JPH10108202A - Method for compressing image data and print data or the like - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain the inexpensive data compression method for a high resolution image and print image data having high image quality at high speed. SOLUTION: Since the difference between adjacent pixels of high resolution image data or the like is small with high probability probabilistically, after the data are smoothed and quantized for the purpose, the difference in the scanning direction 3 is taken as 4 and 5 in the figure, data are coded with an encoding method at a high compression rate that is completely reconstructable and an image within a range of error in smoothing or quantizing the original image is compressed and reconstructed at high efficiency.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[Industrial application] High-resolution image high-speed compression for storing various image data. Create and store low-resolution display of high-resolution data and data for printing devices. Image transmission device. High-resolution image high-speed compression software for computers. High-resolution image high-speed compression device for computers. printer. TV set.
Display device. facsimile.

[0002]

2. Description of the Related Art JPFG and MPEG are available.

[0003]

SUMMARY OF THE INVENTION High-speed, high-quality, inexpensive data compression of high-resolution image and print image data.

[0004]

In order to compress data for images and printing, data is taken as it is at an element point group at an arbitrary determined place in a data array in a predetermined arbitrary scanning direction. Calculate the difference between adjacent data in the scanning direction between the element point groups, and write the difference data as EntropyCod.
Encoding is performed by an encoding method such as ing, Huffman Coding, or an arithmetic encoding method that can be completely restored when the encoding is restored. However, the element point group includes the case where it is the first point in the data array in the scanning direction. For data compression for images and printing, data is taken as it is at the element point group at an arbitrary position in the data array for any given scanning direction, and data adjacent to each other in the scanning direction Calculate the difference between the two, and use the difference code as the Entropy Code
An arbitrary predetermined identifier for identification is added to a data point group which is coded by a coding method which can be completely restored at the time of restoring the coding such as ing, Huffman Coding, arithmetic coding and the like and uses raw data. . However, the element point group includes the case where it is the first point in the data array in the scanning direction. Compression of data for images and printing is performed by dividing the screen or data array into an arbitrary number determined by an arbitrary size determined in a fixed place, and using each arbitrary divided screen or divided data array. The data at the beginning of the scanning direction is taken as it is for a given arbitrary scanning direction, and thereafter, the difference between each adjacent data in the scanning direction is calculated, and the difference data is calculated by Entropy Coding or Huffman Co.
Encoding is performed by an encoding method that can be completely restored when the encoding is restored, such as ding and arithmetic encoding. Compression of data for images and printing is performed by dividing the screen or data array into an arbitrary number determined by an arbitrary size determined in a fixed place, and using each arbitrary divided screen or divided data array. The data is taken as it is at the element point group at an arbitrary position in the determined arbitrary scanning direction, the difference between the adjacent data in the scanning direction is calculated between the element point groups, and the difference data is calculated as Entr.
opy Coding and Huffman Coding
Encoding using a coding method that can be completely restored when restoring the encoding, such as arithmetic coding and arithmetic coding, and if necessary, for a data point group using raw data, an arbitrary identifier for identification is used. Add. 4. The method according to claim 1, 2, 3, or 4, wherein before calculating a difference between data in the whole or divided data array, a moving average or the like is used except for a point group that takes raw data instead of a difference between boundary points. Become Claim 1,
Before calculating the difference between the data in the whole or divided data array in 2, 3, 4, and 5, smoothing using a moving average or the like, except for a point group that takes raw data instead of a difference between boundary points and the like, Quantization is performed on the whole or divided regions. According to the sixth aspect, when quantizing the entire or divided region, the quantization level is set so that the difference between adjacent data is not required in the stochastic sense as much as possible while maintaining the contrast. The quantization is finely set for the values necessary for the quantization, where the maximum and minimum neighbors, the neighborhood with a small frequency of the values, and the high possibility are set, and the others are set roughly. This method is used when giving priority to the compression ratio. Alternatively, in order to maintain the contrast of the quantization level as much as possible, the neighborhood of the maximum and the minimum of the value and the neighborhood of the high frequency of the value can be distinguished from each other. Set up easily. 9. The scanning method according to claim 1, 2, 3, 4, 5, 6, 7, or 8, wherein a difference between adjacent data is equal to or greater than a predetermined threshold value in order to maintain contrast. The point closer to is one of the points that take raw data, the quantization level is made as fine as possible with one point value, the accuracy is increased, and the deterioration of contrast due to smoothing and quantization is reduced. Let it.
The problem is solved by the above method and an apparatus having the above method.

[0005]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The compression of data for images and prints is performed by taking data as it is at an element point group at an arbitrary predetermined place in a data array in a predetermined arbitrary scanning direction. The difference between groups is calculated between adjacent data in the scanning direction, and the difference data is encoded by an encoding method that can be completely restored at the time of restoring the encoding, such as EntropyCoding, Huffman Coding, and arithmetic encoding. However, the element point group includes the case where it is the first point in the data array in the scanning direction. For data compression for images and printing, data is taken as it is at the element point group at an arbitrary position in the data array for any given scanning direction, and data adjacent to each other in the scanning direction Calculate the difference between them, and use the difference data as Entropy Coding or H
An arbitrary identifier determined for identification is added to a data point group which is coded by a coding method which can be completely restored at the time of restoring such encoding, such as uffman coding and arithmetic coding, and which uses raw data. However, the element point group includes the case where it is the first point in the data array in the scanning direction.
Compression of data for images and printing is performed by dividing the screen or data array into an arbitrary number determined by an arbitrary size determined in a fixed place, and using each arbitrary divided screen or divided data array. The first data in the scanning direction is taken as it is with respect to the determined arbitrary scanning direction, and thereafter, the difference between each adjacent data in the scanning direction is calculated, and the difference data is obtained by Entro.
Encoding is performed by an encoding method that can be completely restored when the encoding is restored, such as py coding, Huffman coding, and arithmetic coding. Compression of data for images and printing is performed by dividing the screen or data array into an arbitrary number determined by an arbitrary size determined in a fixed place, and using each arbitrary divided screen or divided data array. The data is taken as it is at the element point group at an arbitrary position in the determined arbitrary scanning direction, the difference between the adjacent data in the scanning direction is calculated between the element point groups, and the difference data is entropy.
Coding, Huffman coding, arithmetic coding, and various other codings, such as coding, can be completely restored at the time of restoration of the coding. If necessary, the data point group using raw data is determined for identification. Add an arbitrary identifier. 4. The method according to claim 1, 2, 3, or 4, wherein before calculating a difference between data in the whole or divided data array, a moving average or the like is used except for a point group that takes raw data instead of a difference between boundary points. Become Claims 1, 2, 3,
In steps 4 and 5, before calculating a difference between data in the whole or divided data array, smoothing is performed using a moving average or the like except for a point group that takes raw data instead of a difference between boundary points or the like, and the whole or divided area is calculated. Quantize with According to the sixth aspect, when quantizing the entire or divided region, the quantization level is set so that the difference between adjacent data is not required in the stochastic sense as much as possible while maintaining the contrast. The quantization is finely set for the values necessary for the quantization, where the maximum and minimum neighbors, the neighborhood with a small frequency of the values, and the high possibility are set, and the others are set roughly. This method is used when giving priority to the compression ratio. Alternatively, in order to maintain the contrast of the quantization level as much as possible, the neighborhood of the maximum and the minimum of the value and the neighborhood of the high frequency of the value can be distinguished from each other. Set up easily. Claims 1, 2,
In 3, 4, 5, 6, 7, and 8, in order to maintain the contrast, if the difference between adjacent data is equal to or more than a predetermined threshold value, the raw data closer to the origin in the scanning direction of the two points is used as the raw data. And the quantization level is made as fine as possible with the value of one of the points, the accuracy is increased, and the deterioration of contrast due to smoothing and quantization is reduced.

[0006]

According to the present invention, since the difference between adjacent pixels is small with a high probability in a stochastic sense, the present invention uses
Smooth the original image and quantize it if necessary by taking the difference in the scanning direction after smoothing and quantizing, which is appropriate for the application, and performing the encoding with high compression ratio that can completely restore. An image in the range of the error can be compressed with high efficiency and further restored.

[Brief description of the drawings]

FIG. 1 Quantization method

FIG. 2 Difference between adjacent data

[Description of sign]

 1 Fine quantization. 2 New quantization. 3 Scan direction. 4 points. 5 points 2.

────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] October 1, 1996

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] Full text

[Correction method] Change

[Correction contents]

[Document Name] Statement

Patent application title: Method for compressing data for images and printing

[Claims]

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[Industrial application] High-resolution image high-speed compression for storing various image data. Create and store low-resolution display of high-resolution data and data for printing devices. Image transmission device. High-resolution image high-speed compression software for computers. High-resolution image high-speed compression device for computers. printer. TV set. Display device. facsimile. Image abstraction, contour extraction, feature extraction.

[0002]

2. Description of the Related Art JPEG and MPEG are known.

[0003]

SUMMARY OF THE INVENTION High-speed, high-quality, inexpensive data compression of high-resolution image and print image data.

[0004]

SUMMARY OF THE INVENTION Data compression for images and printing is performed by the following method. A first method is to use these data points as raw data element points for the determined arbitrary scanning direction 3 and use the data values themselves at arbitrary determined locations in the data array. Between the element point groups, the difference between each adjacent data is calculated along the scanning direction 3, for example, as at points 4 and 5 in FIG.
ropy Coding and Huffman Codin
g, Run Length Coding, arithmetic coding, and other coding methods that can be completely restored when the coding is restored. However, the raw data element point group includes the case where the raw data element point group is the first point of the data array in the scanning direction 3.
The second method is to use these data points themselves as raw data element points for the determined arbitrary scanning direction 3, and use these data points themselves as raw data element points. The interval between the adjacent data is calculated along the scanning direction 3, for example, at points 4 and 5 in FIG. 2, and the difference data is calculated by Entropy Coding or Huffman.
For the raw data element point group using raw data, which is coded by various coding methods which can be completely restored at the time of decoding the coding, such as coding, run length coding, arithmetic coding, etc., an arbitrary predetermined identification is used. How to add an identifier. However, the raw data element point group includes the case where the raw data element point group is the first point of the data array in the scanning direction 3. The third method is to divide the screen or data array into an arbitrary number determined by an arbitrary size determined in a fixed place, and to perform any scanning determined by each arbitrary divided screen or divided data array The first data value itself in each split data array is used for direction 3 and then the difference between each adjacent data is calculated along scan direction 3 as at points 4 and 5 in FIG. , Difference data by Entropy Coding or Hu
ffman Coding and Run Length C
A coding method using various coding methods, such as coding and arithmetic coding, which can be completely restored when the coding is restored. The fourth method is to divide the screen or data array into an arbitrary number determined by an arbitrary size determined in a fixed place, and to perform an arbitrary scan determined by each arbitrary divided screen or divided data array. In the direction 3, the data value itself at an arbitrary position in each divided data array is used, and these data points are regarded as raw data element points.
Then, the difference between each adjacent data is calculated, for example, as shown at points 4 and 5 in FIG.
Ding, Huffman Coding, Run Le
Encoding is performed by various encoding methods that can be completely restored when the encoding is restored, such as ngth coding or arithmetic encoding method. If necessary, a raw data element point group using raw data is determined for identification. How to add arbitrary identifiers. A fifth method is that, in each of the above methods, before calculating a difference between data in the whole or divided data array, raw data element points using data values themselves instead of differences such as boundary points of the divided data array. A method that incorporates smoothing processing using a moving average or the like, excluding groups. A sixth method is that, in each of the above methods, before calculating a difference between data in the whole or divided data array, raw data element points using data values themselves instead of differences such as boundary points of the divided data array. A method that incorporates processing for smoothing using a moving average or the like, excluding groups, and quantizing the whole or divided regions. Further, here, the quantization in each of the above-mentioned methods includes the following method.
The seventh method is that, when performing quantization in the whole or divided region in the quantization in each of the above methods, the setting of the quantization level is
In order to maintain the contrast and minimize the difference between adjacent data in the stochastic sense as much as possible,
A method in which the quantization is finely set in a value region where quantization is necessary for the vicinity where the value is the maximum and minimum, the value where the value is infrequent, and the value is small, and the other regions are roughly set. If necessary, use this method when priority is given to the compression ratio. In the eighth method, when performing quantization in the whole or divided regions in the quantization in each of the above-described methods, the setting of the quantization level is performed.
In the data area where fine and high-precision changes are required, in order to make the data smooth, the neighborhood of the maximum and minimum of the value and the neighborhood of the high frequency of the value and the high frequency can be divided. How to set roughly in the area. If necessary, this method can be used globally (Glo
This is used when priority is given to improving accuracy (for bal). As a method of reducing the deterioration of the contrast in each of the above methods, a method of incorporating the following method in each of the first to eighth methods is adopted. As the ninth method, if the difference between adjacent data is equal to or greater than the determined threshold, the element closer to the origin in the scanning direction of the two points is used as the element of the raw data element point group using the data value itself, If necessary, it is excluded from the smoothing processing, and the quantization level is reduced as much as possible near the value of one point, or the value is set to one quantization level, and quantization is performed with high accuracy, or A method that incorporates a method of excluding from quantization processing using the data value itself as an element of the data element point group. By combining these methods, data sledding, contour extraction, and feature extraction are performed. The problem is solved by the above method and an apparatus having the above method.

[0005]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Data compression for images and prints is performed by the following method. A first method is to use these data points as raw data element points for the determined arbitrary scanning direction 3 and use the data values themselves at arbitrary determined locations in the data array. Between the element point groups, the difference between each adjacent data is calculated along the scanning direction 3, for example, at points 4 and 5 in FIG.
Coding and Huffman Coding and Ru
A coding method using various coding methods such as n Length Coding and arithmetic coding, which can be completely restored when the coding is restored. However, the raw data element point group includes the case where the raw data element point group is the first point of the data array in the scanning direction 3. The second method is to use these data points themselves as raw data element points for the determined arbitrary scanning direction 3, and use these data points themselves as raw data element points. In the interval, the difference between each adjacent data is calculated along the scanning direction 3, for example, as shown by points 4 and 5 in FIG.
Tropy Coding and Huffman Codi
ng, Run Length Coding, arithmetic coding, etc., for the raw data element point group using raw data, which is coded by various coding methods that can be completely restored at the time of decoding of the coding, an arbitrary predetermined identification is used. How to add an identifier. However, the raw data element point group includes the case where the raw data element point group is the first point of the data array in the scanning direction 3. The third method is to divide the screen or data array into an arbitrary number determined by an arbitrary size determined in a fixed place, and to perform any scanning determined by each arbitrary divided screen or divided data array For the direction 3, the value of the first data in each divided data array itself is used, and thereafter, for example, the point 4 in FIG.
The difference between each adjacent data is calculated as in steps 5 and 5, and the difference data is calculated by Entropy Coding or Huffman.
Coding and Run Length Coding
A method of encoding using various encoding methods that can be completely restored when the encoding is restored, such as the arithmetic encoding method and the arithmetic encoding method. The fourth method is
The screen or data array is divided into an arbitrary number determined by an arbitrary size determined in a fixed place, and an arbitrary scanning direction 3 determined by each arbitrary divided screen or the divided data array
On the other hand, these data points are used as raw data element points by using the data values themselves at arbitrary positions in each divided data array. Calculate the difference between each adjacent data, as at points 4 and 5, and use the difference data as Entropy Coding, Huffman Coding, or Run Length.
Encoding is performed using various encoding methods that can be completely restored when the encoding is restored, such as coding or arithmetic encoding, and if necessary, a raw data element point group using raw data is determined for identification. How to add an arbitrary identifier. A fifth method is that, in each of the above methods, before calculating a difference between data in the whole or divided data array, raw data element points using data values themselves instead of differences such as boundary points of the divided data array. A method that incorporates smoothing processing using a moving average or the like, excluding groups. A sixth method is that, in each of the above methods, before calculating a difference between data in the whole or divided data array, raw data element points using data values themselves instead of differences such as boundary points of the divided data array. A method that incorporates processing for smoothing using a moving average or the like, excluding groups, and quantizing the whole or divided regions. Further, here, the quantization in each of the above-mentioned methods includes the following method. In the seventh method, when performing quantization in the whole or divided region in the quantization in each of the above methods, it is necessary to set a quantization level by maintaining a contrast and, as much as possible, by using a difference between adjacent data in a probabilistic sense. In order to make the nonexistent portion smaller smoothly, a method is used in which the quantization is finely set in a value area where quantization is necessary for the neighborhood of the maximum and minimum values and the neighborhood of a small value frequency and which can be highly divided. If necessary, use this method when priority is given to the compression ratio. In the eighth method, when performing quantization in the whole or divided regions in the quantization in each of the above methods, the quantization level is set, the contrast is maintained, and the difference between adjacent data in a probabilistic sense is reduced. In the data area where high-precision change is required, in order to make the data smooth, the neighborhood of the maximum and minimum of the value, the neighborhood of the high frequency of the value, and the value that can be highly divided can be finely quantized in the value area where quantization is required. Then how to set roughly. If necessary, this method is used when priority is given to global (global) accuracy improvement in the whole and divided regions. As a method of reducing the deterioration of the contrast in each of the above methods, a method of incorporating the following method in each of the first to eighth methods is adopted. As a ninth method, if the difference between adjacent data is equal to or greater than the determined threshold, the element closer to the origin in the scanning direction of the two points is used as the element of the raw data element point group using the data value itself, Exclude from the smoothing process if necessary,
The quantization level is made as small as possible in the vicinity of the value of one point, or the value is made one quantization level and quantized with high precision, or the data value itself is made an element of the raw data element point group. A method that incorporates a method that is used and excluded from quantization processing. By combining these methods, data abstraction, contour extraction, and feature extraction are performed.

[0006]

According to the present invention, since the difference between adjacent pixels is small with a high probability in a stochastic sense, high-resolution image data and the like are used according to the present invention.
Smooth the original image and quantize it if necessary by taking the difference in the scanning direction after smoothing and quantizing, which is appropriate for the application, and performing the encoding with high compression ratio that can completely restore. An image in the range of the error can be compressed with high efficiency and further restored.

[Brief description of the drawings]

FIG. 1 is an outline of a quantization method.

FIG. 2 is an outline of a difference between adjacent data.

[Description of Signs] 1 Fine quantization area. 2 A rough quantization area. 3 Scan direction. 4 points. 5 points 2.

[Procedure amendment 2]

[Document name to be amended] Drawing

[Correction target item name] All figures

[Correction method] Change

[Correction contents]

FIG.

FIG. 2

Claims (9)

[Claims] 1. A method for compressing data for an image, printing, and the like, by taking data as it is at an element point group at an arbitrary predetermined position in a data array in a predetermined arbitrary scanning direction, Calculates the difference between each adjacent data in the scanning direction, and calculates the difference data by EntropyCoding or Huf.
A method of encoding using various encoding methods such as fman coding and arithmetic encoding, which can be completely restored when the encoding is restored, and an apparatus using the same. However, the element point group includes the case where it is the first point in the data array in the scanning direction. 2. Compression of data for images and printing is performed by taking data as it is in an element point group at an arbitrary position in a data array with respect to a predetermined arbitrary scanning direction, and interpolating the scanning direction between the element point groups. The difference between each adjacent data is calculated, and the difference data is calculated by Entropy Coding or Huffman.
A method of adding an arbitrary predetermined identifier for identification to a data point group encoded using a coding method that can be completely restored at the time of restoring the encoding such as coding and arithmetic encoding method and using raw data, and the like. Equipment using. However, the element point group includes the case where it is the first point in the data array in the scanning direction. 3. Compression of data for images and printing is performed by dividing a screen or data array into an arbitrary number determined by an arbitrary size determined at a fixed place, and each arbitrary divided screen. Alternatively, the first data in the scanning direction is taken as it is for an arbitrary scanning direction determined by the divided data array, and thereafter, the difference between each adjacent data in the scanning direction is calculated, and the difference data is calculated by Entropy Coding or Huffman C.
An encoding method using an encoding method that can be completely restored at the time of restoring the encoding, such as coding and arithmetic encoding, and an apparatus using the encoding method. 4. Compression of data for images and printing is performed by dividing a screen or a data array into an arbitrary number determined by an arbitrary size determined at a fixed place, and each arbitrary divided screen. Alternatively, the data is taken as it is at an element point group at an arbitrary position in an arbitrary scanning direction determined by a divided data array, and a difference between each adjacent data in the scanning direction is calculated between the element point groups, and the difference data is calculated. Entropy Coding and Hu
ffman Coding, arithmetic coding, etc., are encoded by a coding method that can be completely restored at the time of decoding the coding, and if necessary, a data point group using raw data can be arbitrarily determined for identification. A method of adding an identifier and an apparatus using the method. 5. The method according to claim 1, wherein before calculating a difference between data in the whole or divided data array,
A method for smoothing using a moving average and the like, except for a point group that takes raw data instead of a difference between boundary points and the like, and an apparatus using the method. 6. A method according to claim 1, wherein, before calculating a difference between data in the whole or divided data array, a point group which takes raw data instead of a difference at a boundary point or the like is used. , A method of smoothing using a moving average or the like, and quantizing the whole or divided area, and an apparatus using the method. 7. In the sixth aspect, when quantizing the entire or divided area, the quantization level is set so that the difference between adjacent data is not required in a stochastic sense while maintaining contrast. For this reason, a method of setting the quantization finely for values necessary for quantization that can be distinguished from the maximum and minimum values of the values and the values with low frequency of values, and roughly setting the other values roughly, and an apparatus using the method. This method is used when giving priority to the compression ratio. 8. In the method according to claim 6, when quantizing the whole or divided areas, the setting of the quantization level is made such that the difference between adjacent data is not required in a stochastic sense as much as possible while maintaining the contrast. Therefore, a method for setting the quantization to be fine for the values required for quantization, where the maximum and minimum values of the value and the frequency of the value are high and for which the value can be distinguished, and an apparatus using the method. 9. The method according to claim 1, wherein the difference between adjacent data is equal to or greater than a predetermined threshold value in order to maintain the contrast. The one closer to the origin in the scanning direction of is taken as one of the point cloud that takes raw data,
A method of making the quantization level as fine as possible by using the value of one of the points, increasing the accuracy, and reducing the deterioration of contrast due to smoothing and quantization, and a device using this method.