JP4696738B2 - Image data compression method and apparatus, and pseudo gradation image restoration method - Google Patents

Description

  The present invention relates to a method and apparatus for compressing pseudo gradation image data using a dither pattern, and a pseudo gradation image restoration method.

  In general, a printing system including a computer and a printer is used. In such a printing system, image processing such as pseudo gradation processing is performed on the computer side on the image data to be printed, and special processing is performed on the image data transmitted from the computer on the printer side. However, there is a method in which the data is output as it is and printed on paper.

  Conventionally, a raster printer is known as a printer used in this method. Since the raster printer does not perform special image processing on the printer side, it can be realized at low cost using small-scale hardware. However, on the other hand, since image data that can be printed on paper as it is is transmitted to the printer, there is a drawback that the amount of data to be transmitted becomes large. In addition, image data has become larger due to the recent demand for higher image quality, and as a result, the amount of data to be transmitted to the printer becomes even larger. Therefore, it takes a lot of time to transmit image data from the computer to the printer.

  In order to solve this problem, it is necessary to compress the image data on the computer side. However, if the processing for compression is complicated, it takes time for the compression processing itself, and it takes time for the expansion processing on the printer side as well, and the scale of hardware for that becomes large and expensive. It becomes.

  Therefore, in order to realize a printing system that enables the use of inexpensive printers and performs high-speed processing as a whole, the compression processing is light, the processing at the time of compression is light, can be developed with small hardware, and a high compression rate can be realized. Technology is desired.

  By the way, in order to print multi-gradation image data with a printer, in many cases, the image data must be converted into pseudo gradations. A dither method has been well known as a pseudo gradation method. In the dither method, a dither pattern is used, and each threshold value of the dither pattern is compared with the value of each pixel of the image data to be binarized or multi-valued.

  As a compression technique focusing on the dither pattern, there is a method disclosed in Patent Document 1. According to the method disclosed in Patent Document 1, the binarized image data is divided into blocks of a predetermined size. The reference pattern corresponding to the number of dots included in an arbitrary image block is compared with the image block, and if the image block and the reference pattern match, a code indicating the reference pattern is assigned. Otherwise, the comparison result Is encoded.

  Further, as a compression technique for binarized image data, the JBIG system, which has been approved for adoption by ITU (International Telecommunication Union) -T recommendation, is known.

Further, Patent Document 2 discloses a method of compressing image data that has been pseudo-gradated with a dither pattern. According to the method described in Patent Document 2, the pixel value for each pixel is compared with a threshold value, thereby obtaining a range for each pixel.
JP-A-6-152986 USP 6201614

  However, in the method of Patent Document 1 described above, when the designated image block matches the reference pattern, the reference pattern is encoded, so the amount of data to be transmitted increases by the amount of the reference pattern code. . Furthermore, since the image data changes, it is unlikely that all the pixels match in the block. In this case, the difference data is transmitted, so that the amount of data further increases.

  On the other hand, in the JBIG method, the compression process is relatively light, and the compression process can be performed at high speed even on a computer. However, according to the JBIG method, the compression rate is not so high as compared with other methods such as the LZS method or the LZW method, and the compression rate decreases particularly when halftone data increases.

  Further, in the method of Patent Document 2, since the value is determined for each pixel, the amount of information is not reduced by itself. Therefore, in order to reduce the amount of information, a means for separately compressing is necessary.

  The present invention has been made in view of the above-described problems, and can compress image data that has been pseudo-graded using a dither pattern at a high speed and with a high compression rate, and can be performed at high speed with a small amount of hardware. It is an object of the present invention to provide a method and apparatus for compressing image data and a method for restoring a pseudo gradation image that can be restored by the above method.

  A compression method according to the present invention is a method for compressing pseudo gradation image data that is a pseudo gradation image using a dither pattern, wherein the pseudo gradation image is partitioned into predetermined regions. 1 step, a second step of determining a representative value of each region, and a third step of creating compressed data using the determined representative value. In the second step, For a region of interest that is one of the regions, a representative value of the region of interest and a representative value of one or a plurality of regions adjacent to the region of interest are used according to a predetermined interpolation rule. Such a representative value is obtained such that the pseudo gradation image in the region of interest is reproduced when the density of each pixel is interpolated and pseudo gradation is performed using the dither pattern. Umate determined as the representative value of the region of interest.

  Alternatively, in the second step, the attention area that is one of the areas is determined in advance using a representative value of the attention area and a representative value of one or more areas adjacent to the attention area. In accordance with the interpolation rule, the density of each pixel in the region of interest is interpolated, and the dot arrangement pattern of the image data obtained by performing pseudo gradation using the dither pattern and the image data in the region of interest Such a representative value that matches the dot arrangement pattern within a predetermined error range is obtained and determined as the representative value of the region of interest.

  Preferably, in the second step, the representative value is a value corresponding to a density value of a pixel located at an end of each region.

  In the second step, when a value that can be selected as the representative value has a predetermined range, a median value in the range is determined as the representative value.

  Further, the interpolation rule interpolates the density of each pixel by a plane defined by a total of three representative values of the representative value of the region of interest and the representative values of two regions adjacent to the region of interest. .

  In the second step, when a representative value cannot be determined in the attention area, the image data in the attention area is determined as an alternative representative value of the attention area.

  In the second step, when the representative value cannot be determined in the attention area, the attention area is further divided into a plurality of small areas, and the representative value is determined in each small area.

  In the second step, when the representative value cannot be determined in the small area, the image data in the small area is determined as an alternative representative value of the small area.

  The restoration method according to the present invention obtains a representative value of each region from the compressed data, and for a region of interest that is one of the regions, the representative value of the region of interest and 1 adjacent to the region of interest. A pseudo gradation image obtained by interpolating the density of each pixel in the region of interest according to a predetermined interpolation rule using a representative value of one or a plurality of areas and further performing pseudo gradation using the dither pattern To restore.

  A compression apparatus according to the present invention is an apparatus that compresses pseudo gradation image data that is a pseudo gradation image using a dither pattern, and divides the pseudo gradation image into predetermined areas. 1 means, second means for determining the representative value of each area, and third means for creating compressed data using the determined representative value, wherein the second means For the attention area that is one of the areas, the attention area according to a predetermined interpolation rule using the representative value of the attention area and the representative value of one or more areas adjacent to the attention area Obtaining such representative values so that the pseudo gradation image in the region of interest can be reproduced when pseudo gradation is performed using the dither pattern. The area of interest It is determined as the representative value.

  According to the present invention, image data that has been pseudo-gradated using a dither pattern can be compressed at a high speed and with a high compression rate, and can be restored at high speed with a small amount of hardware. As a result, the entire process can be performed in a short time while using an inexpensive printer.

In the description of the present embodiment, first, the case of compressing pseudo-graded binary image data will be described, and second, the case of compressing pseudo-gradated 4-level image data. Will be described. In the second description, only the main points of the differences from the first will be described.
[First]
Embodiment for compressing binary image data [Configuration of entire system]
FIG. 1 is a diagram showing an overall configuration of a print system 1 to which an image data compression method according to the present invention is applied.

  In FIG. 1, the print system 1 includes a computer main body 11, a display 12, a printer 13, a keyboard 14, a mouse 15, and the like.

  The computer main body 11 pseudo-gradates image data input from the outside or image data generated inside (original image data) using a dither pattern by a dither method. The original image data is, for example, gradation image data of 256 gradations for each color of RGB or CMY. Pseudo gradation is sometimes called halftoning. The pseudo gradation image is referred to as a pseudo gradation image, a halftone image, or the like. In the present embodiment, binarization is performed as pseudo gradation processing, and binary image data D1 is generated. Then, the generated binary image data D1 is compressed to generate compressed data D2, and the generated compressed data D2 is transmitted to the printer 13. The printer 13 restores binary image data (halftone image) with respect to the received compressed data D2 by referring to the dither pattern used for binarization and prints it out on paper or the like. Note that FIG. 29 can be referred to for the relationship between the binary image data D1 and the dither pattern.

  In the present embodiment, during the compression process, the binary image data D1 is separated into a dither pattern information component and an image information component. Then, only the image information component is extracted and encoded by the Huffman method to obtain compressed data D2. The dither pattern information component is not included in the compressed data D2. Incidentally, conventionally, both the dither pattern information component and the image information component have been compressed together.

  More specifically, in the course of the compression processing of the present embodiment, the binary image data D1 is partitioned for each predetermined area (a low-resolution area TL described later) (first step). Next, the representative value of each area is determined (second step). Then, compressed data is created using the determined representative value (third step). The representative value may be represented by “a” or “AP”.

  In this second step, for an attention area that is one of the areas, a predetermined interpolation rule using a representative value of the attention area and a representative value of one or more areas adjacent to the attention area Such that the pseudo gradation image in the region of interest is reproduced when pseudo-gradation is performed using the dither pattern obtained by interpolating the density of each pixel in the region of interest Is determined as a representative value a of the region of interest.

  The region adjacent to the region for which the representative value a is determined is set as the next region of interest, and the region of interest is sequentially moved to determine the representative value a for all regions. Since there is no area for which the representative value “a” has already been determined, the area that has been initially set as the attention area is representative in consideration of the density value of the pixel located at the periphery or end of the attention area, particularly at the apex. The value a may be determined.

  In the second step, the method for obtaining the representative value can also be said as follows. That is, for a region of interest that is one of the regions, the region of interest according to a predetermined interpolation rule using a representative value of the region of interest and a representative value of one or more regions adjacent to the region of interest. Is obtained by interpolating the density of each pixel in the image, and the dot arrangement pattern of the image data obtained by performing pseudo gradation using the dither pattern and the dot arrangement pattern of the image data in the region of interest have a predetermined error. Such a representative value that matches within the range is obtained and determined as the representative value a of the region of interest.

  In the second step, the representative value a is set to a value corresponding to the density value of the pixel located at the end of each region. Further, when a value that can be selected as the representative value has a predetermined range, for example, the median value is determined as the representative value a in the range. The predetermined range that can be selected as the representative value may be referred to as a “representative value range”.

In the second step, an interpolation rule is used when obtaining the representative value a. The following is an example of the interpolation rule.
(1) The density of each pixel is interpolated by a plane defined by a total of three representative values of the representative value of the attention area and the representative values of two areas adjacent to the attention area. In addition, linear interpolation using a plane other than a plane is used.
(2) The density of each pixel is interpolated by a curved surface (twisted surface) defined by a total of four representative values of the representative value of the attention area and the representative values of three areas adjacent to the attention area.
(3) The density of each pixel is interpolated by the spline curved surface defined by the representative value of the attention area and the representative values of a plurality of areas adjacent to the attention area. Also, polynomial interpolation is used instead of spline interpolation.

  In this way, the interpolation rule can be expressed by an interpolation formula, an interpolation table, or the like in addition to what has been described. The interpolation rule may be determined in advance as a precondition for the compression process and the decompression process, and the compression process is performed by applying a different interpolation rule depending on the type of the original image data. The used interpolation rule may be notified to the restoration processing side. Specifically, the interpolation rule can be realized, for example, by a processing circuit, a program, a conversion table, or the like that executes the interpolation rule.

  By the way, as another method for obtaining the representative value a, it is determined whether the dot arrangement pattern of the image data in each area matches the dither pattern used for binarization. There is a method of obtaining a representative value related to the density of image data. This method was previously proposed as Japanese Patent Application No. 2004-377575. Therefore, this method is sometimes referred to as “prior application method”. Although the prior application method is different from the method of the present embodiment, the basic concept of “representative value” in compression is the same as that of the present embodiment, and is very helpful in understanding the significance of “representative value”. Therefore, in this specification, reference is made by writing or the like as necessary.

  Returning to the description, when the representative value a of each region is determined, in the third step, difference information of the representative value a between adjacent regions is obtained, and the obtained difference information is encoded by the Huffman method. To obtain compressed data D2. The dither pattern used for binarization is a dither pattern information component, and the representative value a is an image information component.

  In the binarization process, a plurality of dither patterns are switched and used in accordance with the type of area such as a character area or an image area. Therefore, in the compression process, the plurality of dither patterns are switched. In the prior application method, a plurality of dither patterns are switched to determine whether or not the pattern matches the dot arrangement pattern of the image data in each region.

  In the second step, when the representative value a cannot be determined in the attention area, the attention area is further divided into a plurality of small areas (a medium resolution area TM described later), and the representative value a is determined in each small area. To do.

  In the prior application method, if the pattern does not match any dither pattern, the area is further divided into a plurality of small areas, and the dot pattern of the image data in each small area matches the dither pattern. Determine whether or not. If they match, the representative value is the number of dots hit in the small area. If they do not match again, the compressed data D2 is obtained by encoding the dot arrangement of these areas or small areas.

  The computer main body 11 transmits the obtained compressed data D2 to the printer 13. The computer main body 11 also transmits in advance to the printer 13 information related to the dither pattern and a decoding table DC that is a table for restoration (decoding) by the Huffman method. The interpolation rule described above for interpolating the density of each pixel in each region is also transmitted in advance to the printer 13 as necessary. As the decoding table DC, there are decoding tables DC1 to DC3 for the low resolution area, the medium resolution area, and the high resolution area, respectively. Details will be described later.

  In the printer 13, the representative value a of each area is acquired from the compressed data D2 transmitted from the computer main body 11, and for the attention area that is one of the areas, the representative value a of the attention area and the attention area. Is obtained by interpolating the density of each pixel in the region of interest according to a predetermined interpolation rule using a representative value a of one or a plurality of regions adjacent to, thereby obtaining a virtual gradation image, The pseudo gradation image is restored by performing pseudo gradation using the dither pattern for the virtual gradation image.

  In the prior application method, the printer 13 acquires the representative value a of each area from the compressed data D2 transmitted from the computer main body 11, compares the acquired representative value a with the corresponding dither pattern, The same dot pattern (dot arrangement pattern) as the binary image data D1 is reproduced.

  In such restoration processing (decoding processing), the decoding table DC is used for restoration by the Huffman method.

As the computer main body 11, a personal computer, a workstation, and other various computers are used. As the printer 13, a raster printer, a GDI printer, and other various printers are used.
[Compression processing]
Next, the compression process in the computer main body 11 will be described in more detail.

  2 is a block diagram showing a functional configuration of the computer main body 11, FIG. 3 is a block diagram functionally showing the compressed data creating unit 100 of the computer main body 11, and FIG. 4 is a block diagram showing a functional configuration of the printer 13. 5 is a diagram for explaining the partitioned low resolution region TL, FIG. 6 is a diagram illustrating an example of the low resolution region TL and the dither pattern DP, and FIG. 7 is a diagram illustrating the low resolution region TL and the dither pattern DP having different sizes. FIG. 8 is a diagram showing the position of the representative value a in each low resolution region TL, and FIG. 9 is a diagram for explaining an example of the interpolation rule HR when the density of each pixel in the low resolution region TL is obtained by interpolation. FIG. 10 is a diagram for explaining how to obtain the representative value a. The image data is two-dimensional, and each low-resolution region TL is also two-dimensional. In FIG. 10, a one-dimensional region is schematically shown for the sake of simplicity of explanation.

  As shown in FIG. 2, the computer main body 11 includes an image database 150, a dither pattern storage unit 160, a pseudo gradation unit 170, a compressed data creation unit 100, and the like.

  The image database 150 holds a large number of image data (original image data) FD1, FD2,. All or part of the image data FD1 to FD may be described as “image data FD”. Note that most of the image data FD is a color image. In this case, one image data FD is composed of image data of each color of C (cyan), M (magenta), Y (yellow), and K (black). Is done. The image data FD is obtained, for example, by inputting from a scanner, a digital camera, or other external source, or by editing or generating inside.

  The dither pattern storage unit 160 holds a plurality of types of dither patterns DP11 to DP13, DP21 to 23, DP31 to 33, and DP41 to 43. All or part of these may be described as “dither pattern DP”.

  The dither pattern DP is a character area dither pattern DP11, 21, 31, 41, an image area dither pattern DP12, 22, 32, 42, and a chart area dither pattern DP13 for each color of C, M, Y, and K. 23, 33, 43, etc. are provided.

  The pseudo gradation unit 170 performs binarization processing on the image data FD stored in the image database 150 using the dither pattern DP selected from the dither pattern storage unit 160, and the obtained binary image data D1. Is sent to the compressed data creation unit 100. The binary image data D1 is also composed of binary image data of C, M, Y, and K colors.

  The compressed data creation unit 100 compresses the binary image data D1 to generate compressed data D2, and outputs the generated compressed data D2, the dither pattern DP, the interpolation rule HR, and the decode table DC to be transmitted to the printer 13. To do. However, when an interpolation rule is predetermined between the computer main body 11 and the printer 13, it is not necessary to transmit the interpolation rule HR.

In FIG. 3, the compressed data creation unit 100 includes a low resolution region partition unit 101, a low resolution representative value calculation unit 103, a medium resolution region partition unit 104, a medium resolution representative value calculation unit 106, a dot pattern extraction unit 107, and an encoding unit. 108, a data transmission unit 109, an interpolation rule holding unit 110, and the like.
[Low-resolution area processing]
The low resolution area dividing unit 101 divides the binary image data D1 sent from the pseudo gradation unit 170 into a low resolution area TL, which is a large number of small areas. The low resolution region TL is a rectangular region including 96 (6 × 16) pixels TP as shown in FIG. 5, for example. In the example of FIG. 5, the entire area of the binary image data D1 is TL21, TL22, TL23,... For the next row, such as the low resolution areas TL11, TL12, TL13,. Thus, each block is divided up to TLmn.

  In the low resolution region TL shown in FIG. 6A, the black pixels are pixels on which dots are formed. When the density value of each pixel of the image data FD is equal to or greater than the threshold value of the corresponding pixel of the dither pattern DP shown in FIG. 6B, a dot is formed on that pixel. That is, the larger the density value, the higher the density.

  In the example of FIG. 6, an example in which the size of the low resolution region TL is divided so as to be the same as the size of the dither pattern DP is shown, but these sizes may be different from each other.

  For example, as shown in FIG. 7, the size of the low resolution region TLb may be smaller than the size of the dither pattern DPb. Contrary to the example of FIG. 7, the size of the low resolution area TL may be larger than the size of the dither pattern DP.

The low resolution representative value calculation unit 103 obtains a representative value a for each low resolution region TL partitioned by the low resolution region partition unit 101 with reference to the interpolation rule HR stored in the interpolation rule holding unit 110. The method for obtaining the representative value a is as follows.
[How to find representative values]
In describing how to obtain the representative value a, first, the representative value position AP, which is the position of the representative value a, will be described.

  In FIG. 8, a representative value position AP, which is the position of the representative value a in each low resolution region TL, is shown for a part of the binary image data D1. As shown in FIG. 8, the representative value position APmn of each low resolution area TLmn is located at the lower right end of each low resolution area TLmn. Outside the area of the binary image data D1, there is a representative value position AP at the lower right corner when it is assumed that the same division as the low resolution area TL is performed. Note that the value of the representative value a may be indicated by “APmn” which is a code indicating the representative value position.

Now, how to obtain the representative value a will be described on the assumption that the low resolution region TL is one-dimensional.
[Representative value in the case of one dimension]
10, for the binary image data D1, one-dimensional low-resolution areas TL1, 2, 3,... Each including five pixels TP are shown. Since each pixel TP is binary, it is shown in white or black. For example, in the low resolution region TL1, the value of each pixel TP is white, white, white, black, black. The binary image data D1 is obtained by performing binarization processing on the image data (original image data) FD shown in the figure using the dither pattern DP shown in the figure. In other words, if the image data FD is less than or equal to the threshold value of the dither pattern DP, it is “white”, and if it exceeds the threshold value, it is “black”.

  First, the first representative value AP0 is determined. Since the first representative value AP0 is outside the area of the binary image data D1, any value can be taken. However, the density value (halftone data) of the leftmost pixel TP in the low resolution area TL1, the dither pattern DP An appropriate value may be set corresponding to the threshold value. For example, it may be an intermediate value of the range of image data calculated from the density value and the threshold value.

  Then, the low resolution region TL1, which is the first block, is set as the attention region, and attention is paid to white pixels in the attention region. From the first representative value AP0, straight lines LG1, 2, and 3 are drawn toward the threshold value of each white pixel in the region of interest. The straight line LG3 at the lowest position is selected, the straight line LG3 is extended, and the intersection with the position of the last pixel TP of the target area is set as the upper limit position. The density value at this upper limit position is set as the representative value upper limit APmax in the target area. And

  Processing similar to that for white pixels is also performed for black pixels. That is, straight lines LG4, LG4 are drawn from the first representative value AP0 toward the threshold value of each black pixel in the region of interest. The straight line LG4 at the highest position is selected, the straight line LG4 is extended, and the intersection with the position of the last pixel TP of the target area is set as the lower limit position. The density value at this lower limit position is set as the representative value lower limit APmin in the target area. And

When the representative value upper limit APmax thus determined is larger than the representative value lower limit APmin, that is, the following equation (1):
APmax ≧ APmin (1)
Is established, the value between the representative value upper limit APmax and the representative value lower limit APmin (representative value range) can be selected as the representative value of the region of interest. In the present embodiment, the median value between the representative value upper limit APmax and the representative value lower limit APmin is determined as the representative value AP1 of the low resolution region TL1.

  If the representative value AP1 is determined in this way, the previous representative value AP0 corresponding to the adjacent area and the current representative value AP1 corresponding to the attention area are connected by the straight line LA1, and the straight line LA1 and the dither are connected. By using the pattern DP, the binary image data D1 in the low resolution region TL1 can be completely reproduced.

  That is, by interpolating the density of each pixel TP in the low resolution region TL1 by a straight line LA1 connecting the previous representative value AP0 and the current representative value AP1, a virtual gradation image composed of the obtained density of each pixel TP is obtained. By performing pseudo gradation using the dither pattern DP, reconstructed image data FS shown in FIG. 10 is obtained. The reconstructed image data FS is the same as the original binary image data D1 in the low resolution area TL1. That is, the binary image data D1 is reproduced based on the representative values AP0,1. Note that the virtual gradation image composed of the density of each pixel TP is a straight line LA1 when this is represented continuously. Therefore, the virtual gradation image data may be indicated by “LA”.

  Therefore, in this case, the representative value AP1 obtained as described above is determined as the representative value a of the low resolution region TL1. The interpolation rule HR in this case is that the density of each pixel TP is interpolated by a straight line LA1 defined by a total of two representative values of the representative value AP1 of the attention area and the representative value AP0 of the area adjacent to the attention area. is there. The straight line LA1 usually has an inclination with respect to the horizontal. However, in reality, the low resolution region TL is often two-dimensional, and in this case, the interpolation is not performed by the straight line LA1 but by a plane. This corresponds to (1) in the example of the interpolation rule HR described above.

If the above equation (1) does not hold,
APmax <APmin
In this case, the previous representative value AP0 is corrected within the range between the representative value upper limit APmax and the representative value lower limit APmin. Using the corrected representative value AP0, the representative value AP1 of the attention area is obtained again as described above.

  If the representative value AP or the representative value range still cannot be obtained, a separate process is performed after recording that fact. As a separate process, for example, the attention area is further divided into a plurality of small areas, and the above processing is performed in each small area to determine the representative value APS. The small area is, for example, a half area of the attention area or a quarter area. By using a small area, the representative value APS is obtained in many cases.

  Further, when the representative value AP cannot be obtained in the small region, the representative value upper limit APmax and the representative value lower limit APmin are appropriately determined, and an intermediate value within the range is determined as the representative value AP. In such a case, the original binary image data D1 cannot be accurately reproduced in that region, but it is possible to tolerate even if there is some error.

  When the representative value APS cannot be determined in the small area, the image data in the small area may be determined as an alternative representative value APD for the small area. Further, the image data in the region of interest may be determined as an alternative representative value APD of the region of interest without being divided into small regions. In this case, as a matter of course, the alternative representative value APD and the representative value APS of the small area are different in restoration method from the normal representative value AP of the entire attention area.

  In this way, the representative value AP1 of the low resolution region TL1 that is one region of interest is determined. Next, the low resolution region TL2 adjacent thereto is set as a region of interest, the representative value upper limit APmax and the representative value lower limit APmin are obtained in the same manner as described above using the previously obtained representative value AP1, and the median value is used as the representative value. Let it be AP2. Such processing is repeated to sequentially obtain the representative value AP, and the representative value AP is obtained for all regions.

  For the low resolution region TL3 in FIG. 10, since the normal representative value AP was not obtained, the representative value AP2 in the low resolution region TL2 is corrected to the representative value upper limit APmax, and an appropriate representative value AP3 is set. A selected example is shown. In this case, reconstruction using the corrected representative value AP2 and representative value AP3 provides reconstructed image data FS as shown in the figure. The reconstructed image data FS for the low resolution region TL3 does not match the binary image data D1. That is, the values of the fourth and fifth pixels TP are opposite to each other. However, even this level of error may be acceptable in reproducibility.

  When the image data in the attention area or the small area is used as the alternative representative value APD, the difference between the representative value APD (image data) and the original binary image data D1 is taken, and the difference is used as the representative value AP. Also good.

Further, since the representative value AP is a value related to the image density, the relationship between the adjacent areas is deep and the representative values AP are close to each other in the adjacent areas. That is, there is a property that the density difference is small in the neighboring area, and therefore the difference in the representative value AP is small. Accordingly, when the difference between the representative value AP of a certain low resolution region TL and the representative value AP of the low resolution region TL obtained immediately before is taken, the difference is “0” in many cases, and the statistical frequency of the difference The bias becomes larger. That is, by using the difference between the representative values AP in place of the representative value AP itself, it is possible to perform compression at a high compression rate in the Huffman method encoding. In other words, it is desirable to select such a representative value a within a possible range so that the difference is as close to 0 as possible.
[Typical values for 2D]
By the way, in the above description, the method for obtaining the representative value AP has been described on the assumption that the divided low-resolution area TL is one-dimensional. However, in practice, the low resolution region TL is often two-dimensional as shown in FIGS. For example, the low resolution region TL may be 6 × 16 pixels or 4 × 16 pixels (in the case of binary) or 4 × 4 pixels (in the case of 16 values) as described above. As described above, when the low-resolution area TL is two-dimensional, the “straight line” in the interpolation rule HR in the one-dimensional case may be replaced with a “plane”. Next, a brief description will be given with reference to FIGS.

  First, as shown in FIGS. 9A and 9B, first representative values AP00, 10, 01 are determined. Since the first representative values AP00, 10, 01 are outside the area of the binary image data D1, any value can be taken, but the density value for the pixel TP closest to the representative value position AP in the low resolution area TL11. It is preferable to set the intermediate value of the range of image data calculated based on (halftone data) and the threshold value of the corresponding dither pattern DP.

  Then, the low resolution area TL11 which is the first block is set as the attention area, and the attention area is divided into two triangles PA1 and PA2 as shown in FIG. In the area of one triangle PA1, attention is paid to white pixels in the area. Of the first representative values AP00,10,01, a plane that passes through two representative value positions AP00,10 close to the apex of the triangle PA1 and passes through a threshold value corresponding to a white pixel in the attention area is obtained. This plane corresponds to the straight line LG in FIG.

  The plane at the lowest position among these planes is selected, the plane is extended, and the intersection with the position of the last pixel TP in the region of interest is set as the upper limit candidate position. The upper limit candidate position is obtained in the same manner for the area of the other triangle PA2. The smaller of the two obtained upper limit candidate positions is set as the representative value upper limit APmax in the attention area.

  Processing similar to that for white pixels is also performed for black pixels, and a representative value lower limit APmin in the region of interest is obtained. In FIG. 9B, TP indicates a position corresponding to each pixel.

  When the representative value upper limit APmax thus determined is larger than the representative value lower limit APmin, that is, when the above equation (1) holds, a value between the representative value upper limit APmax and the representative value lower limit APmin, for example, the median value thereof. Is determined as the representative value AP11 of the low resolution region TL11.

  By using the representative value AP11 determined in this way, by using the triangles PA1 and PA2 defined by the representative values AP00 and 10 or the representative values AP00 and 01 of the adjacent areas, and the dither pattern DP, The binary image data D1 in the low resolution region TL11 can be completely reproduced.

  If the above equation (1) does not hold, the previous representative values AP00, 10, 01 are corrected. Using the corrected representative value AP, the representative value AP11 of the attention area is obtained again as described above.

  If the representative value AP is still not obtained, a separate process is performed after recording that the representative value AP cannot be determined. As a separate process, as in the one-dimensional case, for example, the attention area is divided into half or quarter small areas, and the above processing is performed in each small area to determine the representative value APS. .

  Further, when the representative value AP cannot be obtained in the small region, the representative value upper limit APmax and the representative value lower limit APmin are appropriately determined, and an intermediate value within the range is determined as the representative value AP.

  Further, the processing in the case where the representative value APS cannot be determined in the small region may be considered in the same manner as in the one-dimensional case.

  In this way, the representative value AP11 of the low resolution region TL11 that is one region of interest is determined. Next, using the low resolution region TL12 or TL21 adjacent thereto as a region of interest, the representative value AP12 or the representative value AP21 is obtained by the same processing as above using the representative value AP11 obtained previously. Such processing is repeated to sequentially obtain the representative value AP, and the representative value AP is obtained for all regions.

  Although the method for obtaining the representative value AP in the case of two dimensions has been described above, in addition to this method, an image reconstructed from the representative value AP may be a slope inclined only in one direction. Good. That is, in this method, a straight line connecting pixels in a one-dimensional case is referred to as a plane connecting pixels. In this case, the area is divided in two directions when the representative value AP is not obtained. According to this method, the processing is simple and can be performed at high speed, but the compression rate is inferior to the method described above.

  In the prior application method, when the dot arrangement pattern of the image data in the attention area matches the dither pattern DP, the representative value a is obtained from the density of the image data in that area. This is a dot pattern obtained when the assumed image data is binarized by the dither pattern DP, assuming that all the pixels in the attention area are uniform image data having a certain density value. However, when it becomes equal to the dot pattern of the attention area, the dot pattern of the attention area coincides with the dither pattern DP. Any value may be used as the density value used for such determination. In other words, it can be said that when there is a density value such that the dot patterns are equal to each other when the dot pattern is generated using the dither pattern DP, it is determined that they match. Therefore, such a density value may be a specific value or may have a plurality of values in a certain range. This density value is the representative value a in the prior application method.

  The representative value a in this embodiment and the representative value a in the prior application method are common in that the dot pattern of the low resolution region TL can be restored from the dither pattern DP if the representative value a is known. That is, it is possible to reproduce the binary image data D1 of the area by using the dither pattern DP by obtaining only one representative value a in each area as the image information component.

  Therefore, by transmitting this representative value a to the printer 13, the printer 13 can reproduce the dot pattern of the binary image data D1 using the same appropriate dither pattern DP as when the binary image data D1 was obtained. is there.

  The interpolation rule HR in this embodiment is to interpolate with a straight line or a plane defined by a plurality of representative values a as described above, but in the prior application method, a straight line or a plane used for the interpolation is used. Can be viewed as a horizontal or flat line or plane with one representative value a as the height position.

  If the image data FD has 256 gradations for each color, the representative value a has an 8-bit information amount. In the present embodiment, the low resolution area TL has an information amount of 6 × 16 = 96 bits, and this can be expressed by 8 bits. Therefore, the information amount can be compressed to about 1/10. In addition, by using the difference of the representative value a, it is possible to further compress to, for example, a fraction in encoding.

  By the way, since the range of values that the representative value a can take is 0 to 255, when a simple subtraction is performed, the range of the difference is −255 to +255, and the amount of information necessary to express this is 9 A bit. However, in the present embodiment, the difference is not obtained by simple subtraction, but the difference is obtained assuming that the representative value a is a cyclic integer of 0 to 255. As a result, all the differences are represented by integers from 0 to 255, and the amount of information necessary to represent this is 8 bits, which is reduced to half.

  Next, a method for obtaining the difference assuming that it is a cyclic integer is described. In order to simplify the description, a cyclic integer of 0 to 8 will be described.

  FIG. 11 is a diagram showing a cyclic integer of 0 to 8.

  In FIG. 11, in order to obtain a difference C between a certain number A and a certain number B, the position number corresponding to the number B is rotated leftward from the position indicated by the number A in the cyclic integer. The numerical value of the stopped position is defined as a difference C. For example, the difference between “5” and “3” is “2” obtained by rotating three rotations from the position of the cyclic integer “5” to the left. In another example, the difference between “0” and “8” is “1”, which is obtained by rotating eight leftward from the position of the cyclic integer “0”. In yet another example, the difference between “3” and “5” is “7”, which is obtained by rotating left by five from the position of the cyclic integer “3”. In this way, the difference is obtained as an integer value in the range of 0-8.

  Further, a certain number A is obtained by adding the difference C to the certain number B. In this case, the number of positions corresponding to the difference C is rotated to the right from the position indicated by the number B having the cyclic integer, and the numerical value of the stopped position is set to the number A. In this way, it can be completely restored.

  FIG. 12 is a diagram showing the distribution of the appearance frequency of the difference of the representative value a in the image.

  As shown in FIG. 12, the appearance frequency is high in the range where the difference is about “0” to “20” and about “230” to “255”. Such a distribution is almost the same regardless of the type of image, and the deviation is very large, so that most images can be greatly compressed by the Huffman method by using the difference.

  Depending on the image, one low resolution area TL can be compressed to about 1 to 2 bits. The probability that the dot pattern of the 6 × 16 pixel low-resolution area TL matches the dither pattern DP is about 95% in most images. That is, in an area of approximately 95% of the entire image, data originally 96 bits can be expressed by 1 to 2 bits, and compression can be performed at a compression rate as high as several tens of times at this point.

  In FIG. 12, the appearance frequency for three values “256”, “257”, and “258” is shown after 0 to 255, which is the difference of the representative value a. This is to indicate the state with a specific three numerical values when a predetermined state occurs, thereby further reducing the amount of information and further increasing the compression rate.

  For example, since pixels with the same density frequently continue in the image for a long time, as described later, when the same density continues over 50 regions, that is, when the difference “0” continues 50 times, it is set to “257”. It is expressed by one attribute code. “256” is an attribute code indicating that the representative value a could not be determined in a certain low resolution region TL, and “258” indicates that the dither pattern DP was switched, that is, switched in a certain low resolution region TL. This is an attribute code indicating that the representative value a is determined by another dither pattern DP. When switched twice, “258” is output twice.

  As described above, in the low resolution representative value calculation unit 103, the representative value a is obtained in order for the low resolution region TL, and a difference between adjacent regions is obtained. Data to be sent to the encoding unit 108 is created based on the difference and the attribute value.

For example, when “0” continues as the difference 50 times, the attribute value “257” is output to the encoding unit 108. The value of the difference “0” is not output. If the difference “0” has not been repeated 50 times, “0” is output for the number that has been continued until then, and then a difference that is not “0” is output. When the dither pattern DP is switched, the attribute code “258” is output for the number of times of switching. If the representative value a is not determined, the attribute code “256” is output. In this case, the low resolution area TL is processed by the medium resolution area dividing unit 104 described below.
[Processing of medium resolution area and high resolution area]
13 is a diagram showing the low resolution region TL and the medium resolution region TM, FIG. 14 is a diagram for explaining the digitization of the dot pattern, and FIG. 15 is a diagram showing the distribution of the appearance frequency of each dot pattern.

  In the medium resolution area dividing unit 104, the low resolution area TL is further divided into four medium resolution areas TM of 3 × 8 pixels as shown in FIG.

  The medium resolution representative value calculation unit 106 obtains the representative value a in the medium resolution region TM by the same process as described above. Then, based on the obtained representative value, the difference in the representative value between the adjacent medium resolution areas TM is calculated in the same manner as in the case of the low resolution area TL described above. The difference here is in the range of 0 to 255 because the cyclic integer described above is used.

  When the processing of the four medium resolution areas TM for one low resolution area TL is completed, the process returns to the processing for the low resolution area TL again.

  When the representative value a is not found in the medium resolution area TM, the dot pattern extraction unit 107 further divides the medium resolution area TM into three high resolution areas TS of 8 pixels. Digitize the dot pattern. In the digitization of the dot pattern, for example, as shown in FIG. 14, the doted pixel is represented by a binary number “1”, and the non-printed pixel is represented by “0”, which is converted into a decimal number. This is output as an attribute code.

  As shown in FIG. 15, it can be seen that the distribution is considerably biased and concentrated on a specific value. Accordingly, even in this case, a high compression rate can be obtained by encoding using the Huffman method. For example, a compression ratio of about twice can be obtained.

  When the processing of the three high resolution areas TS for one medium resolution area TM is completed, the process returns to the process for the medium resolution area TM again.

The encoding unit 108 encodes input data, that is, a difference in the representative value a, a difference in the number of dots, an attribute code, and the like by the Huffman method. The compression by the Huffman method is very effective for an object whose signal frequency distribution does not change greatly. Since the Huffman method itself is publicly known, detailed description thereof is omitted here.
[Explanation of Huffman code table]
In this embodiment, the difference and attribute code of the representative value a for the low resolution area TL, the difference and attribute code of the number of dots for the medium resolution area TM, and the attribute code for the high resolution area TS are respectively independent codes. 1 to 3 are encoded. Therefore, three dedicated Huffman code tables FH1 to FH3 are provided for each.

  FIG. 16 is a diagram showing an example of the Huffman code tables FH1 to FH3. FIG. 16A shows the Huffman code table FH1 for code 1 for the low resolution area, FIG. 16B shows the Huffman code table FH2 for code 2 for the medium resolution area, and FIG. 16C for the high resolution area. A Huffman code table FH3 for code 3 is shown. These will be briefly described.

  In FIG. 16A, the Huffman code table FH1 includes items such as an input, an output code, and the number of output digits. The relationship between the input and the output code is determined in advance by encoding by the Huffman method based on the appearance frequency of each data. The output code is actually stored in the memory as, for example, 4 bytes of data, and a code that is substantially less than 4 bytes is padded with “0” in the upper digit.

  For the input data, the data of the term of the output code corresponding to the input term corresponding to the input data is read from the least significant digit by the number of digits indicated by the term of the number of output digits. For example, when the input is “3”, “1000100” which is 7 digits of the output code “. The read “1000100” becomes the compressed data D2.

  Similarly, the Huffman code tables FH2 and 3 shown in FIGS. 16B and 16C are also provided with items such as input, output code, and number of output digits. The relationship between these input and output codes is determined in advance by encoding by the Huffman method for the three Huffman code tables FH1 to FH3.

  For example, in the Huffman code table FH2, when the input is “1”, “1010” that is four digits of the output code “. In the Huffman code table FH3, when the input is “1”, “1010”, which is the four digits of the output code “.

  Although these Huffman code tables FH1 to 3 may be converted into the same code, compression is performed on which code area corresponds to which code 1 to 3 is converted. Since it can be recognized, the decoding table DC corresponding to it is selected, and complete restoration is possible.

  For example, in encoding, code 1 is encoded using the Huffman code table FH1 for the low resolution area until “256” appears in the data. When “256” appears, after “256” is encoded with code 1, it is switched to the Huffman code table FH2 for the medium resolution area, and the code of code 2 is encoded until encoding of the four medium resolution areas TM is completed. Will continue. If “25” appears during encoding of code 2, “25” is encoded with code 2 and then switched to the Huffman code table FH3 for the high-resolution area, and the three high-resolution areas TS The encoding of code 3 is continued until the encoding of is completed. When the encoding of the three high resolution areas TS is completed, the process returns to the encoding of the code 2, and when the encoding of the four medium resolution areas TM is completed, the process returns to the encoding of the code 1 again. Thereafter, the input data is sequentially encoded according to this rule, and the compressed data D2 is created.

  Further, the compressed data creation unit 100 creates decode tables DC1 to DC3 for restoring the compressed data D2 encoded by the Huffman code tables FH1 to FH3.

  The data transmission unit 109 transmits the compressed data D2 created by the encoding unit 108 to the printer 13. In addition, the data transmission unit 109 transmits all the dither patterns DP and the decode table DC to the printer 13 in advance.

The decoding table DC is created based on the appearance frequency of data. The appearance frequency of each data can also be obtained by actually executing the algorithm in this embodiment. For example, as such a frequency table, a table created by summing all the frequencies obtained for the dither patterns DP for the character area, the image area, and the chart area for each color of CMYK can be used. . Ideally, the frequency table should be obtained for each image to be compressed, but even if it is not done, sufficient compression performance can be obtained on the basis of the frequency table created as a sum as described above. For the same reason, the same frequency table created as described above can be used for each region such as a character region, an image region, and a chart region for each color of CMYK.
[Restore processing]
Next, restoration processing in the printer 13 will be described.

  4, the printer 13 includes a dither pattern storage unit 360, a data restoration unit 300, a decode table storage unit 370, a line buffer 380, and the like.

  The dither pattern storage unit 360 holds the same dither pattern DP stored in the dither pattern storage unit 160 of the computer main body 11. The dither pattern DP is transmitted from the computer main body 11 in advance, but the same dither pattern DP stored in an appropriate memory may be built in the printer 13 in advance.

  The decode table storage unit 370 stores the decode table DC transmitted from the computer 11.

  The data restoration unit 300 refers to a predetermined dither pattern DP, decode table DC, and interpolation rule HR, restores the compressed data D2 received from the computer 11, and reproduces a dot pattern. At the time of decompression of the compressed data D2, first, the decompression process by the decode 1 is performed using the decode table DC1 for the low resolution area, and the decompression process by the decode 2 using the decode table DC2 for the medium resolution area according to the conditions. Alternatively, the restoration process by the decode 3 using the decode table DC3 for the high resolution area is performed. That is, the compressed data D2 encoded by the codes 1, 2, and 3 are restored by the decodes 1, 2, and 3, respectively. In the data restoration unit 300, the same interpolation rule HR as the interpolation rule HR used in the compressed data creation unit 100 is provided as a processing circuit or a program for executing the same.

  FIG. 17 is a diagram showing an example of the decode table DC2.

  In FIG. 17, the decode table DC2 has items such as the number of symbols, Root, child 0, child 1 and the like. Based on the input data, the original data can be obtained by sequentially tracing the table from the position of Root = 24. For example, if the input data is “011011”, the top of the data is “0”, so the process starts with the child 0 of Root = 24. Since the value of the child 0 of Root = 24 is “23” and the next value of the data is “1”, next, the child 1 of Root = 23 is seen. Since the value of the child 1 of Root = 23 is “22” and the next value of the data is “1”, the child 1 of Root = 22 is seen next. Thereafter, the same operation is repeated, and the process ends when the value of child 0 or child 1 becomes negative. The sum of the last obtained value (negative value) and the number of symbols (in this case “26”) is the original data. Thereafter, the same processing is performed from the next value of the data, and it is repeated until the data expansion is completed.

  In the reproduction of the dot pattern, the restoration process is first performed by the decode 1. When the value restored by the decoding table DC1 is “258”, it indicates that the dither pattern DP has been switched. When a plurality of “258” continues, the dither pattern DP is sequentially switched as many times as the number of times followed, and the dither pattern DP after the switching is set as a dither pattern (current dither pattern) DP to be used thereafter.

  When the restored value is “256”, the subsequent processing is switched to the restoration processing by decoding 2 using the decoding table DC2 for the medium resolution area. If the restored value is “257”, it means that “0” has continued 50 times for the difference between the representative values, and therefore, the low-resolution area TL previously calculated for 50 areas by the current dither pattern DP. A dot pattern is reproduced by using the same value as the representative value of, and sent to the line buffer 380. When the end of the line is reached before the processing for 50 areas is completed, the rest is sent to the line buffer 380 as data of the next line. If the restored value is none of these, the value is a difference between representative values. In this case, the representative value a of the low resolution region TL calculated last time is set as an initial value, and the current representative value a is obtained by adding the difference to the initial value. In this manner, the representative values a of all the low resolution areas TL are obtained by accumulating the differences. If this is the first time, the initial value is set to “0” to obtain the representative value a. Then, a dot pattern is reproduced from the obtained representative value a, current dither pattern DP, and interpolation rule HR, and sent to the line buffer 380.

  In the restoration process by the decode 2, the restoration is performed by the decode table DC2 for the medium resolution area. When the value restored by the decode 2 is “256”, the subsequent processing is switched to the restoration processing of the decode 3 for the high resolution area. When the restored value is not “256”, the value is the difference between the representative values of the medium resolution area TM. In this case, the representative value is obtained by accumulation as in the case of the low resolution region. From the obtained representative value and the current dither pattern DP, the dot pattern of the medium resolution region TM is reproduced. When the processing for the four medium resolution areas TM is completed, the process returns to the decode 1.

  In the restoration process by the decode 3, the restoration is performed by the decode table DC3 for the high resolution area. The result restored by the decode 3 represents an array of dots for 8 pixels. The restoration for the three high resolution areas TS is repeated, thereby reproducing the dot pattern of one medium resolution area TM. Thereby, it is counted that one restoration process of the medium resolution area TM is completed, and the process returns to the decoding 2 of the calling process.

  As described above, the attribute code “256” functions as a code switching signal between areas, and when these attribute codes appear, it is possible to switch from the code 1 to the code 2 or from the code 2 to the code 3 without fail.

In the line buffer 380, the binary image data sent from the data restoration unit 300 is temporarily stored. The binary image data stored in the line buffer 380 is output every six lines, for example, and printed out on paper.
[Explanation by flowchart]
Next, processing in the computer main body 11 and the printer 13 will be described with reference to flowcharts.

  FIG. 17 is a flowchart showing a general flow of the compression processing in the computer main body 11, and FIG. 18 is a flowchart showing a general flow of the decompression processing in the printer 13.

  In FIG. 17, after reading the dither pattern DP as a preparation (# 11), the binary image data D1 is read (# 12). The binary image data D1 is divided into low resolution areas TL (# 13), and it is determined whether or not there is a range (representative value range) that can be selected as a representative value for one low resolution area TL ( # 14).

  In step # 14, as described above, the representative value upper limit APmax and the representative value lower limit APmin in the attention area are obtained based on the representative value AP of the area adjacent to the attention area and the threshold value of the dither pattern DP. From these, it is determined whether or not a representative value range exists, and if it exists, the representative value range is acquired.

  When there are a plurality of dither patterns DP, the dither pattern DP is switched until a representative value range is found. If the representative value range is found (Yes in # 14), the representative value a is obtained within the representative value range (# 15). In this case, for example, the median value of the representative value range is determined as the representative value a. Then, the difference between the representative value a determined here and the previously obtained representative value a is taken (# 16) and encoded (# 17).

  If there is no representative value range in step # 14, it is divided into medium resolution areas TM (# 19), and it is determined whether or not there is a representative value range for one medium resolution area TM (# 20). If there is a representative value range, the number of dots is obtained (# 21), the difference is taken (# 22), and encoding is performed (# 23). If the representative value range is not found in step # 20, it is partitioned into high-resolution areas TS (# 25), and a dot array is obtained for one high-resolution area TS (# 26) and encoded (##). 27).

  Such a process is performed until all of the read binary image data D1 is completed (# 18, 24, 28).

  In FIG. 19, the dither pattern DP and the decode table DC are read as preparation (# 31, 32). The compressed data D2 is read (# 33), the representative value a is obtained (# 34), the density of each pixel is calculated using the interpolation rule HR, and the dot pattern is obtained by referring to the dither pattern DP (# 35). 36). Note that the description of the restoration processing of the medium resolution area TM and the high resolution area TS is omitted.

  Next, it demonstrates using a more detailed flowchart.

  20 and 21 are flowcharts of the compression process in the compressed data creation unit 100, FIG. 22 is a flowchart showing an example of the representative value range determination process, and FIGS. 23 and 24 are flowcharts of the decompression process in the data restoration unit 300. Note that these flowcharts show processing for one color component in CMYK.

  20 and 21, first, the dither pattern DP and the decode table DC are read (# 101, 102). Further, the dither pattern DP and the decode table DC are transmitted to the printer 13 in advance. Input 6 lines of the binary image data D1 (# 104), and use the adjacent representative value a for the one low-resolution area TL from the left end to determine whether there is a representative value range in that area. It is determined whether or not (# 105). When there is a representative value range, a representative value a is determined within the representative value range (# 106), and a representative value difference is calculated (# 107).

  If the representative value difference is “0” (Yes in # 108), one variable n indicating the number of consecutive times is added (# 109). When the variable n becomes “50” (Yes in # 110), the attribute code “257” indicating that the representative value difference “0” has continued 50 times is encoded with the code 1 and output as the compressed data D2. (# 111), variable n is initialized to “0” (# 112).

  If no in step # 105, that is, if there is no representative value range, the representative value difference “0” is encoded and output by the number of times up to that time with code 1 (# 120), and the variable n is set to “0”. Initialization is performed (# 121). Then, it is checked whether or not there is a representative value range in the case of another dither pattern DP (# 122). If there is a dither pattern DP in which the representative value range can be obtained (Yes in # 122), switching to that dither pattern DP (# 123), the attribute code "258" indicating that the dither pattern DP has been switched is switched by code 1 The partial encoding is performed (# 124), the representative value a is determined (# 225), and the process proceeds to step # 107.

  If NO in step # 108, that is, if the representative value difference is not "0" (NO in # 108), the representative value difference "0" is encoded with code 1 for the number of times so far and output (# 126). The variable n is initialized to “0” (# 127). The representative value difference is encoded with code 1 and output (# 128), and the process returns to step # 113.

  If the processing has not been completed up to the right end of the inputted 6 lines (No in # 113), step # 105 and the subsequent steps are repeated in order to sequentially process the low resolution region TL on the right for the same 6 lines. When the processing for 6 lines is completed (Yes in # 113), if there is an unprocessed line (No in # 114), the next 6 lines of the binary image data D1 are input (# 104). The process is repeated until the entire processing of the value image data D1 is completed (# 114).

  If NO in step # 122, that is, if there is no dither pattern DP from which the representative value range is obtained, the attribute code “256” indicating that the low resolution area TL has not obtained the representative value range in any dither pattern DP is displayed. The code 1 is encoded and output (# 129), and the process proceeds to step # 131 and subsequent steps, which are the processing of the medium resolution area TM.

  In the process of the medium resolution area TM, it is determined whether or not a representative value range can be obtained with any dither pattern DP for one medium resolution area TM obtained by dividing the low resolution area TL into four (# 131). When the representative value range is obtained (Yes in # 131), the number of dots in the medium resolution area TM is acquired and the difference is calculated (# 132), and the dot number difference is encoded with code 2 and output ( # 133). This is repeated four times to return to the processing of the low resolution area TL in step # 113.

  If NO in step # 131, that is, if the representative value range is not obtained for any dither pattern DP in the medium resolution area TM, the attribute code “25” indicating that it was not obtained is encoded with code 2 and output. However, the process proceeds to step # 141 and subsequent steps, which are the processing of the high resolution area TS.

  In the processing of the high-resolution area TS, the dot pattern is digitized for one high-resolution area TS obtained by dividing the medium-resolution area TM into three (# 141), and the numerical value is encoded with code 3 and output (#). 142). This is repeated three times (# 143), and the process returns to the process of the medium resolution area TM in step # 134.

  In FIG. 22, the representative value a of the area adjacent to one attention area is acquired (# 151). A temporary representative value in the attention area is set (# 152). The temporary representative value may be set with reference to the binary image data D1 and the dither pattern DP, or a gradation value that is a range of values that the representative value a can take, in the above example, a value from 0 to 255. It may be set in order. Using the representative value and the temporary representative value of the adjacent region, the interpolation rule HR is applied to determine the density of each pixel in the region of interest, and a virtual gradation image is generated (# 153). The generated virtual gradation image is converted into pseudo gradation using the dither pattern DP, and a reconstructed image (reconstructed image data FS in the above example) is generated (# 154). The reconstructed image is compared with the original pseudo gradation image (binary image data D1 in the above example) (# 155), and it is determined from the degree of coincidence whether the reproducibility is good (# 156). ).

  When determining whether or not the reproducibility is good, for example, the dot arrangement pattern of the reconstructed image matches the dot arrangement pattern of the binary image data D1, or falls within a predetermined error range. Judge whether or not. You may judge that reproducibility is favorable only when it corresponds. Alternatively, it may be determined that the reproducibility is good when it is within a predetermined error range. As the predetermined error range, for example, the ratio of the number of pixels that do not match in the region of interest to the total number of pixels or the number of pixels that do not match can be used as a threshold value for determination. The threshold value of the ratio of the mismatched pixels to the total number of pixels is, for example, about 15 to 50% for a photographic image and about 5 to 15% for a character image.

  If the reproducibility is good (Yes in # 156), the temporary representative value is put in the representative value range (# 157). If the reproducibility is not good (No in # 156), the temporary representative value is discarded.

  When the processing for all scheduled temporary representative values is completed (Yes in # 158), the representative value range obtained so far is determined (# 159). Thereafter, for example, in step # 106 of FIG. 20, one representative value a is determined from the representative value range. Note that the processing shown in FIG. 23 is merely an example of processing for determining the representative value range, and various other processing methods can be applied.

  23 and 24, first, the dither pattern DP and the decode table DC are read (# 201, 202). Then, the compressed data D2 transmitted from the computer main body 11 is received, and processing for restoration is performed in order from the head. Initially, it is restored by decode 1 (# 203).

  When the value dc1 obtained by the restoration is not the attribute codes “258”, “256”, and “257” (No in # 204 to 206), the value dc1 is the difference of the representative value a, and thus the cumulative value k1 so far The value dc1 is added to the value, and the obtained value is set as a new cumulative value k1 (# 206). If the cumulative value k1 exceeds “255” (Yes in # 208), since it exceeds the maximum value of the representative value a, “256” is subtracted from the cumulative value k1 (# 209). Then, with the cumulative value k1 as the representative value a, a predetermined interpolation rule HR is applied to determine the density value of each pixel, and the dot pattern is restored using the dither pattern DP specified by the pattern number pn (# 210). . The dot pattern is transferred to the line buffer (# 211), and 1 is added to the area number nb (# 212). The area number nb is a number indicating the position in the 6 lines of the low resolution area TL to be processed.

  Step # 203 and subsequent steps are repeated until the development for the same six lines is completed (No in # 213). When the development for 6 lines is completed (Yes in # 213), the area number nb is initialized to "0" (# 214), and the developed data for 6 lines (binary image data) is printed out (# 215). The process is repeated until all the compressed data D2 is processed (# 216).

  If yes in step # 204, that is, if the value dc1 obtained by restoration is “258”, “1” is added to the pattern number pn (# 221). If the pattern number pn is “3” (Yes in # 222), “3” is subtracted from the pattern number pn. That is, the pattern number pn is any one of “0”, “1”, and “2”, and this specifies one of the three types of dither patterns DP.

  If YES in step # 206, that is, if the value dc1 is “257”, the density value of each pixel is obtained by applying the interpolation rule HR using the cumulative value k1 as the representative value a, and is specified by the pattern number pn. The dot pattern is restored using the dither pattern DP (# 231). The dot pattern is transferred to the line buffer (# 232), and 1 is added to the area number nb (# 233). This is repeated 50 times (# 234). Accordingly, the binary image data of 50 low resolution areas TL are reproduced based on the representative value a when the difference “0” continues 50 times. When the right end of the line is reached during this process, the process is temporarily interrupted, the process from step # 214 is performed, and a new dot pattern is written from the beginning of the next line.

  If YES in step # 205, that is, if the value dc1 is "256", the process proceeds to step # 251 which is the process for the medium resolution area TM. In step # 251, the data is restored by the decode 2. If the value dc2 obtained by decoding 2 is not “25” (No in # 252), the value dc2 is the difference in the number of dots, so the value dc2 is added to the cumulative value k2 so far, and the obtained value Is set as a new accumulated value k2 (# 253). If the cumulative value k2 exceeds “24” (Yes in # 254), the maximum value of the number of dots is exceeded, so “25” is subtracted from the cumulative value k2 (# 255). Then, using the accumulated value k2 as the number of dots, the dot pattern is reproduced using the dither pattern DP specified by the pattern number pn (# 256). This is repeated four times (# 257). The dot pattern for four times is transferred to the line buffer (# 258), and 1 is added to the area number nb (# 259).

  If YES in step # 252, that is, if the value dc2 is "25", the process proceeds to step # 261, which is processing of the high resolution area TS. In step # 261, the data is restored by decoding 3 (# 261). From the restored value, an 8-dot pattern of the high-resolution area TS is reproduced (# 262). This is repeated three times (# 263).

  As described above, during the compression process, the binary image data D1 is separated into a dither pattern information component and an image information component. Then, only the image information component is extracted and subjected to encoding processing by the Huffman method to obtain compressed data D2. Since the dither pattern information component is not included in the compressed data D2, a high compression rate and high speed processing can be realized.

  Further, in the processing of the high resolution region TS, since the image data of 8 pixels is read with one numerical value, the address instruction is fast. This also shortens the overall processing time.

  In the printer 13, the cost of the ASIC used for restoring the compressed data D2 is greatly influenced by the size of the work area of the on-chip memory. As a work area for performing the data expansion process, a storage area for the read dither pattern DP and decode table DC and a data storage area for the processing target area are required. For example, the dither pattern DP is about several hundred bytes to several kilobytes, the decoding table DC is about 2 kilobytes, and the total amount of data is about 3 to 4 kilobytes including other buffers. is there. Therefore, by providing the printer 13 with a small ASIC of about 3 to 4 Kbytes, the decompression process of the compressed data D2 can be performed at high speed. Further, the capacity can be further reduced by storing the decode table DC in the ROM. In this manner, the printer 13 can expand the compressed data D2 at high speed with small hardware. Thus, the entire process can be performed in a short time while using an inexpensive printer 13.

  In the embodiment described above, the low-resolution area dividing unit 101 shown in FIG. 3 divides the binary image data D1 into the low-resolution area TL. Various methods are used for the dividing method. Is possible. That is, since the partition may be virtual, it is not necessary to actually divide the binary image data D1 for each region. For example, the low resolution comparison unit 102 may specify an address for indicating a position to be compared. The configuration of the compressed data creation unit 100 shown in FIG. 3 is an example, and the functions described above can be realized by various configurations.

  In the embodiment described above, the computer main body 11 is provided with the pseudo gradation unit 170, and the binarization processing of the image data FD is performed here, but the binarization processing is performed inside the computer main body 11. Alternatively, the binary image data D1 may be input from the outside. In this case, information such as the dither pattern DP used for the binarization process may be acquired.

In the above-described embodiment, the binary image data D1 can express whether or not a dot is applied to each pixel constituting the image without a flag. Dots may be applied to pixels with high density, or dots may be applied to pixels with low density.
[Second]
Embodiment for Compressing 4-Level Image Data Next, a case where the pseudo gradation image is multi-level 4-level image data will be described. This case is basically the same as the case of compressing binary image data. Therefore, the description and drawings in the first can be applied to the second in substantially the same manner, and for many parts, the “two values” in the first description are changed to “four values” in the second. It can be applied by replacing. Only the main differences will be described here.

FIG. 25 is a diagram for explaining the partitioned low-resolution area TL, FIG. 26 is a diagram showing an example of the low-resolution area TL and the dither pattern DP, and FIG. 27 shows four-value image data (four-value image data) D5. FIG. 28 is a diagram for explaining binary image data (binary image data) D1.
[Explanation of quaternary image data]
In the example of FIG. 25, the entire area of the quaternary image data D5 is partitioned from the upper left to TLm, such as low resolution areas TL1, TL2, TL3,. In FIG. 26, one low-resolution area TL is a rectangular area composed of 48 (6 × 8) pixels TP.

  In the low resolution region TL shown in FIG. 26A, the black pixel is a pixel on which a dot is formed. The size of the dot changes in three steps according to the pixel value. In other words, if the state where dots are not applied is “0”, the state where dots are applied to about one third of the area of one pixel is “1”, and dots are applied to about two thirds of the area. The state in which the dots are applied to the entire area of one pixel can be represented as “3”.

  As shown in FIG. 27, in correspondence with each state “0”, “1”, “2”, “3”, for example, by controlling the laser light emission time (pulse width control) of the printer, the pixel density is set in four levels. Can be changed. Thus, these states “0”, “1”, “2”, and “3” correspond to the gradation values.

  That is, in the state “0”, the laser does not emit light, and the pixel is “white”. In the state “1”, the laser light is emitted for about one third of the pixel width, and the region of about one third of the pixel is “black”. Similarly, in state “2” or “3”, the laser light is emitted for a period of about two-thirds or about three-thirds of the pixel width, respectively, and about two-thirds of the pixel area or the entire region is exposed. “Black”. That is, a density of 4 gradations is obtained corresponding to the state value. Four states representing four gradations can be represented by 2 bits. Therefore, four pixels of four gradations can be represented by 8 bits.

  The relationship between the laser emission time, the size of the dots hit by the laser, and the density at which it is viewed macroscopically is complex, and the above example is an example for explanation. Only.

  The threshold value of the dither pattern DP is a value in the range of 0 to 255, and three threshold values are set for each pixel. In each pixel, indexes (numbers) “1”, “2”, and “3” are given to the three threshold values in ascending order. The image data FD having a density smaller than the lowest threshold value with the index “1” is in the state “0”. The image data FD having a density between the threshold value of the index “1” and the threshold value of the index “2” is in the state “1”. Image data FD having a density between the threshold value of the index “2” and the threshold value of the index “3” is in the state “2”. The image data FD having a density larger than the highest threshold value whose index is “3” is in the state “3”. Usually, different pixels have different thresholds for the same index.

  In FIG. 27, the virtual gradation image data LA is indicated by a substantially horizontal straight line. The virtual gradation image data LA is used in place of the original image data FD when reconstructing the pseudo gradation image, and is reproduced using the representative value a.

  In the prior application method, the virtual gradation image data LA is the representative value a itself in the area.

  Here, binary image data is also described for comparison.

  As shown in FIG. 28, in the binary image data, the pixel density can be changed in two stages of black and white by controlling on / off the laser of the printer, for example, corresponding to each state “0” “1”. .

  That is, in the state “0”, the laser does not emit light, and the pixel is “white”. In the state “1”, laser light is emitted, and the pixel is “black”. That is, a density of two gradations is obtained corresponding to the state value.

Also in FIG. 28, the virtual gradation image data LA similar to the case of FIG. 27 is shown by a substantially horizontal straight line.
[How to find representative value a]
FIG. 29 is a diagram for explaining how to obtain the representative value a. As in the first case, it is assumed here that the low resolution region TL is a one-dimensional region.

  The representative value a can be obtained by almost the same processing as in the case of binary even in the case of multivalue. However, in the case of multi-value, there is a difference in that there is an upper limit and a lower limit for each pixel TP depending on the image data of the pseudo gradation image.

  In FIG. 29, a one-dimensional low-resolution region TL1 composed of four pixels TP is shown for the four-value image data D5. The gradation values of the respective pixels TP in the figure are “1”, “2”, “3”, and “3” in order from the left pixel TP. This quaternary image data D5 is obtained by subjecting the image data FD shown in the figure to a quaternary process using the dither pattern DP shown in the figure. That is, for each index, if the image data FD is less than or equal to the threshold value of the dither pattern DP, “white” is obtained, and if the image data FD exceeds the threshold value, “black” is obtained. Determined.

  First, the first representative value AP0 is determined. Since the first representative value AP0 is outside the region of the quaternary image data D5, any value can be taken, but it may be an appropriate value as in the case of the binary value.

  Then, the low resolution region TL1, which is the first block, is set as the attention region, and straight lines LG1, 2, 3, 4 are drawn from the first representative value AP0 toward the threshold value of each white index of the attention region. The straight line LG4 at the lowest position is selected, the straight line LG4 is extended, and the intersection with the position of the last pixel TP of the target area is set as the upper limit position, and the density value at this upper limit position is set as the representative value upper limit APmax in the target area. And

  The same processing as for the white index is performed for the black index. That is, straight lines LG5 and LG6 are drawn from the first representative value AP0 toward the threshold value of each black index in the region of interest. The straight line LG5 at the highest position is selected, the straight line LG5 is extended, and the intersection with the position of the last pixel TP of the target area is set as the lower limit position. The density value at this lower limit position is set as the representative value lower limit APmin in the target area. And

When the representative value upper limit APmax thus obtained is larger than the representative value lower limit APmin, that is, when the above equation (1) is satisfied, the range between the representative value upper limit APmax and the representative value lower limit APmin is set as the representative value range. Then, a value within the representative value range is determined as the representative value a of the attention area.
[Low resolution area, medium resolution area, high resolution area]
FIG. 30 is a diagram showing a low resolution area TL, a medium resolution area TM, and a high resolution area TS.

As shown in FIG. 30, the low resolution area TL is an area of 6 × 8 pixels. The low resolution area TL is divided into four medium resolution areas TM of 3 × 4 pixels by the medium resolution area dividing section 104. The medium resolution area TM is further divided into three high resolution areas TS of 4 pixels as necessary.
[Medium resolution area processing]
FIG. 31 is a diagram showing the distribution of the appearance frequency of the difference in the number of dots in an image.

  The medium resolution representative value calculation unit 106 obtains the representative value a in the medium resolution region TM by the same process as described above.

  Then, based on the obtained representative value a, the difference between the representative values between the adjacent medium resolution areas TM is calculated in the same manner as in the low resolution area TL. The difference here is in the range of 0 to 255 using the cyclic integer described above.

  As shown in FIG. 31, the appearance frequency is high in the range where the difference is about “0” to “30” and “200” to “255”. After the difference of 0 to 255, the appearance frequency for the value “256” is shown. The meaning of this is the same as the attribute code in the case of the low resolution area TL, and “256” is an attribute code indicating that the representative value a cannot be determined in a certain medium resolution area TM. In this case, “256” is output with the code 2, and the region is further divided into high-resolution regions TS.

In the case of the binary value shown in the first example, the medium resolution area TM has 24 pixels, and the maximum number of dots is 24. Therefore, the number of dots in the area is counted as a representative value. However, in the case of the four values shown in the second example, priority is given to the sharing and the processing speed, and the density value is used as the representative value as described above.
[Explanation by flowchart]
Next, the second process will be described with reference to a flowchart.

  FIG. 32 is a flowchart showing a rough flow of compression processing in the computer main body 11. Note that FIG. 32 corresponds to FIG. 18 and only step # 21B is different.

  That is, in FIG. 32, when there is a representative value range for one medium resolution area TM (Yes in # 20), the representative value a is obtained in the same manner as in Step # 15 (# 21B). Then, the difference is taken (# 22) and encoded (# 23).

  33 and 34 are flowcharts of the compression processing in the compressed data creation unit 100, and FIGS. 35 and 36 are flowcharts of the decompression processing in the data restoration unit 300. 33 to 36 correspond to the first FIGS. 20, 21, 23, and 24.

  In FIG. 34, when the representative value range is obtained with any dither pattern DP (Yes in # 131), the representative value a is obtained in the medium resolution region TM and the difference is calculated (# 132B). The value difference is encoded with code 2 and output (# 133B). This is repeated four times, and the process returns to the processing of the low resolution region TL in step # 113 in FIG.

  If NO in step # 131, that is, if no representative value range is obtained for any dither pattern DP in the medium resolution area TM, the attribute code “256” indicating that it was not obtained is encoded with code 2 and output. (# 135B), the process proceeds to step # 141 and subsequent steps, which are the processing of the high resolution area TS.

  In FIG. 36, if the value dc2 obtained by decoding 2 is not “256” in step # 252B, the value dc2 is the difference between the representative values, so the value dc2 is added to the cumulative value k2 obtained so far. The obtained value is set as a new cumulative value k2 (# 253). If the cumulative value k2 exceeds “255” (Yes in # 254B), the maximum value of the representative value is exceeded, so “256” is subtracted from the cumulative value k2 (# 255B). Then, using the accumulated value k2 as a representative value, the dot pattern is reproduced using the dither pattern DP specified by the pattern number pn (# 256).

  If YES in step # 252B, that is, if the value dc2 is "256", the process proceeds to step # 261, which is processing of the high resolution area TS.

  In the decompression process, the dither process is performed using the dither pattern DP attached to the header portion of the compressed data D2.

  In the above embodiment, binarization and quaternarization have been performed as pseudo-gradation, but various values such as ternarization, 8-level conversion, and 16-level conversion can also be performed. . The representative value position AP in each region may be a position other than those described above, for example, the center of the region, another vertex of the region, or the like.

  In addition, the configuration, function, number, number of bits, processing contents or order of the entire computer main body 11, printer 13, and printing system 1 or each part can be appropriately changed in accordance with the spirit of the present invention.

  The present invention is used in a printing system including a raster printer, a GDI printer, and the like, and can be used as a method for compressing pseudo gradation image data.

1 is a diagram illustrating an overall configuration of a print system according to the present invention. It is a block diagram which shows the functional structure of a computer main body. It is a block diagram which shows the compressed data production part of a computer main body functionally. FIG. 2 is a block diagram illustrating a functional configuration of a printer. It is a figure for demonstrating the divided low resolution area | region. It is a figure which shows the example of a low-resolution area | region and a dither pattern. It is a figure which shows the example of the low resolution area | region and dither pattern from which size differs. It is a figure which shows the position of the representative value in each low-resolution area | region. It is a figure for demonstrating the example of the interpolation rule in a low-resolution area | region. It is a figure for demonstrating how to obtain | require a representative value. It is a figure which shows the example of a cyclic integer. It is a figure showing distribution of the appearance frequency of the difference of the representative value in an image. It is a figure which shows a low-resolution area | region and a medium-resolution area | region. It is a figure for demonstrating the digitization of a dot pattern. It is a figure showing distribution of the appearance frequency of each dot pattern. It is a figure which shows the example of a Huffman code table. It is a figure which shows the example of a decoding table. It is a flowchart which shows the rough flow of a compression process. It is a flowchart which shows the rough flow of a decompression | restoration process. It is a flowchart of the compression process in a compression data creation part. It is a flowchart of the compression process in a compression data creation part. It is a flowchart which shows the example of the determination process of a representative value range. It is a flowchart of the decompression | restoration process in a data decompression | restoration part. It is a flowchart of the decompression | restoration process in a data decompression | restoration part. It is a figure for demonstrating the divided low resolution area | region. It is a figure which shows the example of a low-resolution area | region and a dither pattern. It is a figure for demonstrating 4 value image data. It is a figure for demonstrating binary image data. It is a figure for demonstrating how to obtain | require a representative value. It is a figure which shows a low resolution area | region and a medium resolution area | region. It is a figure showing distribution of the appearance frequency of the difference of the number of dots in an image. It is a flowchart which shows the rough flow of a compression process. It is a flowchart of the compression process in a compression data creation part. It is a flowchart of the compression process in a compression data creation part. It is a flowchart of the decompression | restoration process in a data decompression | restoration part. It is a flowchart of the restoration process in a data restoration part.

Explanation of symbols

1 Print System 11 Computer Body (Image Data Compression Device)
13 Printer 100 Compressed data creation unit 101 Low resolution area partition unit (first means)
103 Low resolution representative value calculation unit (second means)
104 Medium resolution area section (means to divide into small areas)
106 Medium Resolution Representative Value Calculation Unit 107 Dot Pattern Extraction Unit 108 Encoding Unit (Third Means)
109 Data transmission unit 300 Data restoration unit D1 Binary image data (pseudo gradation image data, image data)
D2 Compressed data D5 Four-value image data (pseudo gradation image data, image data)
DP dither pattern TL Low resolution area (predetermined area)
TM Medium resolution area (predetermined area)
TS High resolution area (predetermined area)
a Representative value AP Representative value HR Interpolation rule AL Straight line (pseudo gradation image)
PA triangle (plane)
FS reconstruction image data (pseudo gradation image, image data obtained by performing pseudo gradation process)

Claims (11)

A method of compressing pseudo-gradation image data, which is a pseudo-gradation image using a dither pattern,
A first step of dividing the pseudo gradation image into predetermined regions;
A second step of determining a representative value for each region;
And a third step of creating compressed data using the determined representative value,
In the second step,
For a region of interest that is one of the regions, a representative value of the region of interest and a representative value of one or a plurality of regions adjacent to the region of interest are used according to a predetermined interpolation rule. Such a representative value is obtained by interpolating the density of each pixel and further reproducing the pseudo gradation image in the region of interest when pseudo gradation is performed using the dither pattern. Determine as the representative value of the region of interest,
A method for compressing image data. A method of compressing pseudo-gradation image data, which is a pseudo-gradation image using a dither pattern,
A first step of dividing the pseudo gradation image into predetermined regions;
A second step of determining a representative value for each region;
And a third step of creating compressed data using the determined representative value,
In the second step,
For a region of interest that is one of the regions, a representative value of the region of interest and a representative value of one or a plurality of regions adjacent to the region of interest are used according to a predetermined interpolation rule. Interpolation of the density of each pixel and the pseudo-gradation using the dither pattern, and the dot arrangement pattern of the image data obtained by performing pseudo gradation processing and the dot arrangement pattern of the image data in the region of interest have a predetermined error. Find such a representative value that matches within the range and determine it as the representative value of the region of interest,
A method for compressing image data. In the second step, the representative value is set to a value corresponding to a density value of a pixel located at an end of each region.
The image data compression method according to claim 1 or 2. In the second step, when a value that can be selected as the representative value has a predetermined range, a median value in the range is determined as the representative value.
The image data compression method according to claim 1. The interpolation rule interpolates the density of each pixel by a plane defined by a total of three representative values of the representative value of the region of interest and the representative values of two regions adjacent to the region of interest.
5. A method for compressing image data according to claim 1. In the second step, when a representative value cannot be determined in the attention area, image data in the attention area is determined as an alternative representative value of the attention area.
6. A method for compressing image data according to claim 1. In the second step, when a representative value cannot be determined in the attention area, the attention area is further divided into a plurality of small areas, and a representative value is determined in each small area.
6. A method for compressing image data according to claim 1. In the second step, when the representative value cannot be determined in the small area, the image data in the small area is determined as an alternative representative value of the small area.
The image data compression method according to claim 7. A method for decompressing data compressed by the method according to claim 1, comprising:
A representative value of each area is obtained from the compressed data, and for the attention area that is one of the areas, the representative value of the attention area and the representative value of one or more areas adjacent to the attention area Using a predetermined interpolation rule to interpolate the density of each pixel in the region of interest to obtain a pseudo-gradation using the dither pattern to restore a pseudo-gradation image,
A pseudo gradation image restoration method characterized by the above. An apparatus for compressing pseudo gradation image data, which is an image that has been pseudo gradation using a dither pattern,
First means for dividing the pseudo gradation image into predetermined regions;
A second means for determining a representative value of each region;
And a third means for creating compressed data using the determined representative value,
The second means includes
For a region of interest that is one of the regions, a representative value of the region of interest and a representative value of one or a plurality of regions adjacent to the region of interest are used according to a predetermined interpolation rule. Such a representative value is obtained by interpolating the density of each pixel and further reproducing the pseudo gradation image in the region of interest when pseudo gradation is performed using the dither pattern. Determine as the representative value of the region of interest,
A compression apparatus for image data. In the case where the representative value cannot be determined in the attention area in the second means, the second means further includes means for dividing the attention area into a plurality of small areas, and the second means determines the representative value in each small area. decide,
The image data compression apparatus according to claim 10.