JP2006197178A - Image processor, image processing method, program, and storage medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor capable of attaining high compression of a color image, while suppressing degradation of image quality, and to provide an image processing method, a program, and a storage medium. <P>SOLUTION: The image processor is provided with a means for acquiring a multivalued processing target image, a means for acquiring a binary image generated on the basis of the multivalued image, a means for identifying a region having a specific attribute on the basis of either of the images, a means for generating a specific attribute binary image in which a part corresponding to a region, without having the specific attribute is replaced with white pixels in the binary image, a means for generating a non-specific attribute multivalued image, in which pixels at the specific attribute in the multivalued image are filled with a background color, a means for deciding colors composing the specific attribute, a means for generating the specific attribute binary image composed of each decided color, and a means for putting the non-specific attribute multivalued image and the specific attribute binary image comprising each decided color into one file by compressing and encoding each of the images. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a technique for highly compressing a color image.

Conventionally, as a method for highly compressing a color image, for example, a process of separating a character part and a picture part and compressing them by JPEG and MMR, as described in Patent Document 1, is performed in the picture part. Like recurring ones or those described in Patent Document 2, the color image is binarized, the binary image is used to remove the edge from the color image, and the binary image and the color from which the edge has been removed There have been proposed ones in which images are encoded, and those in which image quality is improved by expanding a line drawing and filling a background hole, such as those described in Patent Document 3. .
Japanese Patent No. 026111012 Japanese Patent No. 2872334 Japanese Patent No. 0287748

  The present invention relates to such a technique, and an object of the present invention is to provide an image processing apparatus, an image processing method, a program, and a storage medium that can highly compress a color image while suppressing deterioration in image quality. .

  The present invention provides a means for acquiring a multi-valued image as a processing target image, a means for acquiring a binary image generated based on the multi-valued image, and a region having a specific attribute based on any one of the images Means for generating a specific attribute binary image in which a portion corresponding to a region that is not the specific attribute in the binary image is replaced with a white pixel, and the specific attribute portion in the multi-value image Means for generating a non-specific attribute multi-valued image in which pixels are filled with a background color, means for determining a color constituting the specific attribute part, and means for generating the specific attribute binary image composed of each defined color; The non-characteristic attribute multi-valued image and the specific attribute binary image composed of the determined colors are each compression-encoded and provided in one file.

  Further, a means for acquiring a multi-valued image as a processing target image, a means for acquiring a binary image generated based on the multi-valued image, and an area having a specific attribute based on any one of the images Means for generating a specific attribute binary image in which a portion corresponding to the region that is not the specific attribute is replaced with a white pixel in the binary image, and a pixel of the specific attribute portion in the multi-value image Means for generating a non-specific attribute multi-valued image filled with a background color, means for determining a color constituting the specific attribute part, means for generating the specific attribute multi-valued image consisting of each defined color, A non-characteristic multi-valued image and a unit for compressing and encoding the specific attribute multi-valued image made up of each determined color and combining them into one file.

  Further, a means for acquiring a multi-valued image as a processing target image, a means for acquiring a binary image generated based on the multi-valued image, and an area having a specific attribute based on any one of the images Means for generating a specific attribute binary image in which a portion corresponding to the region that is not the specific attribute is replaced with a white pixel in the binary image, and a pixel of the specific attribute portion in the multi-value image Generating a non-specific attribute multi-value image in which the specific attribute portion is embedded, means for determining a color constituting the specific attribute portion, and generating color information of each specific attribute portion of the specific attribute binary image Means, and means for compressing and encoding the non-characteristic attribute multi-valued image, the specific attribute binary image, and the color information of each specific attribute part, respectively, and consolidating them into one file.

  Therefore, according to the present invention, it is possible to obtain an effect that a color image can be highly compressed while suppressing deterioration in image quality. In addition, the multi-value document image can be greatly reduced in size without sacrificing the character visibility.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

  In the following description, “specific attribute” represents “character area”, “specific attribute binary image” represents “binary image of characters only”, and “non-specific attribute multi-valued image” “Color image with characters erased” is represented, and “color information of specific attribute portion” represents “image representing character color”. The image data format targeted by the present invention is PDF (Portable Document Format; a registered trademark of Adobe Corporation), which provides a method for highly compressing a PDF image.

  FIG. 1 shows the configuration of an image processing apparatus according to an embodiment of the present invention.

  This image processing apparatus is a general-purpose device comprising a CPU (central processing unit) 1, a memory 2, a hard disk 3, an input device 4 such as a scanner for inputting images, a CD-ROM drive 5, a display 6, a mouse (not shown), and the like. A data processing apparatus (computer system).

  Here, the recording medium 7 such as the hard disk 3 or the CD-ROM drive 5 stores a program for realizing the processing functions and processing procedures of the present invention. The document image to be processed is stored in the hard disk 3 or the like through the input device 4, for example. The CPU 1 reads out and executes a program for realizing the processing functions and procedures described above from the recording medium 7, stores the result of the image processing in the hard disk 3, and outputs it to the display 6 as necessary.

  FIG. 2 is a flowchart showing a flow of image processing according to an embodiment of the present invention, FIG. 3 is a schematic diagram showing an outline of image processing according to an embodiment of the present invention, and FIG. 3 shows an example of a configuration for realizing the image processing of FIG. 3 in hardware.

  First, a multi-value document image is acquired using a multi-value image input means 101 such as an image scanner (step S101).

  Next, the position of a part having a specific attribute, for example, a character part is specified from the image by the specific attribute part extracting means 104 (step S103). Here, in order to know the position of the character, the method described in Japanese Patent Laid-Open No. 2002-288589 can be adopted when directly acquiring from the multi-valued image, and a binary image is generated from the multi-valued image. (Step S102) In the case of obtaining from a binary image that has been made, a character region extraction technique that has already been published, such as the method described in Japanese Patent Laid-Open No. 6-20092, may be used. The following description will be made assuming that the character area is acquired from the above. However, the character area may be extracted from multiple values (including color).

  Then, the white pixel replacement unit 105 erases black pixels other than characters in the binary image (step S104). With this process, the position of the character can be known in pixel units.

  Next, the specific attribute portion erased image generating means 103 creates an image from which the specific attribute portion has been erased (step S105). In a color image, an image in which pixels in a character portion are replaced with surrounding colors may be created.

  Next, the color of the specific attribute portion is determined by the specific attribute portion color determining means 106 (step S106). All the pixel colors of the color image at the position of the black pixel constituting the character are obtained, and several frequently used colors are selected from this data as representative colors. Then, it is determined for each pixel or for each connected component which representative color the pixel constituting the character is closest to.

  Next, the specific attribute pixel image generation unit 107 generates an image in which the pixel having the specific attribute has a color determined for each pixel or for each connected component (step S107). A multi-valued image having only a limited color (for example, PNG format) may be used, or one binary image may be provided for each color, but here it will be described as having a binary image for each color. I do.

  Next, the image encoding unit 108 generates a compressed image from an image in which the generated specific attribute pixel is deleted and an image including only the specific attribute pixel (step S108). For example, if the former is JPEG compression and the latter is MMR compression, the file size is effectively reduced.

  Next, the summary file creation unit 109 synthesizes the image having only the specific attribute pixels on the image from which the specific attribute portion has been erased (step S109). If you superimpose these, the text will be pasted on the background and you can see it like the original image.

  With the above processing, the file size can be greatly compressed without significantly reducing the visibility. The reason is as follows.

  (1) JPEG compression is not very efficient for images with large pixel value fluctuations, but if the character part is erased by the method described here, the pixel value fluctuations in the character part are eliminated and the efficiency is improved.

  (2) Since the number of colors of the character portion is greatly reduced, this also improves the compression efficiency.

  In addition, processing for further improving the image quality and compression rate can be added. FIG. 5 is a processing flow when processing is added, and FIG. 6 is a configuration diagram when processing is added.

  First, a multi-value document image is acquired using multi-value image input means 201 such as an image scanner (step S201).

  Then, the sharpening unit 210 smoothes the color image obtained in step S201 (step S202). This may be expressed as an image with weak edges in a color image. If binarization is performed as it is, the character becomes faint and the readability and compression efficiency of the character image are reduced. Because. This is because sharpening helps to improve the readability of characters.

  Further, the smoothing unit 211 smoothes the color image obtained by the process of step S201 (step S203). This is because in a color image, the intermediate color may be expressed by a set of pixels of different colors. If binarization is performed as it is, the inside of the character stroke becomes a binary image full of holes. This is because the compression efficiency of the character image is lowered. Further, the smoothing of the image may be performed on a color image (background image) from which characters are erased. This is because the background image is subjected to JPEG compression, so that the compression efficiency is increased by smoothing and it is useful for suppressing moire that may occur due to the reduction in resolution.

  Next, the position of a part having a specific attribute, for example, a character part is specified from the image by the specific attribute part extracting means 204 (step S205). Here, in order to know the position of the character, the method described in Japanese Patent Laid-Open No. 2002-288589 can be adopted when directly acquiring from the multi-valued image, and a binary image is generated from the multi-valued image. (Step S204) In the case of obtaining from a binary image that has been made, a character region extraction technique that has already been published, such as the method described in Japanese Patent Laid-Open No. 6-20092, may be used. The following description will be made assuming that the character area is acquired from the above. However, the character area may be extracted from multiple values (including color).

  Then, the white pixel replacement unit 105 erases black pixels other than characters in the binary image (step S206). With this process, the position of the character can be known in pixel units.

  Next, the connected component size examining unit 212 may extract the connected components of black pixels from the binary image from which characters other than characters have been deleted, and further delete those that are too large and too small (step S207). This is because a connected component that is too small is not likely to be a character such as noise, and the compression efficiency decreases if the binary image is used as it is. Also, it is technically difficult to specify a character area, and a correct character area is not always extracted. For this reason, if the original image has a figure area or a photo area, the possibility of erroneously setting it as a character cannot be discarded. Connected components that are too large are likely to be other than these characters. Also, if the large connected component is actually a character, it will be mistakenly classified as background in this process, but it is large enough to be read from the human eye.

  Next, the specific attribute portion erased image generating means 103 creates an image from which the specific attribute portion has been erased (step S208). In a color image, an image in which pixels in a character portion are replaced with surrounding colors may be created.

  Next, the color of the specific attribute portion is determined by the specific attribute portion color determining means 206 (step S209). All the pixel colors of the color image at the position of the black pixel constituting the character are obtained, and several frequently used colors are selected from this data as representative colors. Then, it is determined for each pixel or for each connected component which representative color the pixel constituting the character is closest to.

  Next, the specific attribute pixel image generation unit 207 generates an image in which the pixel having the specific attribute has a color determined for each pixel or for each connected component (step S210). A multi-valued image having only a limited color (for example, PNG format) may be used, or one binary image may be provided for each color, but here it will be described as having a binary image for each color. I do.

  Next, the contrast adjustment unit 213 performs contrast conversion of the image (background image) from which the specific attribute pixel is deleted and the binary image (character image) including only the specific attribute pixel (step S211), thereby reducing the contrast. Make a smooth image. This is because in the case of JPEG compression in the background image, the smaller the change in pixel value, the higher the compression efficiency. Since the character image is MMR, the contrast conversion does not affect the size, but the same contrast conversion as that of the background is performed so that the color image does not appear unnatural when superimposed.

  Next, the resolution conversion unit 214 converts the resolution of the image from which the specific attribute pixel is erased, and lowers the resolution (step S212). Compared to a character image, the background image has a lower effect on the visibility even if the resolution is somewhat lower, so the resolution is reduced in consideration of compression efficiency.

  Next, the image encoding unit 208 generates a compressed image from an image in which the generated specific attribute pixel is deleted and an image including only the specific attribute pixel (step S213). For example, if the former is JPEG compression and the latter is MMR compression, the file size is effectively reduced.

  Next, the summary file creation unit 209 combines the image with the specific attribute portion deleted with the image including only the specific attribute pixel so as to be displayed in an overlapping manner (step S214). If you superimpose these, the text will be pasted on the background and you can see it like the original image.

  FIG. 7 describes a method of superimposing the generated multi-valued background image and a binary image of only a plurality of characters. The background image is mainly compressed by JPEG (DCT) compression. If there is no character, the compression ratio can be increased. Further, if necessary, resolution conversion may be performed before compression.

  The binary image is a method of compressing one image at a time in the MMR format. The images generated in this way are superimposed and saved in one file. Then, by decoding (decompressing) the stored file and displaying it, a highly compressed file can be generated and displayed without impairing the readability of characters.

  FIG. 8 describes a method of superimposing the generated background multi-value image and one color image having 16 colors corresponding to 15 character colors. The background image is mainly compressed by JPEG (DCT) compression. In an image with only characters, an image clustered up to a maximum of 15 colors excluding the background color is assigned to the corresponding character color.

  Thereafter, unlike the processing flow having a plurality of binary images as shown in FIG. 7, the clustered character color has a maximum of 15 colors and a background color of 1 color, and a single color having a total of 16 colors. Generate an image. The generated 16 color character images are compressed by deflate compression (PNG format). Then, the background image and the character image are superimposed and stored in one file. There is a merit that the display speed of an image file in which images are superimposed is improved.

  FIG. 9 is a schematic diagram of a method of superimposing a generated background image with only a background, a foreground image indicating a character color, and a binary image for a mask. 10 and 11 are a process flow diagram and a configuration diagram thereof.

  First, a multi-value document image is acquired using a multi-value image input unit 301 such as an image scanner (step S301).

  Further, the position of a part having a specific attribute, for example, a character part is specified from the image by the specific attribute part extracting means 303 (step S303). In order to know the position of the character, the method disclosed in Japanese Patent Application Laid-Open No. 2002-288589 can be adopted in the case of obtaining directly from a multi-valued image. In addition, when a binary image is generated from a multi-valued image (step S304) and acquired from the generated binary image, a character area extraction technique that has already been published, such as Japanese Patent Laid-Open No. 6-20092, may be used. Here, the following description will be made on the assumption that the image is acquired from the binary image. However, the character region may be extracted from multiple values (including color).

  In addition, the white pixel replacement unit 304 deletes black pixels other than characters in the binary image (step S306). With this process, the position of the character can be known in pixel units.

  Next, the specific attribute portion erased image generating means 305 creates an image from which the specific attribute portion has been erased (step S308). In a color image, an image in which pixels in a character portion are replaced with surrounding colors may be created.

  Further, the specific attribute portion erasure image generation means 306 generates a binary image of only the characters generated in step S304 (step S309). Bit ON (here, black pixel) is used for processing as a pixel to be left in mask processing, and bit OFF (here, white pixel) is used as a pixel not to be left in mask processing.

  Further, the specific attribute portion color determination means 307 takes out the character color corresponding to the black pixel of the specific attribute portion from the input color image and paints it (step S310). The specific attribute color determination unit 308 generates the image drawn in step S310 as a foreground image. At this time, pixel portions other than characters are drawn as a background color.

  In addition, the image encoding unit 309 draws a DCT (JPEG) compressed image on the generated background image from which the specific attribute pixel is deleted, an MMR compressed image on the character mask image including only the specific attribute pixel, and only the specific attribute portion. A DCT (JPEG) compressed image is generated for the foreground image thus obtained (step S314). When compression is performed in this way, the file size is effectively reduced.

  In addition, the summary file creation unit 310 applies a character mask image to the background image from which the specific attribute portion has been deleted and the foreground image including only the specific attribute pixels, and obtains the foreground image remaining in the pixels corresponding to the black pixel portion. The images are combined in a form that can be superimposed and displayed (step S315). If you superimpose these, the text will be pasted on the background and you can see it like the original image.

  With the above processing, the file size can be greatly compressed without significantly reducing the visibility. The reason is as follows.

  (3) DCT (JPEG) compression is not very efficient for an image with a large variation in pixel value, but if the character portion is erased by the method described here, the pixel value variation in the character portion is eliminated, so the efficiency is high. Get better.

  (4) Although the character portion is DCT (JPEG) compressed, the compression efficiency can be improved without impairing the readability of the character by mask processing.

  FIG. 12 is a processing flow relating to decoding. First, a superimposed image is acquired (step S401). Next, each constituent image is decoded (step S402).

  Then, the decoded image data is superimposed and displayed (step S404). Here, if the resolution is converted in step S403, a scaling process for returning to the original resolution is performed (step S403).

  In summary, by decoding (decompressing) the saved file and displaying it, a highly compressed file can be generated and displayed without impairing the readability of the characters.

  Of course, the created file may be transmitted via a communication means such as the Internet, or may be output to an image display device such as a display or an external output device such as a printer.

  Now, as shown in FIG. 13, the points of the present invention are as follows.

  First, in the multi-valued image, the edge portion of the character changes smoothly as shown in FIG. When binarization is performed, a black-and-white boundary is created somewhere in the position where the threshold value is smoothly changed using some threshold ((B) in FIG. 13). If is used as it is, the outline of the character remains without being erased ((C) of FIG. 13). Since the remaining portion also has a difference in pixel value from the surrounding area, the compression efficiency is lowered and the image quality is lowered due to mosquito noise.

  For this purpose, it is possible to suppress a marginal area by filling a region including a specific region (character region) with a wide background color.

  That is, in the case of a character composed of one character block as shown in the upper flow of FIG. 14A, the character region extraction method is performed by looking at a black pixel block (= connected component). Can be extracted.

  A plurality of methods for extracting a character area from the coordinate position of a character line or a character line from the coordinate positions of these character blocks have been proposed in the past, and these general methods may be used. Does not reach.

  As shown in FIGS. 14 (a) and 14 (b), the background color is obtained by using the average color of the peripheral pixels or by calculating the distribution of the peripheral pixels using a histogram, and determining the intermediate value and the brightest / darkest color. And so on.

  As an example of embedding with a bright color, in the example of black characters on a white background, it is possible to eliminate a color difference from the background by embedding with a bright color. On the other hand, as an example of filling with a dark color, there is a black / white inverted character (a white character on a black background).

  Furthermore, when the distribution of the surrounding colors is within a predetermined range, it may be determined that they are the same color, and only when they are the same color, a process of filling with the calculated representative color may be performed.

1 is a block diagram showing a configuration of an image processing apparatus according to an embodiment of the present invention. 3 is a flowchart showing a flow of image processing according to an embodiment of the present invention. 1 is a schematic diagram showing an outline of image processing according to an embodiment of the present invention. The block diagram which showed an example of the structure at the time of implement | achieving the image processing of FIG. 3 by hardware. 6 is a flowchart showing a flow of image processing according to another embodiment of the present invention. The block diagram which showed an example of the structure at the time of implement | achieving the image processing of FIG. 5 by hardware. Schematic which described the method of superimposing the produced | generated multi-value image of the background and the binary image of only the character for several sheets. Schematic which described the method of superimposing the produced | generated multi-valued image of a background and one color image which has 16 colors for 15 character colors. It is a schematic diagram about the method of superimposing from the background image only of the produced | generated background, the foreground image which shows a character color, and the binary image for a mask. The processing flowchart of FIG. The block diagram of FIG. The figure which showed the processing flow regarding a decoding. Schematic for demonstrating the point of this invention. Schematic for demonstrating the extraction method of a character area.

Explanation of symbols

1 CPU (Central Processing Unit)
2 memory

Claims (15)

Means for acquiring a multi-valued image as a processing target image;
Means for obtaining a binary image generated based on the multi-valued image;
Means for identifying a region having a specific attribute based on any one of the images;
Means for generating a specific attribute binary image in which a portion corresponding to a region that is not the specific attribute is replaced with a white pixel in the binary image;
Means for generating a non-specific attribute multi-value image in which pixels of the specific attribute portion in the multi-value image are filled with a background color;
Means for determining a color constituting the specific attribute part;
Means for generating the specific attribute binary image composed of each determined color;
An image processing apparatus comprising: means for compressing and encoding the non-characteristic attribute multi-valued image and the specific attribute binary image made up of each determined color into a single file. Means for acquiring a multi-valued image as a processing target image;
Means for obtaining a binary image generated based on the multi-valued image;
Means for identifying a region having a specific attribute based on any one of the images;
Means for generating a specific attribute binary image in which a portion corresponding to a region that is not the specific attribute is replaced with a white pixel in the binary image;
Means for generating a non-specific attribute multi-value image in which pixels of the specific attribute portion in the multi-value image are filled with a background color;
Means for determining a color constituting the specific attribute part;
Means for generating the specific attribute multi-valued image comprising the determined colors;
An image processing apparatus comprising: means for compressing and encoding the non-characteristic multi-valued image and the specific attribute multi-valued image having each determined color, and combining them into one file. Means for acquiring a multi-valued image as a processing target image;
Means for obtaining a binary image generated based on the multi-valued image;
Means for identifying a region having a specific attribute based on any one of the images;
Means for generating a specific attribute binary image in which a portion corresponding to a region that is not the specific attribute is replaced with a white pixel in the binary image;
Means for generating a non-specific attribute multi-value image in which pixels of the specific attribute portion in the multi-value image are filled with a background color;
Means for determining a color constituting the specific attribute part;
Means for generating color information of each specific attribute portion of the specific attribute binary image;
An image processing apparatus comprising: means for compressing and encoding color information of the non-characteristic attribute multi-valued image, the specific attribute binary image, and the specific attribute parts, and combining them into one file. Means for generating a non-specific attribute multi-valued image in which pixels of the specific attribute part are filled with a background color,
The image processing apparatus according to claim 1, wherein the image processing apparatus is an area including a specifying unit.   The image processing apparatus according to claim 1, wherein the specific attribute area is a black pixel block (= black pixel connected component) constituting a character.   The specific attribute area is a character area, a character word area, or a line area, which is composed of one or a plurality of black pixel blocks (= black pixel connected components) constituting a character. Alternatively, the image processing apparatus according to claim 2, claim 3, or claim 4.   The image processing apparatus according to claim 1, wherein the specific attribute area is an area having the same surrounding color of the specific attribute.   When the surrounding color is darker than the specific attribute area (= reverse character), it is the darkest color of the surrounding color. Conversely, when the surrounding color is light (= normal character), it is filled with the background color with the brightest color of the surrounding color. 5. The image processing apparatus according to claim 1, wherein the non-specific attribute multi-valued image is generated.   The image processing apparatus according to claim 1, wherein the one file is transmitted. 5. The image processing apparatus according to claim 4, 5, 6, 7, or 8.   Means for acquiring the one file generated by the image processing apparatus according to claim 1, claim 2, claim 3, claim 4, claim 5, claim 6, claim 7, or claim 8; An image processing apparatus comprising: means for decoding information constituting the file.   The image processing apparatus according to claim 10, further comprising means for printing the decrypted information.   12. The image processing apparatus according to claim 10, further comprising means for displaying the decrypted information. Obtaining a multi-valued image as a processing target image;
Obtaining a binary image generated based on the multi-valued image;
Identifying a region having specific attributes based on any of the images;
Generating a specific attribute binary image in which a portion corresponding to a region that is not the specific attribute is replaced with a white pixel in the binary image;
Generating a non-specific attribute multi-valued image in which pixels of the specific attribute part in the multi-valued image are filled with a background color;
Determining a color constituting the specific attribute portion;
Generating the specific attribute binary image composed of the determined colors;
An image processing method comprising: a step of compressing and encoding the specific attribute binary image composed of the non-characteristic attribute multi-valued image and each determined color and combining them into one file. Obtaining a multi-valued image as a processing target image;
Obtaining a binary image generated based on the multi-valued image;
Identifying a region having specific attributes based on any of the images;
Generating a specific attribute binary image in which a portion corresponding to a region that is not the specific attribute is replaced with a white pixel in the binary image;
Generating a non-specific attribute multi-valued image in which pixels of the specific attribute part in the multi-valued image are filled with a background color;
Determining a color constituting the specific attribute portion;
Generating the specific attribute binary image composed of the determined colors;
A program that includes a step of compressing and encoding the specific attribute binary image composed of the non-characteristic attribute multi-valued image and each determined color and grouping them into one file, and executable by a computer system. Obtaining a multi-valued image as a processing target image;
Obtaining a binary image generated based on the multi-valued image;
Identifying a region having specific attributes based on any of the images;
Generating a specific attribute binary image in which a portion corresponding to a region that is not the specific attribute is replaced with a white pixel in the binary image;
Generating a non-specific attribute multi-valued image in which pixels of the specific attribute part in the multi-valued image are filled with a background color;
Determining a color constituting the specific attribute portion;
Generating the specific attribute binary image composed of the determined colors;
A storage medium storing a program executable by a computer system, the program having a step of compressing and encoding the specific attribute binary image composed of the non-characteristic attribute multi-valued image and each determined color and combining them into one file .