 gradations, the values in the reference pixel value table 1707 for the portion corresponding to 4 gradations are as follows. That is, U [4] [0] = 0, U [4] [1] = 85, U [4] [2] = 170, U [4] [3] = 255.

Although the reference pixel value table 1707 is a two-dimensional array, it has already been described for the second color reduction table 1003 shown in FIG. 12 that it includes many unnecessary elements and elements that can be omitted. It is the same as that.

  In preparation for later explanation, here, the values of the reference pixel value table 1707 are similarly shown for the eight gradations. That is, in this case, U [8] [0] to U [8] [7] are 0, 36, 73, 109, 146, 182, 219, and 255, respectively.

  Referring to FIG. 23 again, the controller 1701 compares the value of the pixel in the small area of the input image buffer 301 and the value of the pixel in the small area of the temporary color-reduced image buffer 1706 corresponding to the small area currently being processed. (S1803). If the difference between the two is large, it is determined that the image after color reduction is greatly deteriorated compared to the original image, and the control advances to S1805. Otherwise, control proceeds to S1804.

  The process performed in S1803 is basically a process of calculating an average value of differences between corresponding pixels and comparing it with a predetermined threshold value. However, the contents of the temporary color-reduced image buffer 1706 are reduced to S [i] gradation. For this reason, it is impossible to make a meaningful comparison with the original image having 256 gradations. Therefore, in the present embodiment, the following method is used.

  The pixel value of one pixel extracted from the temporary color-reduced image buffer 1706 is set as w. This pixel value w is converted into U [S [i]] [w]. That is, the pixel value w is converted into the reference pixel value used for color reduction. The converted value is compared with the corresponding pixel of the image stored in the temporary subtractive color buffer 1706.

  As a result of the determination performed in S1803, if the difference between the small area pixel of the input image buffer 301 and the small area pixel value of the temporary color-reduced image buffer 1706 is smaller than a predetermined threshold, the controller 1701 The value i of the 1 counter 1704 is decremented (S1804). Then, the value i of the first counter 1704 is compared with 0 (S1805). If i is smaller than 0, the control proceeds to S1806, and if not, the control returns to S1802.

  If it is determined in S1803 that the image has greatly deteriorated, and if it is determined in S1805 that the value i of the first counter 1704 is smaller than 0, the controller 1701 outputs S [i + 1] as the output of the small area color reducer 616. Then, the process is terminated (S1806).

  Referring to FIG. 24, the small region gradation number verifier 619 shown in FIG. 3 includes a controller 2001, a reference gradation number register 2002, a pixel value converter 2003, a reference pixel value table 2004, and a subtractive color image buffer. 2005, a first counter 2006, a second counter 2007, a label buffer 2008, a label changer 2009, a small area label buffer 2010, and a third counter 2011 are included.

  The controller 2001 controls the other components of the small region gradation number verifier 619 and performs data communication between the other components or between each component and the outside of the small region gradation number verifier 619. It is for performing control. The controller 2001 is always involved in the operation of the other components of the small area gradation number verifier 619. However, here, for simplification of description, the involvement of the controller 2001 may not be mentioned to the extent that it may be understood.

The subtractive color image buffer 2005 is an image buffer having the same horizontal width and vertical width as the input image buffer 302 (see FIG. 1) and having the number of gradations given by the reference gradation number register 2002. The label buffer 2008 is a one-dimensional integer array that has a one-to-one correspondence with each pixel of the subtractive color image buffer 2005. Here, each pixel of the subtractive color image buffer 2005 corresponds to each element of the label buffer 2008 in the order according to the raster scan. The i-th element of the label buffer 2008 is written as PB [i].

  The small area label buffer 2010 is a one-dimensional integer array that has a one-to-one correspondence with the small area according to the following rules. The k-th element of the small area label buffer 2010 is written as ALB [k]. Elements of the small area label buffer 2010 corresponding to the j-th small area in the horizontal direction and the i-th small area in the vertical direction are given by the following equations.

ALB [i × number of small areas in the horizontal direction + j]
Hereinafter, processing performed by the controller 2001 in order to realize the function of the small region gradation number verifier 619 will be described with reference to FIG. The controller 2001 reduces the color of each pixel of the input image buffer 302 to the number of gradations stored in the reference gradation number register 2002 using the pixel value converter 2003 and stores it in the reduced color image buffer 2005 (S2101). Here, it is assumed that the pixel value converter 2003 performs exactly the same operation as the pixel value converter 1705 shown in FIG. Further, it is assumed that the reference pixel value table 2004 referred to by the pixel value converter 2003 has exactly the same contents as the reference pixel value table 1707 shown in FIG. The number of gradations stored in the reference gradation number register 2002 is a predetermined value, and the number of gradations that provides good image quality is assumed. Here, it is assumed that “8” is stored in the reference gradation number register 2002 in the same manner as the contents stored in the line area gradation number register 618 shown in FIG.

  Subsequently, the controller 2001 initializes the label buffer 2008 (S2102). This initialization method will be described later.

  The controller 2001 further generates a value of the label buffer 2000 (S2103). The value of the label buffer 2008 is generated so that pixels having the same pixel value among the adjacent pixels in the so-called 8-connected meaning of the image stored in the subtractive color image buffer 2005 have a common value. Details of this method will be described later.

  Subsequently, the controller 2001 initializes the small area label buffer 2010 (S2104). Details of this method will also be described later.

  Further, the controller 2001 generates a value of the small area label buffer 2010 (S2105). The value of the small area label buffer 2010 is generated so that two small areas having the same pixel in the label buffer 2000 have the same value in the corresponding elements of the small area label buffer 2010. Details of this method will be described later.

  The process performed in S2102 shown in FIG. 25 will be described in detail with reference to FIG. In the label buffer initialization process, the controller 2001 initializes the value i of the first counter 2006 to 0 (S2201). Hereinafter, the value of the first counter 2006 is i.

  Subsequently, the controller 2001 initializes the value of LB [i] in the label buffer 2000 to i (S2202). Further, the controller 2001 increments the value i of the first counter 2006 (S2203), and compares the value i of the first counter 2006 with the number of pixels (= horizontal width × vertical width) of the reduced color image buffer 2005 (S2204). If they are equal, the process ends. Otherwise, the control returns to S2202.

Details of the process of generating the label buffer value performed in S2103 of FIG. 25 will be described with reference to FIG. Referring to FIG. 27, the controller 2001 initializes the value i of the first counter 2006 to 0 (S2301). Hereinafter, the value of the first counter 2006 is referred to as i. Subsequently, the controller 2001 initializes the value j of the second counter 2007 to 0 (S2302). Hereinafter, the value of the second counter 2007 is j.

  In the following description, a pixel corresponding to the coordinate (j, i) in the subtractive color image buffer 2005 (see FIG. 24) is referred to as a “target pixel”. The controller 2001 checks whether the pixel of interest and the pixel having the coordinates (j−1, i−1) at the upper left have the same pixel value (S2303). If both have the same pixel value, control proceeds to S2304, otherwise control proceeds to S2305.

  In step S2304, the label buffer 2008 (see FIG. 24) is rewritten using the label changer 2009 (see FIG. 24) (S2304). Here, rewriting of the label buffer 2008 is performed as follows. That is, the element LB [k], LB [m] of the label buffer 2008 corresponding to the pixel of interest and each of the upper left pixels, and the representative pattern label buffer 2008 having the same value as either of them until then. A common value min (LB [k], LB [m]) is substituted for all elements. Here, k and m are integers obtained by the following equations, respectively.

k = i × width of image + j
m = (i−1) × width of image + (j−1)
When there is no upper left pixel of the target pixel, the condition of S2303 is not always satisfied. That is, if j <1 or i <1, the process of S2304 is not performed.

  Next, instead of the upper left pixel of the target pixel, the same determination as in S2303 is performed in S2305 for the pixel immediately above. That is, the controller 2001 checks whether the pixel of interest and the pixel having the coordinate (j, i−1) directly above have the same pixel value (S2305). If both have the same pixel value, control proceeds to S2306, otherwise control proceeds to S2307.

  In step S <b> 2306, the label buffer 2008 is rewritten using the label changer 2009. Here, the rewriting of the label buffer 2008 is performed as follows. That is, the element LB [k] and LB [m] of the label buffer 2008 corresponding to the pixel of interest and the pixel immediately above the pixel, and the representative pattern label buffer 2008 that had the same value as either of them until then. A common value min (LB [k], LB [m]) is substituted for all the elements. However, k and m are values obtained by the following equations, respectively.

k = i × width of image + j
m = (i−1) × width of image + j
When there is no pixel directly above the target pixel, that is, when i <1, it is considered that the condition in S2305 is not always satisfied.

  Next, in S2307, the same determination as in S2303 and S2305 is performed for the upper right pixel instead of the pixel immediately above the target pixel. That is, the controller 2001 checks whether the pixel of interest and the pixel having the coordinates (j + 1, i−1) at the upper right of the pixel have the same pixel value (step S2307) and have the same pixel value. If so, control proceeds to S2308; otherwise, control proceeds to S2309.

In S2308, the label buffer 2008 is rewritten using the label re-setter 2009 (S2308). Here, rewriting of the label buffer 2008 is performed as follows. That is, the elements LB [k] and LB [m] of the label buffer 2008 corresponding to the pixel of interest and the upper right pixel respectively, and the representative pattern label buffer 2008 having the same value as either of them until then. A common value min (LB [k], LB [m]) is substituted for all elements. Here, k and m are integers determined according to the following equations, respectively.

k = i × width of image + j
m = (i−1) × width of image + (j + 1)
Again, if there is no upper right pixel of the target pixel, that is, if j = horizontal width of the image −1 or i <1, the condition of step S2307 is not always satisfied.

  Subsequently, in S2309, the same determination as that in S2307 is performed on the left adjacent pixel instead of the upper right pixel of the target pixel. In other words, the controller 2001 checks whether the pixel of interest and the pixel having the coordinates (j−1, i) adjacent to the left have the same pixel value (S2309). If both have the same pixel value, control proceeds to S2310, otherwise control proceeds to S2311.

  In S2310, the label rewriter 2009 is used to rewrite the label buffer 2008 (S2310). The rewriting of the label buffer 2008 is performed as follows. That is, the element LB [k] and LB [m] of the label buffer 2008 corresponding to the pixel of interest and each of the pixels adjacent to the left, and the representative pattern label buffer 2008 that had the same value as either of them until then. A common value min (LB [k], LB [m]) is assigned to all the elements. However, k and m are integers calculated according to the following equations, respectively.

k = i × width of image + j
m = i × width of image + (j−1)
Here, when there is no pixel on the left of the target pixel, that is, when j <1, it is assumed that the condition of S2113 does not always hold.

  After the process of S2310, in S2311, the controller 2001 increments the value j of the substitution counter 2007. The controller 2001 further compares the value j of the substitution counter 2007 with the horizontal width of the image (S2312). If they are equal, the control proceeds to S2313, otherwise the control returns to S2303.

  If the result of determination in S2312 is YES, the controller 2001 increments the value i of the first counter 2006 (S2313). Then, it is determined whether or not the value i of the first counter 2006 is equal to the vertical width of the image (S2314). If they are equal, the process is terminated; otherwise, the control returns to S2302.

  In the above processing, the connection relation is inspected only for the upper left, right above, upper right, and left adjacent for each pixel of interest. The connection relations for the right side, lower left, right below, and lower right are not inspected. However, the adjacency relationship between the pixels located in them is processed when each of the adjacent pixels becomes the pixel of interest, and therefore the adjacent pixels in the sense of 8-connection are processed by the process shown in FIG. All connectivity is inspected.

When the processing shown in FIG. 27 ends, the image stored in the subtractive color image buffer 2005 (FIG. 24) is the same in the label buffer 2008 (see FIG. 24) and the same in the adjacent pixels. Contains the result of labeling with the rule of labeling. The contents of the label buffer 2008 at this point will be referred to as a “subtractive color connection area image” hereinafter. In addition, each of the connected areas included therein is referred to as a “subtractive color connected area”.

  Next, the process for initializing the small area label buffer shown in S2104 of FIG. 25 will be described in detail with reference to FIG. First, the controller 2001 initializes the value i of the first counter 2006 shown in FIG. 24 to 0 (S2401). Hereinafter, the value of the first counter 2006 is referred to as i. Subsequently, the controller 2001 initializes the contents of the small area label buffer 210 so that ALB [i] = i (S2402).

  Subsequently, the controller 2001 increments the value i of the first counter 2006 (S2403). Then, the value i of the first counter 2006 is compared with the value of the small area number register 6003 shown in FIG. 3 (S2404). If they are equal, the process is terminated, and if they are not equal, the control returns to S2402.

  Next, details of the process of generating the value of the small area label buffer performed in S2105 of FIG. 25 will be described with reference to FIG. What is performed here is a process for determining whether or not the contents of the small area gradation number buffer 617 (FIG. 3) are consistent with the subtractive color connection area image.

  Here, “consistency” has the following meaning. Consider a case where a certain subtractive color connection region extends over a plurality of small regions. Consider a case where such a small area is reduced to a different number of gradations depending on the contents of the small area gradation number buffer 617. In such a case, the difference in the number of gradations may cause a step that was not originally present at the boundary of the small area. This has already been explained using FIG. 59 and FIG. 60 in the section of the problem to be solved by the invention. It should be noted here that this phenomenon is not caused by the original image itself, but because the image is divided into small blocks and the number of gradations in each small region is different. It has occurred.

  If the color is simply reduced according to the contents of the small area gradation number buffer 617, such a situation may occur. Therefore, in such a case, the contents of the small area gradation number buffer 617 are regarded as being inconsistent.

  Such a situation can, of course, be avoided if the number of gradations of all the small regions is made equal to the value of the reference gradation number register 2002. However, if such a method is adopted, it becomes necessary to allocate an unnecessary number of gradations to a small area with a small amount of information, for example, an area where no figure exists on the right side of FIG. So you can't do that.

  The purpose of the processing here is to make the pixels belonging to the same connected region have the same pixel value in many cases even after color reduction for each small region. Note that pixels belonging to different connected areas always have the same pixel value after color reduction for each small area, even if the image stored in the color-reduced image buffer 2005 as shown in FIG. 59 has the same pixel value. It is not intended to have For example, in FIG. 59, the arrow-shaped region 102 and the star-shaped region 103 having the same pixel value have different pixel values in the example shown in FIG.

In the following processing, it is assumed that a common label is attached to small areas sharing the same subtractive color connection area. However, here, a connected area having a pixel value of 0 or 7 in the reduced color image buffer is ignored even if it is shared. This is because, under the color reduction algorithm used in the pixel value converter 2003, the pixel values at both ends are also reduced at both ends in the color reduction from 256 gradations to a lower gradation number such as 4 gradations. This is because the pixel value shifts to the pixel value, and therefore, the pixel value at the time of restoring and displaying 256 gradations is not affected. As described above, this cannot be said for pixel values other than 0 and 7 when the color is reduced to 8 gradations.

  Referring to FIG. 29, first, the controller 2001 initializes the value i of the first counter 2006 to 0 (S2501). Hereinafter, the value of the first counter 2006 is represented by i. In step S2502, the controller 2001 initializes the value j of the second counter 2007 to 0. Hereinafter, the value of the second counter 2007 is referred to as j.

  In this loop, i represents the vertical number of the small area, and j represents the horizontal number of the small area. In the following description, X1 and Y1 are defined as follows.

X1 = (width of image + 7) / 8
Y1 = (Vertical width of image + 7) / 8
As can be easily understood, X1 and Y1 respectively represent the number of small areas in the horizontal direction and the number of small areas in the vertical direction. K is an integer given by the following equation.

k = i × X1
In the following description, the j-th small area counted from the horizontal and the i-th small area counted from the vertical are represented as a small area (j, i). The controller 2001 compares the value j of the second counter 2007 with X1-1 (S2503). If they are equal, control proceeds to S2513, otherwise control proceeds to S2504.

  In S2504, the controller 2001 checks the connection of the boundary between the small area (j, i) and the small area (j + 1, i) (Step S2504). If the two small areas share the subtractive color connection area, the control proceeds to S2505; otherwise, the control proceeds to S2506. The determination method here will be described later.

  In step S <b> 2505, the rewriting process of the small area label buffer 2010 is performed by using the label reassigner 2009. Here, rewriting is performed as follows. That is, the elements ALB [k] and ALB [k + 1] of the small area label buffer 2008 corresponding to the small area (j, i) and the small area (j + 1, i), respectively, and the same as either of them so far A common value min (ALB [k], LB [k + 1]) is assigned to all elements of the representative pattern label buffer 2008 having a value.

  Subsequently, the controller 2001 compares the value i of the first counter 2006 with Y1-1 (S2506). If they are equal, the control proceeds to S2513, otherwise the control proceeds to S2507. This is because the fact that the number i in the vertical direction of the small area is equal to the number Y1-1 of the small areas in the vertical direction means that there is no small area below that.

  In S2507, the controller 2001 inspects the boundary connection between the small area (j, i) and the small area (j + 1, i + 1) (S2507). If the two small areas share the subtractive color connection area, the control proceeds to S2508; otherwise, the control proceeds to S2509. This determination method will be described later.

  In step S <b> 2508, the rewriting process of the small area label buffer 2010 is performed using the label reordering unit 2009. The rewriting here is performed as follows. That is, the elements ALB [k] and ALB [k + X1 + 1] of the small area label buffer 2010 corresponding to the small area (j, i) and the small area (j + 1, i + 1), respectively, A common value min (ALB [k], LB [k + X1 + 1]) is substituted for all elements of the small area label buffer 2010 having the same value.

  In S2509, the controller 2001 checks the connection of the boundary between the small area (j, i) and the small area (j, i + 1), and if the two small areas share the subtractive color connection area, the control proceeds to S2508. Otherwise, control continues to S2510. This determination method will be described later.

  In S2510, the label rewriting unit 2009 is used to rewrite the label in the small area label buffer 2010. Here, common values min (ALB [k], LB [k + X1] are assigned to all elements of the small area label buffer 2010 corresponding to the small area (j, i) and the small area (j, i + 1), respectively. ).

  In step S2511, the controller 2001 increments the value j of the second counter 2007. Furthermore, the controller 2001 compares the value j of the second counter 2007 with X1 (S2512), and if both are equal, the control proceeds to S2513, otherwise the control returns to S2503.

  In step S2513, the controller 2001 increments the value i of the first counter 2006. The controller 2001 further compares the value i of the first counter 2006 with Y1 (S2514). If both are equal, the process ends. If both are not equal, the control returns to S2502.

  Details of the determination processing performed in S2504 of FIG. 29 will be described with reference to FIG. First, the controller 2001 initializes the value m of the third counter 2011 to 0 (S2601). Hereinafter, “m” means the value of the third counter 2011. In the following description, the mth pixel counted from the top in the left and right end columns of the small region (j, i) is referred to as a pixel of interest.

  The controller 2001 checks whether or not the pixel of interest is equal to 0 (S2602). If the two are equal, the control proceeds to S2607, otherwise the control proceeds to S2603. In S2603, the controller 2001 checks whether or not the pixel of interest is equal to 7. If equal, the control proceeds to S2607, otherwise the control proceeds to S2604.

  In step S2604, the controller 2001 determines whether the element of the label buffer 2008 corresponding to each of the pixel of interest and the pixel on the right side has the same value. That is, the controller 2001 determines whether the element of the label buffer 2008 corresponding to the pixel of interest and the m-th pixel counted from the top in the leftmost column of the small region (j + 1, i) have the same value. To do. If both have the same value, control proceeds to S2609, otherwise control proceeds to S2605.

  In step S2605, the controller 2001 controls the label buffer 2008 corresponding to each of the pixel of interest and the upper right pixel thereof, that is, the m−1th pixel counted from the top in the leftmost column of the small area (j + 1, i). It is determined whether the elements have the same value (S2605). If both have the same value, control proceeds to S2609, otherwise control proceeds to S2606. However, if m <1, the condition of S2605 is not satisfied.

  In step S <b> 2606, the controller 2001 controls the elements of the label buffer 2008 corresponding to the pixel of interest and the lower right pixel, that is, the m + 1st pixel counted from the top in the leftmost column of the small region (j + 1, i). Determine whether have the same value. If both have the same value, control proceeds to S2609, otherwise control proceeds to S2607. However, when m is equal to the height of the small area, the condition of S2606 is not satisfied.

  In step S2607, the controller 2001 increments the value m of the third counter 2011. Further, the controller 2001 compares the value m of the third counter 2011 with the vertical width of the small area (S2608). If they are equal, the control proceeds to S2610, otherwise the control returns to S2602.

  If the determination results in S2604, S2605, and S2606 are YES, the controller 2001 determines that the two areas share the reduced color connection area (S2609), and the process ends. If the determination result in S2608 is YES, the controller 2001 determines that the two areas do not share the reduced color connection area (S2610), and ends the process.

  The meaning of the conditions used in S2604 to S2606 will be described below. In these determinations, it is determined whether or not the elements of the label buffer 2008 corresponding to the two pixels have the same value. The fact that both elements have the same value indicates that the two pixels belong to the same subtractive color connection region. Furthermore, the fact that such two pixels belong to different small areas indicates that the small areas to which they belong share one subtractive color connection area. Therefore, if the determination result in S2604 to S2606 is YES, it can be determined that the two areas share the subtractive color connection area as in S2609.

  Next, details of the determination process performed in S2507 of FIG. 29 will be described with reference to FIG. The determination here is whether or not the small area (j, i) and the lower right small area (j + 1, i + 1) share the subtractive color connection area.

  These 2001 first check whether the target pixel is equal to 0 (S2701). If both are equal, control proceeds to S2705, otherwise control proceeds to S2702.

  In S2702, the controller 2001 checks whether or not the target pixel is equal to 7 (S2702). If they are equal, control proceeds to S2705, otherwise control proceeds to S2703.

  In step S2703, the controller 2001 labels the pixel 2901 in the lower right corner of the small area (j, i) (hereinafter referred to as “target pixel”) and the pixel in the upper left corner of the small area (j + 1, i + 1). It is determined whether the elements of the buffer 2008 have the same value. If both have the same value, control proceeds to S2704, otherwise control proceeds to S2705.

  FIG. 32 illustrates a pixel of interest 2901, a pixel 2902 in the upper left corner of the small region (j + 1, i + 1), and block boundaries 2903 and 2904, respectively. Of course, only a part of the whole is shown in FIG.

  In step S2704, the controller 2001 determines that the two areas share the reduced color connection area, and ends the process. On the other hand, in step S2705, the controller 2001 determines that the two areas do not share the reduced color connection area, and ends the process.

  The conditions used in S2703 are the same as the conditions used in S2604 to S2606, and therefore have the same meaning as those.

Details of the small area label rewriting process performed in S2510 of FIG. 29 will be described with reference to FIG. First, the controller 2001 initializes the value m of the third counter 2011 to 0 (S2801). Hereinafter, m means the value of the third counter 2011.
In the following description, the mth pixel counted from the left in the lowermost line of the small region (j, i) is referred to as a pixel of interest.

  The controller 2001 checks whether the target pixel is equal to 0 (S2802). If the two are equal, control proceeds to S2807, otherwise control proceeds to S2803. Similarly, the controller 2001 checks in step S2803 whether or not the target pixel is equal to 7 (S2803). If the two are equal, control proceeds to S2807, otherwise control proceeds to S2804.

  In step S2804, the controller 2001 determines whether the element of the label buffer 2008 corresponding to each pixel of interest and the m-th pixel from the left in the uppermost line of the small region (j, i + 1) is the pixel immediately below it. Determine whether they have the same value. If both have the same value, control proceeds to S2809, otherwise control proceeds to S2805.

  In step S <b> 2805, the controller 2001 controls the elements of the label buffer 2008 corresponding to the pixel of interest and the lower left pixel, that is, the m−1th pixel counted from the left of the uppermost line of the small region (j, i + 1). Determine whether have the same value. If they have the same value, control proceeds to S2809, otherwise control proceeds to S2806. At this time, if m <1, the condition of S2805 is not satisfied.

  In step S <b> 2806, the controller 2001 determines that the element of the label buffer 2008 corresponding to each pixel of interest and the m + 1th pixel counted from the left in the upper rightmost line of the small region (j, i + 1) is the pixel of the label buffer 2008. Determine whether they have the same value. If both have the same value, control proceeds to S2809, otherwise control proceeds to S2807. However, when m is equal to the horizontal width of the small area, the condition of S2806 is not satisfied.

  The meanings of the conditions used in S2804 to S2806 are the same as the conditions used in S2604 to S2606.

  In step S2807, the controller 2001 increments the value m of the third counter 2011. Further, the controller 2001 compares the value m of the third counter 2011 with the horizontal width of the small area (S2808). If they are equal, the control proceeds to S2810, otherwise the control returns to S2802.

  If the determination result in S2804, S2805, or S2806 is YES, the controller 2001 determines in S2809 that the two areas share the reduced color connection area, and ends the process. On the other hand, if the determination in S2808 is YES, the controller 2001 determines in S2810 that the two areas do not share the reduced color connection area, and ends the process.

  In the above description of the first embodiment, it is assumed that a grayscale image of 256 gradations is assumed as an input image, and 1 byte (8 bits) is assigned per pixel. However, the present invention is not limited to such specific values. The present embodiment can be realized for an image having other gradation numbers, and is not limited to a gray scale image but may be a color image. In the case of a color image, the color reduction described above can be performed for each color component.

  In the first embodiment described above, when an image is divided into small areas, the image is divided into blocks each having 8 pixels in the horizontal direction and 8 pixels in the vertical direction. However, it goes without saying that the example of dividing the small area is not limited to this. When a division method different from this embodiment is adopted, the structure of the small area information buffer 604 as shown in FIG. 5 naturally changes accordingly.

  As described in the first embodiment, when an image is divided into grids of a certain size to form small regions, the small region information buffer 604 is actually a size other than the vertical block size and the horizontal block size. There is no need to store it. In this case, however, it is necessary to store the standard block size. In this case, the presence or absence of the block size of the half-size at the left end and the lower end, and the vertical and horizontal sizes when present, the size of the vertical block, the size of the horizontal block, and the entire image This is because it can be calculated from the horizontal width and vertical width. In the case of grid-like division, the number of small areas in the horizontal direction and the number of small areas in the vertical direction can be defined, and it is also conceivable to store such information in the small area information buffer 604.

  The same algorithm as described above for dividing the original image into a grid pattern can cope with the case where the original image is divided into small areas of completely different shapes with only a slight modification. Is clear. Therefore, details thereof will not be described here.

  In the first embodiment described above, a method called “Otsu's method” is used as a method for obtaining the threshold T when the image binarizer 605 binarizes an image. There are various other methods for obtaining the threshold value T. These methods are described in “Image Analysis Handbook” supervised by Takagi and Shimoda, University of Tokyo Press, pages 502 to 504, so please refer to them.

  As a binary image storage method, here, a combination of 0 and 255 is used as the value of the non-background region and the background region. However, it is clear that the binary image storage method is not limited to such combinations of values. In other words, it is possible to reduce the capacity of the binary image buffer 606 by using the fact that one bit per pixel is sufficient for storing the binary image, and applying the lossless compression technique to the binary image. It is also possible to reduce the capacity of the buffer 606. In any case, this does not affect the processing result of the above-described image encoding device.

  In the apparatus according to the first embodiment described above, the straight lines extracted by the line extractor 607 are limited to horizontal lines or vertical lines. However, other lines may be extracted by the line extractor 607 at all. A method for extracting a general line element from such an image is described in the above-mentioned “Image Analysis Handbook”, pages 564 to 573, so please refer to that. Note that these methods generally target multi-value images instead of binary images. Therefore, when such a method is adopted, the line extractor 607 may be changed to refer to the image data buffer 302 instead of referring to the binary image buffer 606.

  In the first embodiment described above, the extracted straight line is limited to a horizontal line or a vertical line. Therefore, the data structure of the line information buffer 609 is also as shown in FIG. However, even when diagonal lines are allowed as straight lines, it is possible to store line information with the same data structure. In this case, however, an oblique curve is approximated by its circumscribed parallelogram. An example of the data structure in this case is shown in FIG. In the example shown in FIG. 34, the coordinates of the four vertices of the circumscribed parallelogram of the diagonal line are all stored in the line information buffer.

  There are various other ways to represent the curve. For example, a combination of (center line start point coordinates, center length information or center line end point information, and line thickness information), or (center line start point coordinates, center line length information, center line angle It may be expressed by a combination of information and straight line thickness information). In any case, necessary information having a data structure similar to that shown in FIGS. 5 and 34 can be stored.

  The method of representing a straight line described so far has been represented as so-called “vector data”. However, the way of representing a straight line is not limited to the way of representing such “vector data”. There is also a method of representing the extracted straight line as image information. As an example of the data structure of the line information buffer 609 in this case, the same structure as that of the image data buffer 302 can be used. In this case, the image included in the line area may have a pixel value of 1 and the other pixels may have a pixel value of 0. Such a representation is not as efficient as a method classified as vectorization from the viewpoint of a required storage area when it is used as final data. However, when it is necessary to use the information as intermediate information, there is an advantage that information of a line having a shape other than a curve can be easily expressed.

  In the above description, for example, it has been described that the two-dimensional array of the second color reduction table 1103 includes many unnecessary elements. In that case, it is assumed that 0 is stored in the element T [j] of the second color reduction table 1103 corresponding to the combination of unnecessary subscripts. However, of course, any value other than 0 may be used for an element corresponding to an unnecessary combination of subscripts. Also, by realizing the second color reduction table 1103 with a data structure different from a simple two-dimensional array, elements corresponding to such unnecessary combinations of subscripts can be eliminated. In that case, the required memory capacity can be reduced. Furthermore, in the first embodiment, the element T [j] [0] of the second color reduction table 1103 is always 0 regardless of j. Accordingly, these elements can be omitted in the second color reduction table 1103 from the beginning.

  Furthermore, in the apparatus of the first embodiment described above, the data structure of the line information buffer 1107 shown in FIG. 12 is the same as that of the line information buffer 609 shown in FIG. However, of course, the data structure of the line information buffer 1107 may be different from the data structure of the line information buffer 609.

  Further, in the apparatus of the first embodiment, the determination in S1205 is that the pixel that becomes the background area in the S [j−1] gradation is the threshold compared to the color reduction to the S [v] gradation. It is determined that no further color reduction is possible based on whether or not the value is greater than or equal to the value. However, the conditions for this determination are not limited to those described above. For example, the determination condition may be whether or not the absolute value of the difference between the two numbers is within a range determined by a predetermined threshold value. Alternatively, a condition that the ratio of the two numbers is within a predetermined range close to 1 may be used as the determination condition. Various other variations can be considered.

  In the above-described apparatus according to the first embodiment, it has been described that the information of each line can be individually extracted in relation to the data structure of the line information buffer 609. However, even if the line information buffer 609 does not have such a structure, as long as it can be determined whether or not each pixel is included in any line, the processing of S5401 to S5409 (FIG. 10) is performed as follows. The same purpose can be achieved by substituting this process. The method described below is effective, for example, when the line information buffer 609 is a buffer for storing image data.

  In the following description, the line information buffer 609 has the same size as that stored in the image data buffer 302, an image having a pixel value of 1 for the pixels included in the line and 0 for the other pixels. Assume that data is stored.

  Referring to FIG. 35, controller 601 first initializes value i of first counter 613 with 0 (S5601). Hereinafter, i means the value of the first counter 613. Subsequently, the controller 601 initializes the value j of the second counter 616 with 0 (S5602). Hereinafter, j means the value of the second counter 616.

  Subsequently, in S5603, the controller 601 determines whether the j-th element in the horizontal direction and the i-th element in the vertical direction are 1 in the line information buffer 609 (S5603). When the value of this pixel is 1, the process proceeds to control S5604, otherwise the process proceeds to S5606.

  In S5604, the controller 601 refers to the value i of the first counter 613, the value j of the second counter 6016, and the contents of the small region information buffer 604, and the small region corresponding to the small region including the pixel currently being processed. The element of the gradation number buffer 617 is extracted (hereinafter referred to as the k-th element) and compared with the value of the line area gradation number register 618. If the latter is larger, the control proceeds to S5605; otherwise, the control proceeds to S5606.

  In step S5605, the controller 601 substitutes the value of the line area gradation number register 618 for the kth element of the small area gradation number buffer 617. In step S5606, the controller 601 increments the value j of the second counter 616, and then compares the value j of the second counter 616 with the horizontal width of the image (S5607). If they are equal, control proceeds to S5608, otherwise control returns to S5603.

  In step S5608, the controller 601 increments the value i of the first counter 613. In step S5609, the controller 601 compares the value i of the first counter 613 with the vertical width of the image. If they are equal, the process ends. Otherwise, the control returns to S5602.

  As described above, even when the content of the line information buffer 609 is image data, it is possible to extract information regarding the line.

  In the description so far, the line extractor 607 and the character rectangle information extractor 610 refer to the binary image buffer 606 when a binary image is necessary. However, the apparatus according to the present embodiment is not limited to such an implementation method. For example, as the image binarizer 605, a device that obtains only a threshold value can be used. In this case, the line extractor 607 and the character rectangle information extractor 610 refer to the threshold obtained by the image binarizer 605 and the image data buffer 302, so that the background is obtained whenever necessary. A pixel belonging to the region and a pixel belonging to the non-background region can be identified. By performing such processing, it is obvious that processing equivalent to the processing in the first embodiment described above can be performed.

  Further, in the apparatus of the first embodiment above, when inspecting the connection relationship between the pixel of interest and the adjacent pixel, the adjacent pixel in the meaning of so-called 8-connection is inspected. However, the present invention is not limited to adjacent pixels in the sense of eight connections. For example, an adjacent pixel in the sense of 4-connection can be adopted as the definition of the adjacent pixel. In this case, it is only necessary to inspect the pixels immediately above and to the left of each pixel of interest. It is obvious that the processing in this case may be simplified from the processing described above in the sense of 8-connection.

  In the first embodiment described above, for example, in S2105 of FIG. 25, it is determined whether or not the content of the small area gradation number buffer 617 is consistent with the subtractive color connected area image. It was. In the above embodiment, when a certain subtractive color connection region extends over a plurality of small regions, such a small region may be reduced to a different number of gradations depending on the contents of the small region gradation number buffer 617. It was to avoid the situation. Of course, such a situation can be avoided if the number of gradations of all the small regions is made equal to the value of the reference gradation number register 2002. However, if such a shape is adopted, it is necessary to allocate an unnecessary number of gradations to a small area with a small amount of information, for example, an area where no figure exists on the right side of FIG. Will occur.

  As another method, when restoring to 256 gradations, the pixel value obtained when the low gradation number is restored to 256 gradations is obtained when the high gradation number is restored to 256 gradations. A method of making a subset of pixel values is also conceivable. For example, pixel values when restoring 4 gradations to 256 gradations are set to 0, 73, 182, and 255. Conversely, pixel values when restoring 8 gradations to 256 gradations are set to 0, 36, 85, 109, 146, 170, 219, 255, and the like. Whichever method is used, pixel values when restoring to 256 gradations are unevenly distributed, which is not preferable.

  As another method for avoiding such a problem, a process of adjusting each element of the small area gradation number buffer 617 is performed so that all the small areas over which one subtractive color connection area extends have the same gradation number. Can be mentioned.

  If two pixels belonging to the same subtractive color connection region are also subtracted from the number of gray levels used to create the subtractive color connection region, the pixel values after subtraction of the two pixels do not match. Sometimes.

  For example, consider a pixel having a pixel value of 200 to 236 in the original image. These pixels will all be determined to be close to the reference pixel value of 219 for color reduction to 8 tones and will be reduced to a pixel value of 6. Accordingly, in the original image, two pixels having pixel values in the range of 200 to 236 may belong to the same subtractive color connection region when subtracting to eight gradations. However, if one of the two pixels has a pixel value of 212 or less and the other has a pixel value of 213 or more, the former is determined to be close to the reference pixel value of 170 in the color reduction to four gradations. Therefore, it is determined that the pixel value after color reduction is 2 and the latter is close to the reference pixel value of 255, so the pixel value after color reduction is 3. Therefore, the pixel values of these two pixels after color reduction to four gradations do not match.

  However, such a result is due to the magnitude relationship between the pixel values originally present in the original image. That is, the result is completely different from the result shown in FIG. 60 in which a step along the boundary of a small area that does not originally exist in the original image is formed. Further, in the case where the small area is combined with the color reducer 616, a color reduction result that is largely different from the original image in a small area unit cannot occur. For this reason, it is considered that there is no possibility of causing a large visual problem due to the color reduction except for the problem of the boundaries between the small areas.

  In the condition determination of S2604 shown in FIG. 30, the information on the subtractive color connection area stored in the label buffer 2008 is used. However, in this case, the determination is made using the information in the primary color image buffer 2005 without using the label buffer 2008. However, in this case, the process of S2604 is changed as follows.

  That is, the controller 2001 has the same value for the pixel of interest in the subtractive color image buffer 2005 and the pixel on the right side thereof, that is, the leftmost column of the small region (j, i), the mth pixel counted from the top. Determine whether or not. If both have the same value, control proceeds to S2609, otherwise control proceeds to S2605. The same applies to steps S2605 and S2703 that perform similar processing.

  Even in this form, the same result as that obtained by using the label buffer 2008 can be obtained for the following reason. That is, on the subtractive color image buffer 2005, two adjacent pixels having the same pixel value means that corresponding elements of the label buffer 2008 have the same value. This is because the determination performed in S2604 actually has the same effect as the determination method in the alternative described above.

  As described above, the image color reduction apparatus according to the present embodiment determines the number of gradations after color reduction in a small area in consideration of the connection between the small areas. For this reason, it is possible to perform a color reduction process for an image in which a step due to a difference in each small area that has a large visual effect is unlikely to occur without limiting the arrangement of pixel values of the number of color gradations.

  Further, the image color reduction apparatus according to the present embodiment performs control to increase the number of color reduction gradations in the small area when there is a line area in the small area. Therefore, it is possible to perform image color reduction processing so that deterioration of the line area such as rattling is not noticeable.

  The image color reduction apparatus according to the present embodiment actually detects deterioration due to excessive color reduction around the vertical and horizontal lines of characters included in a small area, and performs color reduction with an appropriate number of gradations according to the result. To do. Since unreasonable color reduction is not performed, it is possible to perform image color reduction processing so that line portions included in characters are not particularly deteriorated.

[Second Embodiment]
Referring to FIG. 36, an image encoding device according to the second embodiment of the present invention includes an image input device 3001, an input image buffer 3002, an adaptive color reducer 3003, a reduced color image buffer 3004, and an image encoder 3005. , And an encoded image buffer 3006. The image input device 3001, the input image buffer 3002, the adaptive color reducer 3003, and the reduced color image buffer 3004 are the image input device 301, the input image buffer 3002, the adaptive color reducer 303, and the reduced color image described in the first embodiment, respectively. It has the same configuration and operation as the buffer 304. Therefore, detailed description thereof will not be repeated here.

  The image encoder 3005 has a function of encoding an image and storing it in the encoded image buffer 3006 while referring to the subtractive color image buffer 3004, the small region gradation number buffer 617 and the small region number register 603.

  As the format of the encoded image stored in the encoded image buffer 3006, the format shown in FIG. 37 is assumed. In the description of the second embodiment, the specific data structure and numerical examples are merely examples for description, and do not limit the technical scope of the present invention.

  With reference to FIG. 37, the encoded image stored in the encoded image buffer 3006 includes a horizontal pixel number 3101, a vertical pixel number 3102, a small area number (AC) 3103, and the number of gradations of each small area. Size of compressed data (SAC) 3104, compressed data size 3105 of the number of gradations of each small region, compressed data size (SASX) 3106 of each small region upper left corner X coordinate, and upper left corner X coordinate of each small region Compressed data 3007, compression data size (SASY) 3108 of each small region upper left corner Y coordinate, compressed data 3109 of each small region upper left corner Y coordinate, compressed data size (SAEX) 3110 of each small region lower right corner coordinate, Compressed data 3011 of the lower right corner X coordinate of the small area, compressed data size (SAEY) 3112 of the lower right corner Y coordinate of each small area, compressed data 31 of the lower right corner Y coordinate of each small area 3. Number of bit planes (NP) 3114, compressed data size SPS [0] 3115 of the 0th bit plane, compressed data 3116 of the 0th bit plane,..., Compression of the (NP-1) th bit plane The data size SPS [NP-1] 3117 and (NP-1) th bit plane compressed data 3118 are included.

  Of the data shown in FIG. 37, data 3101 to 3104, 3106, 3108, 3110, 3112, 3114, 3115, and 3117 are all 4-byte integers.

  38, an image encoder 3005 includes a controller 3201, an integer array lossless compressor 3102, a temporary integer array buffer 3103, a bitplane divider 3204, a bitplane buffer 3205, a bitplane division rule table 3206, and an image lossless. A compressor 3207, a bit plane number register 3208, a first counter 3209, and a temporary small area information buffer 3210 are included.

  The controller 3201 controls the other components of the image encoder 3005 and controls data communication between the other components or between each component and the outside of the image encoder 3005. Is. The controller 3201 is always involved in the operation of each other component in the image encoder 3005. However, in order to simplify the description, the involvement of the controller 3201 may not be referred to in the following description to the extent that it may be understood.

  The controller 3201 realizes the above-described image encoder function by performing processing having a control structure as shown in FIG. That is, the controller 3201 first extracts the horizontal width of the image from the input image buffer 3001 and writes it in the image horizontal width field 3101 of the encoded image buffer 3006 (S3301). Further, the controller 3201 extracts the vertical length of the image from the input image buffer 3001, and writes it in the image vertical width field 3102 of the encoded image buffer 3006 (S3302). Further, the controller 3201 writes the contents of the small area register 603 in the small area number field 3103 of the encoded image buffer 3006 (S3303).

  In the following description, the value indicated by the element of the small area gradation number buffer 617 corresponding to the small area is referred to as the small area gradation number. The i-th element ALB [i] of the small area gradation number buffer 617 is expressed.

  The controller 3201 determines whether or not all the elements ALB [] of the small area gradation number buffer 617 are equal (S3304). If they are all equal, control proceeds to S3305, otherwise control proceeds to S3307.

  In step S <b> 3305, the controller 3201 writes small area information in which the entire image is one small area as information on the 0th small area in the small area information buffer 604. In step S <b> 3306, the controller 3201 sets 1 in the small area number register 603.

  The meaning of the processing in S3304 to S3306 is as follows. When the number of gradations in the small area is the same, there is no point in dividing the image into small areas. Therefore, in such a case, the processing can be speeded up by making the entire image one small region. At this time, there is an advantage that exception processing is not required at the time of decoding by expressing it on the encoded data as a special case in which the number of small areas is 1 without using a special flag.

  In S3307, the integer array reversible compressor 3102 reversibly compresses the number of elements indicated by the value of the small area number register 603 from the beginning of the small area gradation number buffer 617, and the encoded image buffer 3006 along with the length of the compressed data. Are stored in the fields 3005 and 3004.

  Here, the LZ77 method is assumed as a compression method when the integer array reversible compressor 3102 compresses the integer array. The LZ77 system is described in detail, for example, in Uematsu, “Introduction to Document Data Compression Algorithm” (CQ Publishing Co.), pages 131-141. Since the LZ77 system is a known system, detailed description thereof will not be repeated here. Of course, the reversible compression method employed here is not limited to LZ77.

  In step S <b> 3308, the controller 3201 reads the element corresponding to the upper right corner X coordinate of each small area from the top in the small area information buffer 604 by the number indicated by the small area number register 603, and stores it in the temporary integer array buffer 3103. Subsequently, the integer array reversible compressor 3102 reversibly compresses the contents of the temporary integer array buffer 3103, and stores them in the fields 3107 and 3106 of the encoded image buffer 3006 together with the length of the compressed data (S3309).

  Further, the controller 3201 reads the element corresponding to the upper right corner Y coordinate of each small area in the small area information buffer 604 by the number indicated by the small area number register 603 from the top, and stores it in the temporary integer array buffer 3103 (S3310). In step S3311, the integer array lossless compressor 3102 performs lossless compression on the contents of the temporary integer array buffer 3103, and stores the contents in the fields 3109 and 3108 of the encoded image buffer 3006 together with the length of the compressed data.

  In step S <b> 3312, the controller 3201 reads the element corresponding to each small region upper left corner X coordinate of the small region information buffer 604 from the top by the number indicated by the small region number register 603, and stores it in the temporary integer array buffer 3103. In S3313, the integer array lossless compressor 3102 performs lossless compression on the contents of the temporary integer array buffer 3103, and stores the contents in the fields 3111 and 3110 of the encoded image buffer 3006 together with the length of the compressed data.

  In step S <b> 3314, the controller 3201 reads the element corresponding to the lower left corner Y coordinate of each small area in the small area information buffer 604 from the top by the number indicated by the small area number register 603, and stores it in the temporary integer array buffer 3103. In step S3315, the integer array lossless compressor 3102 performs lossless compression on the contents of the temporary integer array buffer 3103, and stores the contents in the fields 3113 and 3112 of the encoded image buffer 3006 together with the length of the compressed data.

  In step S <b> 3316, the bit plane divider 3204 divides the image stored in the reduced-color image buffer 3004 into bit planes for each small area, and stores the bit plane in the bit plane buffer 3205. In step S3317, the controller 3201 writes the number of bit planes to be encoded in the bit plane number field 3113 of the encoded image buffer 3006. In this example, 3 is written in field 3113. Thereafter, the image reversible compressor 3207 compresses the contents of the bit plane buffer 3205 for each plane (S3318), and the process ends.

  Details of the processing of S3304 in FIG. 39 will be described with reference to FIG. First, in step S3401, the controller 3201 initializes the first counter 3209 to 1. In the following description, i means the value of the first counter 3209. In subsequent S3402, the controller 3201 compares ALB [0] with ALB [i] (S3402). If they are equal, control proceeds to S3403, and if not equal, control proceeds to S3406.

  In S3403, the controller 3201 increments the value i of the first counter 3209. Subsequently, in S3404, the controller 3201 compares the value i of the first counter 3209 with the contents of the small area number register 603 (S3404). If they are equal, control proceeds to S3405, otherwise control returns to S3402.

  In step S3406, the controller 3201 determines that all the elements of the small area gradation number buffer 617 do not match, and ends the process. On the other hand, in S3405, the controller 3201 determines that all the elements of the small area gradation number buffer 617 are equal, and ends the processing.

  Details of the processing of S3316 shown in FIG. 39 will be described with reference to FIG. In S3501, the controller 3201 initializes the first counter 3209 to 0 (S3501).

  In S3502, the bit plane divider 3204 reads the position information of the i-th small area from the small area information buffer 604 and copies it to the temporary small area information buffer 3210. Here, the temporary small area information buffer 3210 is a set of four integers that can store the X and Y coordinates of the upper right corner and the X and Y coordinates of the upper right corner of the small area. In step S3503, the bit plane divider 3204 refers to the bit plane division rule table 3206 and the temporary small area information buffer 3210, divides the i-th small area into bit planes, and writes them in the corresponding areas of the bit plane buffer 3205. Include. The rule of division into bit planes performed here is given by the bit plane division rule table 3206. How to determine the value will be described later.

  In step S3504, the controller 3201 increments the value i of the first counter 3209. In step S <b> 3505, the controller 3201 compares the value i of the first counter 3209 with the value of the small area number register 603. If they are equal, the process ends. Otherwise, control returns to S3502.

  Details of the processing of S3318 in FIG. 39 will be described with reference to FIG. The controller 3201 first initializes the value i of the first counter 3209 to 0 (S5001). Hereinafter, i means the value of the first counter 3209.

  Subsequently, the image lossless compressor 3207 losslessly compresses the data of the i-th plane of the bit plane buffer 3205, and writes it in the encoded image buffer 3006 together with the length of the compressed data (S5002). Here, JBIG (Joint Bi-level Image Group) is assumed as a technique for lossless compression of a binary image. Therefore, the output of the image reversible compressor 3207 is JBIG data. JBIG is detailed in, for example, Journal of the Institute of Image Electronics Engineers of Japan, Vol. 20, No. 1. The JBIG method is already widely used. Therefore, detailed description thereof will not be given here. Here, it does not matter what method is used as the binary image compression method, and a method other than JBIG may be used.

  In step S5003, the controller 3201 increments the value i of the first counter 3209. Further, the controller 3201 compares the value i of the first counter 3209 with the bit plane number register 3208 (S5004). If they are equal, the process ends. Otherwise, the control returns to S5002.

  A bit plane buffer 3205 shown in FIG. 38 is a collection of three planes of buffers having the same data structure and size as the output image buffer 3006. As will be described later, the data of each plane of the bit plane buffer 3205 is a binary image. Therefore, as usual, data may be stored more efficiently by storing 1 bit per pixel. It is assumed that the planes of the bit plane buffer 3205 are numbered 0, 1, 2 from the most significant plane.

  The bit plane buffer 3205 is composed of 3 planes because the maximum number of small area gradations indicated by the small area gradation number buffer 617 is 8, and an 8-gradation image is represented by 3 planes. Therefore, depending on the contents of the small area gradation number buffer 617, it is necessary to change the number of planes. It is assumed that information on the maximum number of planes is stored in the bit plane number register 3208 in advance.

  The bit plane division rule table 3206 shown in FIG. 38 is a two-dimensional integer array. The indexes are 0 to 8 for small area gradation and 0 to 7 for pixel values. According to the bit plane division rule table 3206, the pixel value is subjected to different conversion depending on the number of gradations of the small area, and then converted into an appropriate bit plane according to the value of each digit of the binary representation.

  The reason why the range of the small area gradation number, which is one of the indexes in the bit plane division rule table 3206, is set to 0 to 8 is that the maximum gradation number is assumed to be 8. The reason why the range of pixel values is set to 0 to 7 is that the maximum pixel value becomes the maximum number of gradations -1. The reason why the plane number range is 0 to 2 is that the number of planes is three. Such a range should be changed as appropriate if the parameters listed here are changed.

  Note that the data structure of the bit-plane division rule table 3206 described here is a structure that is somewhat redundant in order to make the explanation easy to understand. For example, elements other than 2, 4, and 8 as the number of small area gradations do not require corresponding elements. Further, there is no combination that satisfies the small area gradation number ≦ pixel value. Although it is naturally possible to consider a data structure with higher efficiency, a data structure including a certain degree of redundancy as described above is used here for easy understanding. For elements of the unnecessary bit-plane division rule table 3206, 0 is stored unless otherwise specified. Actually, no matter what values are set, these values are not used, so the processing result is not affected.

  Hereinafter, how to determine the value of each element of the bit plane division rule table 3206 will be described. As shown in FIG. 43, the simplest method is a method in which pixel values are represented in binary and 0/1 of each digit is directly associated with each plane. The structure of the bit plane division rule table 3206 corresponding to this method is as follows. That is, when the first column shown in FIG. 43 is a pixel value that is one of the indexes, the bit plane division rule table 3206 is shown in the second column of FIG. 43 regardless of the value of the other index. The structure is such that the corresponding value can be read. For example, regardless of the number of small area gradations, the element of the bit plane division rule table 3206 corresponding to each of the plane numbers 0, 1, and 2 corresponding to the pixel value 5 is 5, and the element corresponding to the pixel value 6 is 6 as well. Become.

  The third and subsequent columns in FIG. 43 are shown for the purpose of explaining the values after development into the bit plane buffer 3205. Therefore, these values do not actually appear in the bit plane division rule table 3206. The same applies to FIGS. 44 and 45 described later.

  The second embodiment of the present invention operates even when such a method of determining the value of the bit plane division rule table 3206 is used. However, here, the contents of the bit plane division rule table 3206 are determined by the following rules. That is, the element having the small area gradation number of 8 is given in FIG. 43, the element having the small area gradation number of 4 is given in FIG. 44, and the element having the small area gradation number of 2 is given in FIG. Use what you can.

  This definition is especially for the purpose of efficient coding of text images or similar images. In the case of a text image, the background area often occupies a large area. Among them, the small area that can be expressed in binary is considered to be an image in which the background area occupies a wide range. That is, even if the number of small area gradations is different, in most small areas, the small area with the small area gradation number 2 is 1, the small area with the small gradation number 4 is 3, and the small area gradation number is It is considered that at least the text image has 7 in the area 8 and the area where the pixel has the largest value, that is, the background area occupies most of the area.

  However, when such an image composed of small areas having different small area gradation numbers is developed on a bit plane by the method shown in FIG. 43 regardless of the small area gradation numbers, each bit plane can be obtained even in the same background area. May have different values. For example, the background area is a small area with 8 small area gradations, all bit planes are 1, a small area with 4 small area gradations is 0 in the first bit plane, and 1 in other bit planes. In a small area with 2 gradations, the third bit plane is 1 and the other bit planes are 0.

  Therefore, each bit plane is a mixture of 0 and 1 for each small area having a different number of small gradations. This is not a problem that occurs at the time of display as described in the first embodiment. However, as in the second embodiment, a technique for adaptively changing the number of gradations for each small area is image compression. When combined with technology, it appears as a problem of reducing compression efficiency.

  In order to improve such a situation, in order to restore the pixel value, not only the pixel value after color reduction but also the fact that the number of small area gradations of each small area can be used increases the compression efficiency. For this purpose, as described above, the development rule for the bit plane is changed according to the number of small area gradations. Therefore, bit plane division rules are separately provided for the different gradation levels of each small area, and the bit plane after color reduction is used to improve the compression ratio of the bit plane using the gradation number information. The continuity of values can be increased. This is one of the technically important points in the apparatus of the present embodiment.

  This will be described with reference to a relatively simple example. FIG. 46 shows an input image before color reduction. This image is divided into 3901 to 3906 small areas. A character image 3907 exists across the small areas 3901, 3902, 3904, and 3905.

  In this example, it is assumed that the gradation numbers of the small areas 3901, 3902, 3904, and 3905 are determined to be eight values after the color reduction, and the gradation numbers of the small areas 3903 and 3906 are determined to be binary. This is likely due to the presence of the character 3907. In the background area, the pixel value is 7 in the 8-level area, and the pixel value is 1 in the 2-level area, and the character 3907 has a pixel value of 0.

  In the example shown in FIG. 46, when the division method shown in FIG. 43 is used, binary values are expanded to 0 and 1 as bit planes. Therefore, the most significant plane (0th plane) of the binary area is always 0. This is shown in FIG. For this reason, in the uppermost plane, the value differs in the background region depending on the small region, which is disadvantageous for entropy compression in which compression is performed by using a bias in the distribution of values.

  In such a case, it is conceivable to convert the value in the binary region by the method shown in FIG. In this case, the pixel value is expanded as 0 or 7 in the bit plane. Therefore, the uppermost plane (0th plane) in the binary area is 1 in the background area even in the binary area. This is shown in FIG. For this reason, even if different numbers of gradations are mixed in each small area, the compression efficiency is rarely lowered by that.

  As an example different from the bit plane division rule table 3206 defined in FIG. 36 and FIG. 37, the small area gradation number is 4 as shown in FIG. It is also conceivable to use those shown in FIG. This is particularly effective for an image with a small number of small areas where the number of small area gradations is eight. This is because when the bit plane division rule table 3206 shown in FIGS. 49 and 50 is used, most of the small area images are represented by the upper two planes.

  On the other hand, if the bit plane division rule table 3206 shown in FIGS. 44 and 45 is adopted, there is an advantage. In other words, the subtracted color image can have a function of knowing the outline of the image only by decoding the uppermost plane while giving different gradation numbers for each small area. This is because in this case, the pixel value corresponding to the background image contains 1 in the uppermost plane regardless of the number of small area gradations.

  In the present embodiment, the multi-valued image is divided into bit planes and compressed as a plurality of binary images. However, the multi-valued image may be compressed as it is. Further, the small area gradation number buffer 617 may be compressed by dividing it into bit planes as an image with the number of small areas in the horizontal direction as the horizontal width and the number of small areas in the vertical direction as the vertical width.

  The image compression apparatus described in the second embodiment above is configured to increase the compression effect of the bit plane after color reduction in consideration of small areas having different numbers of gradations. Therefore, more efficient image compression can be performed.

[Third Embodiment]
The apparatus according to the third embodiment of the present invention described below is an image decoding apparatus corresponding to the image encoding apparatus according to the second embodiment. As in the description so far, the specific data structures or numerical examples used in the following description are merely examples for description, and are not of a nature that limits the technical scope of the present invention. .

  Referring to FIG. 51, the image decoding apparatus according to the third embodiment includes an encoded image buffer 4201, an image decoder 4202, an output image buffer 4203, and an image output apparatus 4204.

  52, an image decoder 4202 includes a controller 4301, an integer array decoder 4302, a temporary integer array buffer 4303, a compressed image decoder 4304, a bit plane buffer 4305, a bit plane integrator 4306, and a bit plane integration rule table. 4307, small region information buffer 4308, small region gradation number buffer 4309, small region number register 4310, bit plane number register 4311, image horizontal width register 4312, image vertical width register 4313, first counter 4314, subtractive color image buffer 4315, display An image creator 4316 and a reference pixel value table 4317 are included. In the following description, the i-th element of the small area gradation number buffer 4309 is expressed as ALB [i].

  The encoded image buffer 4201 has the same data structure as that of the encoded image buffer 3006 constituting the image encoding device according to the second embodiment.

  The controller 4301 is for controlling other components of the image decoder 4202 and controlling data communication between the other components or between each component and the outside of the image decoder 4202. . The controller 4301 is always involved in the operation of the other components in the image decoder 4202. However, for simplification of explanation, the following description may not refer to the operation of the controller 4301 unless it is understandable.

A control structure of processing for realizing the function as the image decoder 4202 by the controller 4301 controlling each component shown in FIG. 52 will be described with reference to FIG. First, the controller 4301 copies the contents of the image width field 3101 of the encoded image buffer 4201 to the image width register 4312 (S4501). Subsequently, the controller 4301 copies the contents of the image vertical width field 3102 of the encoded image buffer 4201 to the image vertical width register 4313 (S4502). Further, the controller 4301 copies the contents of the small area number field 3103 of the encoded image buffer 4201 to the small area number register 4312 (S4503).

  The integer array decoder 4302 reads encoded data from the field 3105 by the number of bytes indicated by the value of the field 3204 of the encoded image buffer 4201. Further, the integer array decoder 4302 decodes the encoded data read in this way and writes it into the small area gradation number buffer 4309 (S4504). At this time, the integer array decoder 4302 performs a decoding operation for encoding by the integer array lossless compressor 3102 of the apparatus of the second embodiment. Here, it is assumed that LZ77 is assumed as the compression algorithm in accordance with the example used in the second embodiment.

  As described herein, the decoding operation algorithm employed by the integer array decoder 4302 corresponds to the encoding algorithm used in the integer array lossless compressor 3102 of the apparatus according to the second embodiment. Therefore, if the encoding algorithm changes, the decoding operation employed by the integer array decoder 4302 changes accordingly.

  The integer array decoder 4302 reads encoded data from the field 3107 by the number of bytes indicated by the value of the field 3106 of the encoded image buffer 4201. Further, the integer array decoder 4302 decodes the read encoded data, and writes it into the temporary integer array buffer 4303 (S4505).

  The controller 4301 writes the contents of the temporary integer array buffer 4303 to the portion corresponding to the upper left X coordinate of each small area in the small area information buffer 4308 (S4506).

  The integer array decoder 4302 reads encoded data from the field 3109 by the number of bytes indicated by the value of the field 3108 of the encoded image buffer 4201. The integer array decoder 4302 decodes this encoded data and writes it into the temporary integer array buffer 4303 (S4507).

  The controller 4301 writes the contents of the temporary integer array buffer 4303 into the portion corresponding to the upper left Y coordinate of each small area in the small area information buffer 4308 (S4508).

  The integer array decoder 4302 reads encoded data from the field 3111 by the number of bytes indicated by the value of the field 3110 of the encoded image buffer 4201. The integer array decoder 4302 decodes this encoded data and writes it into the temporary integer array buffer 4303 (S4509).

  The controller 4301 writes the contents of the temporary integer array buffer 4303 into the portion corresponding to the lower right X coordinate of each small area of the small area information buffer 4308 (S4510). The integer array decoder 4302 reads encoded data from the field 3113 by the number of bytes indicated by the value of the field 3112 of the encoded image buffer 4201. The integer array decoder 4302 decodes the read encoded data, and writes it into the temporary integer array buffer 4303 (S4511).

  The controller 4301 writes the contents of the temporary integer array buffer 4303 to the portion corresponding to the lower right Y coordinate of each small area in the small area information buffer 4308 (S4512). The controller 4301 copies the contents of the field 3114 of the encoded image buffer 4201 to the bit plane number register 4311 (S4513).

The compressed image decoder 4304 decodes the image in the encoded image buffer 4201 and writes it in the bit plane buffer 4305 (S4514). Note that the algorithm of the decoding operation employed by the compressed image decoder 4304 corresponds to the encoding algorithm used in the image lossless compressor 3207 of the apparatus of the second embodiment. Therefore, if the encoding algorithm changes, the decoding operation employed by the compressed image decoder 4304 also changes accordingly.

  The bit plane integrator 4306 integrates the images of the bit plane buffer 4305, converts them into an 8-level image, and stores them in the reduced color image buffer 3521 (S4515).

  The display image creator 4316 refers to the small area gradation number buffer 617 and the reference pixel value table 4317, and converts the image of the subtractive color image buffer 4315 having a different gradation number for each small area into an image of 256 gradations. The image is converted and output to the output image buffer 4203 (S4516).

  Details of the processing performed in S4514 shown in FIG. 53 will be described with reference to FIG. The controller 4301 initializes the first counter 4314 to 0 (S4601). Hereinafter, i means the value of the first counter 4314.

  The compressed image decoder 4304 reads and decodes the encoded data by the length indicated by the field 3115 (or 3117, 3119,...) Of the encoded image buffer 4201 and writes it into the i-th plane of the bit plane buffer 4305 ( S4602).

  The controller 4301 increments the value i of the first counter 4314 (S4603). Subsequently, the controller 4301 compares the value i of the first counter 4314 with the bit plane number register 4311 (S4604), and if both are equal, the process ends. Otherwise, control returns to S4602.

  Details of the bit plane integration processing shown in S4515 of FIG. 53 will be described with reference to FIG. The controller 4301 initializes the value of the first counter 4314 to 0 (S4605). In the following description, i means the value of the first counter 4314.

  The bit plane integrator 4306 refers to the first counter 4314, the bit plane integration rule table 4307, the small area information buffer 4308, and ALB [i], and integrates the contents of the bit plane buffer 4305 corresponding to the i th small area. And stored in the corresponding small area on the reduced color image buffer 4315 (S4606). The contents of the bit plane integration rule table 4307 in this case will be described later. The bit plane integration rule table 4307 gives rules for integration of these bit planes.

  In step S5102, the bit plane integrator 4306 converts each pixel of the bit plane into a multi-valued image as a binary representation of the pixel value, and then refers to the pixel value and the bit plane integration rule table 3507 to output pixel values. Convert to The contents of the bit plane integration rule table 4307 will be described later.

  The controller 3501 increments the value i of the first counter 4314 (S4607). Subsequently, the controller 3501 compares the value i of the first counter 4314 with the contents of the small area number register 3510 (S4608). If they are equal, the process ends. If they are not equal, control returns to S4607.

  The contents of the bit plane integration rule table 4307 shown in FIG. 52 are as follows. The bit plane integration rule table 4307 is a two-dimensional integer array. The indexes are 0 to 8 for small area gradation and 0 to 7 for pixel values. According to the bit plane integration rule table 4307, the pixel value undergoes different conversion depending on the number of gradations of the small area.

  The bit plane integration rule table 4307 used here is used in the second embodiment of the invention and is obtained by reversing the aspect of the bit plane division rule table 3206 described with reference to FIGS.

  The cases where the number of small area gradations is 8, 4, and 2 are shown in FIGS. 56, 57, and 58, respectively. The first to third columns in FIGS. 56 to 58 are shown for the purpose of explaining the correspondence between the values of the bit plane buffer 4305 and the pixel values, and form part of the bit plane integration rule table 4307. It is not a thing. In FIGS. 57 and 58, the converted pixel value enclosed in parentheses corresponds to an input pixel value that is not used.

  As described above, the image decoding apparatus according to the third embodiment is configured to skip the decoding of a part of information when the number of gradations of the small areas is the same. Therefore, it is possible to perform image compression with higher encoding efficiency than in the past.

The configurations of the invention disclosed in the embodiments described above are listed below.
(1) An image color reduction device for reducing the number of gradations of input digital image data and outputting it,
Dividing means for dividing the input image into small regions by a predetermined method;
An image color reduction device including, for each small area, adaptive color reduction means for setting the number of gradations for appropriately reducing the color of the small area based on image data included in the small area.

(2) The adaptive color reduction means includes:
Extraction means for extracting a predetermined shape from the input image;
Based on the extraction result of the predetermined shape by the extracting means, the number of gradations of the region where the predetermined shape exists in the input image is the floor of the region where the predetermined shape does not exist. And an image color reduction apparatus according to (1), including means for setting the number of gradations of each of the small regions so as to be equal to or greater than the key number.

(3) The image color reduction device according to (2), wherein the predetermined shape is a straight line.
(4) The adaptive color reduction means includes:
Distribution calculating means for calculating an index representing the distribution of either background pixels or non-background pixels in at least a part of the image under different color reduction methods;
The image color reduction apparatus according to (1), further including suitability determination means for determining suitability of a color reduction method based on a comparison result of distributions calculated by the distribution calculation means.

(5) The distribution calculation means includes:
Extraction means for extracting a predetermined shape from the input image;
Pixel distribution detection means for detecting a distribution of either background pixels or non-background pixels in the vicinity of the predetermined shape under different color reduction methods;
The suitability determining means includes
Comparing the distribution of background pixels and non-background pixels under the first subtractive color gradation number and the second subtractive color gradation number detected by the pixel distribution detecting means, and the comparison result (4) The image color reduction apparatus according to (4), further including means for determining whether the second number of color reduction gradations is appropriate based on whether or not satisfies a predetermined condition.

    (6) The predetermined condition is that a difference between the number of background pixels under the second number of subtractive color gradations and the number of background pixels under the first number of subtractive color gradations is a certain value or more. The image color reduction device according to (5), wherein

(7) The predetermined condition is a difference between the number of non-background pixels under the first number of subtractive color gradations and the number of non-background pixels under the second number of subtractive color gradations. The image color reduction device according to (5), wherein the image color reduction device is a certain value or more.

(8) The adaptive color reduction means includes:
Labeling means for labeling the input image;
A small area labeling means for giving a common label to a small area including a common label based on the output of the labeling means,
The image color reduction device according to (1), wherein a common number of gradations is set in a small region to which a common label is assigned by the small region labeling unit.

(9) The adaptive color reduction means includes:
A small area labeling unit that gives a common label to two small areas to which the two pixels belong when two adjacent pixels across the boundary of the small area of the input image have a common pixel value. ,
The image color reduction device according to (1), wherein a common number of gradations is set in a small region to which a common label is assigned by the small region labeling unit.

(10) The image color reduction device according to any one of (1) to (9), for reducing the number of gradations of input digital image data;
And an image encoding means for entropy encoding the image reduced in color by the image reduction means.

(11) An image encoding device that divides an input image that can have different gradation numbers for each small area into bit planes and compresses the input image.
Image conversion means for performing a different conversion on each pixel value according to the number of gradations of the small area to be encoded and dividing it into bit planes so that continuity between adjacent pixels is increased in the upper plane;
An image encoding device comprising: image encoding means for performing predetermined entropy encoding for each bit plane obtained by conversion by the image conversion means.

(12) An image encoding device that divides an input image that may have different gradation numbers for each small area into bit planes and compresses the input image,
A determination means for determining whether or not the numbers of gradations of all the small areas are equal;
In response to determining that the number of gradations of all the small regions is equal by the determination means, the entire input image is regarded as one small region, and the image is encoded with the number of small regions being 1. First image encoding means;
A second image encoding means for encoding each small area of the input image in response to determining that the gradation numbers of all the small areas are not equal by the determining means; An image encoding device.

(13) Input means for inputting image data encoded by the image encoding device according to (10);
An image decoding apparatus comprising: decoding means for performing decoding corresponding to the entropy encoding described in (10) above with respect to the image data input by the input means.

(14) Input means for inputting image data encoded by the image encoding device according to (11);
Decoding means for performing decoding corresponding to the entropy coding described in (11) for the image data input by the input means;
An image decoding apparatus comprising: an image reverse conversion unit configured to perform reverse conversion corresponding to the image conversion unit according to (11) with respect to the image data decoded by the decoding unit.

(15) An image decoding device for decoding image data encoded by the image encoding device according to (12),
Determination means for determining the number of small regions included in the image data encoded by the image encoding means;
An image decoding apparatus comprising: decoding means for dividing and decoding the image data into the number of small regions determined by the determination means.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is shown not by the above description of the embodiment but by the scope of the claims, and is intended to include all modifications within the meaning and scope equivalent to the scope of the claims.

1 is a diagram illustrating a schematic configuration of an image color reduction device according to a first embodiment. FIG. 3 is a diagram illustrating a data structure of image data stored in an input image buffer 302 of the image color reduction device according to the first embodiment. FIG. 3 is a diagram illustrating a configuration of an adaptive color reducer of the image color reduction device according to the first embodiment. It is a flowchart which shows the control structure of the process performed with an image color reduction apparatus. FIG. 6 is a diagram illustrating a configuration example of a small area information buffer 604 of the image color reduction device according to the first embodiment. 6 is a diagram illustrating a data structure of a character rectangle information buffer 612 of the image color reduction device according to the first embodiment. FIG. FIG. 5 is a diagram showing a control structure of processing performed in S705 shown in FIG. 4. FIG. 5 is a diagram showing a control structure of processing performed in S705A shown in FIG. It is a figure which shows the control structure of the process performed in S706 shown in FIG. It is a figure which shows the control structure of the process performed in S707 of FIG. It is a figure which shows the control structure of the process performed in S709 shown in FIG. FIG. 3 is a diagram illustrating a configuration of a character rectangular color reducer 614 of the adaptive color reducer of the image color reducing apparatus according to the first embodiment. It is a figure which shows distribution of the reference pixel value in which the number of gradations corresponds to 2 values. It is a figure which shows distribution of the reference pixel value corresponding to the number of gradations with 4 values. It is a figure which shows distribution of the reference | standard pixel value corresponding to the number of gradations to 8 values. It is a figure which shows the control structure of the process which the controller 1101 of the character rectangle color reducer 614 controls each component. It is a figure which shows the vertical line of the character with which the line extractor 1105 intends to take out. It is a figure which shows the horizontal line of the character with which the line extractor 1105 intends to take out. It is a figure which shows a horizontal line and the strip | belt-shaped area | region of the upper and lower sides for demonstrating the process which calculates | requires a cumulative histogram. FIG. 20 is a diagram showing an example after the input image shown in FIG. 19 is converted into eight values. FIG. 20 is a diagram illustrating an example after the input image illustrated in FIG. 19 is binarized. It is a figure which shows the structure of the small area color reducer 616 which performs the gradation number determination process of the small area i performed by S5202 shown in FIG. It is a figure which shows the control structure of the process performed by the controller. FIG. 4 is a diagram showing a configuration of a small area gradation number verifier 619 shown in FIG. 3. It is a figure which shows the control structure of the process which the controller 2001 performs. It is a figure which shows the control structure of the process performed by S2102 shown in FIG. It is a figure which shows the control structure of the process which produces | generates the value of the label buffer performed in S2103 of FIG. It is a figure which shows the control structure of the process which initializes the small area | region label buffer shown by S2104 of FIG. It is a figure which shows the control structure of the process which produces | generates the value of the small area | region label buffer performed in S2105 of FIG. It is a figure which shows the control structure of the determination process performed in S2504 of FIG. It is a figure which shows the control structure of the determination process performed in S2507 of FIG. It is a figure which shows the relationship between an attention pixel and an adjacent pixel. FIG. 30 is a diagram showing a control structure of a small region label rewriting process performed in S2510 of FIG. 29. It is a figure which shows another example of the data structure of the line information buffer 609. It is a figure which shows the control structure of the process which the controller 601 performs when the line information buffer 609 is a buffer for storing image data. It is a figure which shows schematic structure of the image coding apparatus which concerns on 2nd Embodiment. It is a figure which shows the format of the encoding image stored in the encoding image buffer 3006 in 2nd Embodiment. It is a figure which shows schematic structure of the image encoder 3005 which concerns on 2nd Embodiment. It is a figure which shows the control structure of the process which the controller 3201 of the image encoder 3005 performs. It is a figure which shows the control structure of the process performed by S3304 of FIG. It is a figure which shows the control structure of the process performed by S3316 shown in FIG. It is a figure which shows the control structure of the process performed by S3318 of FIG. FIG. 10 is a diagram illustrating an example of how to determine the value of each element of a bit plane division rule table 3206. It is a figure which shows another example of the method of determining the value of each element of the bit plane division rule table. It is a figure which shows another example of how to define the value of each element of the bit plane division rule table. It is a figure which shows the input image before a color reduction. FIG. 47 is a diagram showing a reduced color image when the image shown in FIG. 46 is developed on a bit plane by the division method shown in FIG. FIG. 49 is a diagram showing a reduced color image when the image shown in FIG. 46 is developed on a bit plane by the division method shown in FIG. 48. It is a figure which shows the example of the bit plane division | segmentation rule table applied with respect to the small area | region whose number of small area gradations is 4. FIG. It is a figure which shows the example of the bit plane division | segmentation rule table applied with respect to the small area | region whose small area gradation number is 2. FIG. It is a figure which shows schematic structure of the image decoding apparatus concerning 3rd Embodiment. FIG. 52 is a diagram showing a schematic structure of the image decoder shown in FIG. 51. It is a figure which shows the control structure of the process which the controller 4301 of the apparatus concerning 3rd Embodiment performs. It is a figure which shows the control structure of the process performed by S4514 shown in FIG. It is a figure which shows the control structure of the bit plane integration process shown by S4515 of FIG. It is a figure which shows the structure of the bit plane integration rule table 4307 in case a small area | region gradation number is 8. FIG. It is a figure which shows the structure of the bit plane integration rule table 4307 in case the number of small area gradations is 4. FIG. It is a figure which shows the structure of the bit plane integration rule table 4307 in case the small area gradation number is two. It is a figure which shows the input image for demonstrating the problem of a prior art. It is a figure which shows an example of the processing result by a prior art.

Explanation of symbols

  301 image input device, 303 adaptive color reducer, 304 color-reduced image buffer, 305 output image creator, 601 controller, 602 small area information generator, 605 image binarizer, 607 line extractor, 610 character rectangle information extractor 614 character rectangle color reducer, 616 small area color reducer, 619 small area gradation number verifier, 620 small area color reducer, 621 pixel converter converter.

Claims (11)

An image color reduction device that outputs by reducing the number of gradations of input digital image data,
Dividing means for dividing the input image into small regions by a predetermined method;
For each small area, including after-color-reduction gradation number setting means for setting the number of gradations after color reduction of the small area based on the image data included in the small area,
The after-color-reduction gradation number setting means includes
Extraction means for extracting a predetermined shape as vector data from the input image;
The number of gradations after color reduction of the small area where the shape extracted by the extracting means exists is equal to or greater than the number of gradations after color reduction of the small area which is a non-background pixel area and does not have the predetermined shape. As described above, an image color reduction apparatus including gradation number setting means for setting the number of gradations after color reduction in each of the small areas.   The image color reduction device according to claim 1, wherein the predetermined shape is a straight line. An image color reduction device that outputs by reducing the number of gradations of input digital image data,
Dividing means for dividing the input image into small regions by a predetermined method;
For each small area, including after-color-reduction gradation number setting means for setting the number of gradations after color reduction of the small area based on the image data included in the small area,
The after-color-reduction gradation number setting means includes
A reduced color image generating means for generating a reduced color image obtained by reducing the color of the input image;
Of the adjacent pixels in the color-reduced image when the color-reduced image generation means performs color reduction with the same number of gradations, pixels having the same pixel value have a common value, and color reduction is made up of pixels having the common value. Labeling means for generating a reduced color connected area image including the connected area;
A small area common labeling means for applying a common label to the small areas including the same subtractive color connection area based on the output of the labeling means;
An image color reduction device that sets a common number of gradations in a small area to which a common label is assigned by the small area common labeling means. An image color reduction device that outputs by reducing the number of gradations of input digital image data,
Dividing means for dividing the input image into small regions by a predetermined method;
For each small area, including after-color-reduction gradation number setting means for setting the number of gradations after color reduction of the small area based on the image data included in the small area,
The after-color-reduction gradation number setting means includes
A reduced color image generating means for generating a reduced color image obtained by reducing the color of the input image;
In the color-reduced image when the color-reduced image generation unit performs color reduction with the same number of gradations, a boundary adjacent pixel that determines whether two adjacent pixels across a boundary between adjacent small regions have a common pixel value Value judging means;
If the boundary neighboring pixel value determining means determines that have a common pixel value, and a neighbor subregion common labeling means for applying a common label to a small region in contact該隣,
An image color reduction device that sets a common number of gradations in a small area to which a common label is assigned by the adjacent small area common labeling means. An image color reduction device according to any one of claims 1 to 4, for reducing the number of gradations of input digital image data;
An image encoding device comprising: an image encoding means for performing entropy encoding of an image reduced by the image color reduction device. Input means for inputting image data encoded by the image encoding device according to claim 5;
An image decoding apparatus comprising: decoding means for performing decoding corresponding to the entropy encoding according to claim 9 for the image data input by the input means. An image color reduction method for reducing the number of gradations of input digital image data and outputting the image,
Dividing the input image into small regions by a predetermined method;
For each of the small areas, setting the number of gradations after color reduction of the small area based on the image data included in the small area,
The step of setting the number of gradations after the color reduction is as follows:
Extracting a predetermined shape as vector data from the input image;
The number of gradations after color reduction of the small area where the shape extracted by the extracting step is not less than the number of gradations after color reduction of the small area which is a non-background pixel area and does not have the predetermined shape. An image color reduction method including a gradation number adjustment step for setting the number of gradations after color reduction in each of the small regions. An image color reduction method for reducing the number of gradations of input digital image data and outputting the image,
Dividing the input image into small regions by a predetermined method;
For each of the small areas, setting the number of gradations after color reduction of the small area based on the image data included in the small area,
The step of setting the number of gradations after the color reduction is as follows:
Generating a reduced color image obtained by reducing the color of the input image;
Among the adjacent pixels in the color-reduced image when the step of generating the color-reduced image is reduced by the same number of gradations, pixels having the same pixel value have a common value, and the pixels having the common value are used. A labeling step for generating a reduced color connected area image including the reduced color connected area,
Applying a common label to the small areas including the same subtractive color connection area based on the output of the labeling step,
An image color reduction method in which a common number of gradations is set in a small area to which a common label is assigned in the step of assigning a common label to the small area. An image color reduction method for reducing the number of gradations of input digital image data and outputting the image,
Dividing the input image into small regions by a predetermined method;
For each of the small areas, setting the number of gradations after color reduction of the small area based on the image data included in the small area,
The step of setting the number of gradations after the color reduction is as follows:
Generating a reduced color image obtained by reducing the color of the input image;
In the color-reduced image when the step of generating the color-reduced image is reduced by the same number of gradations, a boundary for determining whether or not two adjacent pixels across a boundary of adjacent small regions have a common pixel value An adjacent pixel value determination step;
When it is determined that the boundary adjacent pixel value determination step has a common pixel value, a step of assigning a common label to the adjacent small regions,
An image color reduction method in which a common number of gradations is set in a small area to which a common label is assigned in the step of assigning a common label to the small area. An image color reduction method according to any one of claims 7 to 9, for reducing the number of gradations of input digital image data;
And entropy coding of the image reduced in color by the image color reduction method. Inputting image data encoded by the image encoding method according to claim 10;
An image decoding method comprising: decoding the image data input in the input step corresponding to the step of performing entropy encoding according to claim 10.

Images