Image coding and decoding using mapping coefficients corresponding to class information of pixel blocks

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• means for extracting a plurality of pixels from the orginal image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• means for storing mapping coefficients for each of the plurality of classes;
• means for reading the mapping coefficients that correspond to the generated class information;
• and
• means for producing the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• means for extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• means for storing mapping coefficients for each of the plurality of classes;
• means for reading the mapping coefficients that correspond to the generated class information;
• and
• means for producing the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• means for extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• means for storing mapping coefficients for each of the plurality of classes;
• means for reading the mapping coefficients that correspond to the generated class information;
• and
• means for producing the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated by steps of:
• extracting a plurality of pixels of the original image data for learning and generating class information corresponding to characteristics of the extracted plurality of pixels;
• producing coded data where a number of pixels of the original image data for learning is reduced, responsive to the original image data for learning and mapping coefficients that correspond to class information;
• generating predicted image data for learning in accordance with the coded data and generating predicted data having a plurality of predicted pixels;
• generating a predicted error of the predicted image data for learning relative to the original image data for learning;
• updating the mapping coefficients in accordance with the generated predicted error until the updated mapping coefficients are optimum mapping coefficients;
• and
• determining the optimum mapping coefficients.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• means for extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• means for storing mapping coefficients for each of the plurality of classes;
• means for reading the mapping coefficients that correspond to the generated class information;
• and
• means for producing the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping for each class is determined by
• reducing a number of pixels of the original image data to generate compressed data;
• correcting the compressed data to generate corrected data;
• predicting the original image data in accordance with the corrected data and generating predicted data having a plurality of predicted pixels;
• calculating, with respect to the original image data, a predicted error of the predicted data;
• determining suitability of the corrected data as coded data based upon the predicted error;
• repeating the correcting step until the corrected data is an optimum corrected data;
• generating optimum mapping coefficients for each class using the original image data for learning and the optimum corrected data.

An apparatus for decoding coded image data generated by coding original image data, comprising:

• means for receiving the coded image data;
• and
• means for decoding the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients.

An apparatus for decoding coded image data generated by coding original image data, comprising:

• means for receiving the coded image data;
• and
• means for decoding the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients,
• wherein the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

An apparatus for decoding coded image data generated by coding original image data, comprising:

• means for receiving the coded image data;
• and
• means for decoding the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

A method for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients.

A method for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

A method for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

A method for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated by steps of:
• extracting a plurality of pixels of the original image data for learning and generating class information corresponding to characteristics of the extracted plurality of pixels;
• producing coded data where a number of pixels of the original image data for learning is reduced, responsive to the original image data for learning and mapping coefficients that correspond to class information;
• generating predicted image data for learning in accordance with the coded data and generating predicted data having a plurality of predicted pixels;
• generating a predicted error of the predicted image data for learning relative to the original image data for learning;
• updating the mapping coefficients in accordance with the generated predicted error until the updated mapping coefficients are optimum mapping coefficients;
• and
• determining the optimum mapping coefficients.

A method for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping for each class is determined by
• reducing a number of pixels of the original image data to generate compressed data;
• correcting the compressed data to generate corrected data;
• predicting the original image data in accordance with the corrected data and generating predicted data having a plurality of predicted pixels;
• calculating, with respect to the original image data, a predicted error of the predicted data;
• determining suitability of the corrected data as coded data based upon the predicted error;
• repeating the correcting step until the corrected data is an optimum corrected data;
• generating optimum mapping coefficients for each class using the original image data for learning and the optimum corrected data.

A method for decoding coded image data generated by coding original image data, comprising:

• receiving the coded image data;
• and
• decoding the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients.

A method for decoding coded image data generated by coding original image data, comprising:

• receiving the coded image data;
• and
• decoding the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes:
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

A method for decoding coded image data generated by coding original image data, comprising:

• receiving the coded image data;
• and
• decoding the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

A method for transmitting image data, having a plurality of pixels, comprising:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• transmitting the coded image data.

A method for transmitting image data, having a plurality of pixels, comprising;

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• transmitting the coded image data, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

A method for transmitting image data, having a plurality of pixels, comprising:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• transmitting the coded image data, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

A method for transmitting image data, having a plurality of pixels, comprising:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• transmitting the coded image data, wherein
• the mapping coefficients for each class are generated by steps of:
• extracting a plurality of pixels of the original image data for learning and generating class information corresponding to characteristics of the extracted plurality of pixels;
• producing coded data where a number of pixels of the original image data for learning is reduced, responsive to the original image data for learning and mapping coefficients that correspond to class information;
• generating predicted image data for learning in accordance with the coded data and generating predicted data having a plurality of predicted pixels;
• generating a predicted error of the predicted image data for learning relative to the original image data for learning;
• updating the mapping coefficients in accordance with the generated predicted error until the updated mapping coefficients are optimum mapping coefficients;
• and
• determining the optimum mapping coefficients.

A method for transmitting image data, having a plurality of pixels, comprising:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• transmitting the coded image data, wherein the mapping for each class is determined by
• reducing a number of pixels of the original image data to generate compressed data;
• correcting the compressed data to generate corrected data;
• predicting the original image data in accordance with the corrected data and generating predicted data having a plurality of predicted pixels;
• calculating, with respect to the original image data, a predicted error of the predicted data;
• determining suitability of the corrected data as coded data based upon the predicted error;
• repeating the correcting step until the corrected data is an optimum corrected data;
• generating optimum mapping coefficients for each class using the original image data for learning and the optimum corrected data.

An article of manufacture having recorded thereon coded image data, the article of manufacture produced by the following steps of:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data.

An article of manufacture having recorded thereon coded image data, the article of manufacture produced by the following steps of:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

An article of manufacture having recorded thereon coded image data, the article of manufacture produced by the following steps of:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

An article of manufacture having recorded thereon coded image data, the article of manufacture produced by the following steps of:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein
• the mapping coefficients for each class are generated by steps of:
• extracting a plurality of pixels of the original image data for learning and generating class information corresponding to characteristics of the extracted plurality of pixels;
• producing coded data where a number of pixels of the original image data for learning is reduced, responsive to the original image data for learning and mapping coefficients that correspond to class information;
• generating predicted image data for learning in accordance with the coded data and generating predicted data having a plurality of predicted pixels;
• generating a predicted error of the predicted image data for learning relative to the original image data for learning;
• updating the mapping coefficients in accordance with the generated predicted error until the updated mapping coefficients are optimum mapping coefficients;
• and
• determining the optimum mapping coefficients.

An article of manufacture having recorded thereon coded image data, the article of manufacture produced by the following steps of:

• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein the mapping for each class is determined by
• reducing a number of pixels of the original image data to generate compressed data;
• correcting the compressed data to generate corrected data;
• predicting the original image data in accordance with the corrected data and generating predicted data having a plurality of predicted pixels;
• calculating, with respect to the original image data, a predicted error of the predicted data;
• determining suitability of the corrected data as coded data based upon the predicted error;
• repeating the correcting step until the corrected data is an optimum corrected data;
• generating optimum mapping coefficients for each class using the original image data for learning and the optimum corrected data.

A method of transmitting coded image data, comprising:

• receiving the coded image data;
• and
• transmitting the coded image data,
• wherein the coded image data is produced by the following steps of:
• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data.

A method of transmitting coded image data, comprising:

• receiving the coded image data;
• and
• transmitting the coded image data,
• wherein the coded image data is produced by the following steps of:
• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

A method of transmitting coded image data, comprising:

• receiving the coded image data;
• and
• transmitting the coded image data,
• wherein the coded image data is produced by the following steps of:
• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

A method of transmitting coded image data, comprising:

• receiving the coded image data;
• and
• transmitting the coded image data,
• wherein the coded image data is produced by the following steps of:
• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein
• the mapping coefficients for each class are generated by steps of:
• extracting a plurality of pixels of the original image data for learning and generating class information corresponding to characteristics of the extracted plurality of pixels;
• producing coded data where a number of pixels of the original image data for learning is reduced, responsive to the original image data for learning and mapping coefficients that correspond to class information;
• generating predicted image data for learning in accordance with the coded data and generating predicted data having a plurality of predicted pixels;
• generating a predicted error of the predicted image data for learning relative to the original image data for learning;
• updating the mapping coefficients in accordance with the generated predicted error until the updated mapping coefficients are optimum mapping coefficients;
• and
• determining the optimum mapping coefficients.

A method of transmitting coded image data, comprising:

• receiving the coded image data;
• and
• transmitting the coded image data,
• wherein the coded image data is produced by the following steps of:
• extracting a plurality of pixels from original image data and generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing coded image data responsive to the original image data and the read mapping coefficients;
• and
• recording the coded image data, wherein the mapping for each class is determined by
• reducing a number of pixels of the original image data to generate compressed data;
• correcting the compressed data to generate corrected data;
• predicting the original image data in accordance with the corrected data and generating predicted data having a plurality of predicted pixels;
• calculating, with respect to the original image data, a predicted error of the predicted data;
• determining suitability of the corrected data as coded data based upon the predicted error;
• repeating the correcting step until the corrected data is an optimum corrected data;
• generating optimum mapping coefficients for each class using the original image data for learning and the optimum corrected data.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• an extracting circuit that extracts a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• a storing circuit that stores mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• a reading circuit that reads the mapping coefficients that correspond to the generated class information;
• and
• a coding circuit that produces the coded image data responsive to the original image data and the read mapping coefficients.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• an extracting circuit that extracts a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• a storing circuit that stores mapping coefficients for each of the plurality of classes;
• a reading circuit that reads the mapping coefficients that correspond to the generated class information;
• and
• a coding circuit that produces the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• an extracting circuit that extracts a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• a storing circuit that stores mapping coefficients for each of the plurality of classes;
• a reading circuit that reads the mapping coefficients that correspond to the generated class information;
• and
• a coding circuit that produces the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• an extracting circuit that extracts a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• a storing circuit that stores mapping coefficients for each of the plurality of classes;
• a reading circuit that reads the mapping coefficients that correspond to the generated class information;
• and
• a coding circuit that produces the coded image data responsive to the original image data and the read mapping coefficients, wherein
• the mapping coefficients for each class are generated by steps of:
• extracting a plurality of pixels of the original image data for learning and generating class information corresponding to characteristics of the extracted plurality of pixels;
• producing coded data where a number of pixels of the original image data for learning is reduced, responsive to the original image data for learning and mapping coefficients that correspond to class information;
• generating predicted image data for learning in accordance with the coded data and generating predicted data having a plurality of predicted pixels;
• generating a predicted error of the predicted image data for learning relative to the original image data for learning;
• updating the mapping coefficients in accordance with the generated predicted error until the updated mapping coefficients are optimum mapping coefficients;
• and
• determining the optimum mapping coefficients.

An apparatus for coding original image data, having a plurality of pixels, to generate coded image data, comprising:

• an extracting circuit that extracts a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• a storing circuit that stores mapping coefficients for each of the plurality of classes;
• a reading circuit that reads the mapping coefficients that correspond to the generated class information;
• and
• a coding circuit that produces the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping for each class is determined by
• reducing a number of pixels of the original image data to generate compressed data;
• correcting the compressed data to generate corrected data;
• predicting the original image data in accordance with the corrected data and generating predicted data having a plurality of predicted pixels;
• calculating, with respect to the original image data, a predicted error of the predicted data;
• determining suitability of the corrected data as coded data based upon the predicted error;
• repeating the correcting step until the corrected data is an optimum corrected data;
• generating optimum mapping coefficients for each class using the original image data for learning and the optimum corrected data.

An apparatus for decoding coded image data generated by coding original image data, comprising:

• a receiving circuit that receives the coded image data;
• and
• a decoding circuit that decodes the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes, wherein the mapping coefficients are generated based on a predicted error of predicted image data for learning, and an original image data for learning;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients.

An apparatus for decoding coded image data generated by coding original image data, comprising:

• a receiving circuit that receives the coded image data;
• and
• a decoding circuit that decodes the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is minimized.

An apparatus for decoding coded image data generated by coding original image data, comprising:

• a receiving circuit that receives the coded image data;
• and
• a decoding circuit that decodes the coded image data and producing decoded image data,
• wherein said coded image data is generated by the steps of:
• extracting a plurality of pixels from the original image data and for generating class information corresponding to characteristics of the extracted plurality of pixels, wherein the generated class information indicates one of a plurality of classes;
• storing mapping coefficients for each of the plurality of classes;
• reading the mapping coefficients that correspond to the generated class information;
• and
• producing the coded image data responsive to the original image data and the read mapping coefficients, wherein the mapping coefficients for each class are generated such that a predicted error of predicted image data for learning, determined from a difference between decoding coded data generated from the original image data for learning, and the original image data for learning, is less than a prescribed threshold value.