Method and apparatus for coding and decoding digital image data using image quantization modification

• US 6,044,115 A
• Filed: 12/09/1997
• Issued: 03/28/2000
• Est. Priority Date: 12/13/1996
• Status: Expired due to Fees

• Current Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Original Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Inventors: Mizuguchi, Noboru, Horiike, Kazuyoshi
• Primary Examiner(s): Le, Vu
• Application Number:
  US08/987,243
• Time in Patent Office: 840 Days
• Field of Search: 348/404-405, 348/415-416, 348/419, 386/109, 386/111-112, 386/27, 386/33, 382/239, 382/251, 382/236, 382/238
• US Class Current: 375/240.05
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/14 Coding unit complexity, e.g. amount of activity or edge presence estimation H04N19/146 takes precedence
• CPC Class Codes: H04N 19/146 Data rate or code amount at the encoder output
• CPC Class Codes: H04N 19/152 by measuring the fullness of the transmission buffer
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/196 being specially adapted for the computation of encoding parameters, e.g. by averaging previously computed encoding parameters processing of motion vectors H04N19/513
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

An image coding method and apparatus for transforming and quantizing an image to produce transformed and quantized coefficients. The coefficients are coded and a complexity index value is obtained from the input image to calculate local characteristic values used to modify the quantization scale. In this manner, rate control is obtained on the basis of the local characteristics of the image, whereby the quality of the coded image and the precision of the rate control is improved.

Claims

1. An image coding apparatus that receives an input image per coding unit, which is obtained by dividing an original digital image into coding units, and codes the input image to obtain coded data, said apparatus comprising:

subtraction means for providing a difference image between the input image per coding unit and a reference image generated from a coded image which has already been coded;

image transform means for receiving, as an input image to be transformed, one of the input image per coding unit and the difference image, and performing a prescribed transformation process to the input image to be transformed, thereby to obtain a transformed image;

quantization means for quantizing the transformed image by one of a prescribed quantization scale and a modified quantization scale which is given from a quantization scale decision means, thereby to obtain transformed and quantized coefficients;

coding means for coding the transformed and quantized coefficients to obtain coded data, and outputting the coded data;

local characteristic value calculation means for obtaining a complexity index value from the input image to be transformed and, using the complexity index value, calculating local characteristic values which are used for modification of the prescribed quantization scale;

control information set means for setting control information which is used for modification of the prescribed quantization scale; and

quantization scale decision means for modifying the prescribed quantization scale using the quantity of the coded data output from the coding means, the local characteristic values calculated by the local characteristic value calculation means, and the control information set by the control information set means, and outputting the modified quantization scale toward the quantization means.

2. The image coding apparatus of claim 1 wherein:

the local characteristic value calculation means is a parameter calculation means which provides, as the local characteristic values, quantization scale modification parameters calculated on the basis of the complexity index values, an image significance decision result obtained as a result of comparison of the complexity index value of each input image to be transformed with a prescribed value, and a significant image area ratio showing the number of the input images to be transformed whose image significance decision results are "significant", and the parameter calculation means outputs the quantization scale modification parameters, the image significance decision result, and the significant image area ratio toward the quantization scale decision means;

the control information set means sets a target value of the quantity of the coded data as a target data quantity, and outputs the target data quantity toward the quantization scale decision means; and

the quantization scale decision means decides the modified quantization scale, on the basis of the target data quantity output from the control information set means, and the quantization scale parameters, the significant image area ratio and the image significance decision result which are output from the parameter calculation means.

3. The image coding apparatus of claim 1 further comprising control information storage means for storing control information used for modification of the prescribed quantization scale, wherein:

the local characteristic value calculation means is a complexity calculation means which provides a complexity mean value being an average of the complexity index values, an image significance decision result being a result of comparison of the complexity index value of each input image to be transformed with a prescribed value, and a significant image area ratio showing the number of the input images to be transformed whose image significance decision results are "significant", and the complexity calculation means outputs the complexity mean value and the significant image area ratio toward the control information storage means as control information while outputs the complexity index values toward the quantization scale decision means;

the control information set means sets a target value of the quantity of the coded data as a target data quantity, sets a complexity mean value and a significant image area ratio on the basis of the control information stored in the control information storage means, and outputs the target data quantity, the complexity mean value, and the significant image area ratio toward the quantization scale decision means; and

the quantization scale decision means decides the modified quantization scale, on the basis of quantization scale modification parameters calculated from the complexity index values, and the complexity mean value, the significant image area ratio and the target data quantity which are output from the control information set means.

4. The image coding apparatus of claim 1 wherein:

the local characteristic value calculation means is a parameter calculation means which provides a complexity mean value being an average of the complexity index values with respect to the original image, modulation parameters obtained from the statistical characteristics of the complexity index values, and weighting parameters obtained from the complexity index values using a prescribed function, calculates quantization scale modification parameters from the complexity index values, the complexity index mean value, and the modulation parameters, and outputs the weighting parameters and the quantization scale modification parameters toward the quantization scale decision means;

the control information set means sets a target value of the quantity of the coded data as a target data quantity, sets quantization characteristic information showing quantization characteristics for decision of the quantization scale, and outputs the target data quantity and the quantization characteristic information toward the quantization scale decision means; and

the quantization scale decision means decides the modified quantization scale, on the basis of the quantization scale modification parameters output from the parameter calculation means, and the quantization characteristic information and the target data quantity which are output from the control information set means.

5. The image coding apparatus of claim 1 further comprising control information storage means for storing control information used for modification of the prescribed quantization scale, wherein:

the local characteristic value calculation means is a complexity calculation means which outputs the complexity index values as the control information to the control information storage means;

the control information set means obtains a complexity mean value as an average of the complexity index values with respect to the original image, and modulation parameters from the statistical characteristics of the complexity index values, on the basis of the complexity index values stored in the control information storage means, calculates quantization scale modification parameters from the complexity index values, the complexity index mean value, and the modulation parameters, and obtains weighting parameters from the complexity index values using a prescribed function;

the control information set means sets a target value of the quantity of the coded data as a target data quantity, and sets quantization characteristic information showing characteristics of quantization for decision of the quantization scale;

the control information set means outputs the quantization scale modification parameters, the weighting parameters, the target data quantity, and the quantization characteristic information toward the quantization scale decision means; and

the quantization scale decision means decides the modified quantization scale, on the basis of the quantization scale modification parameters, the weighting parameters, the quantization characteristic information, and the target data quantity which are output from the control information set means.

6. A recording medium containing coded image data, wherein the coded image data are coded data output from the image coding apparatus according to claim 1.

7. An image decoding apparatus for decoding coded image data to obtain an image, comprising:

variable-length decoding means for decoding coded image data to generate a transformed and quantized image, reference image information, and a modified quantization scale;

inverse quantization means for performing inverse quantization to the transformed and quantized image using the modified quantization scale, thereby to obtain an inversely quantized image;

inverse image transform means for performing inverse transform to the inversely quantized image, thereby to obtain an inversely transformed image;

addition means for generating a decoded image using the inversely transformed image and a reference image which is obtained from image storage means described later; and

image storage means for storing the decoded image for a prescribed period of time, generating a reference image from the stored decoded image and the reference image information output from the variable-length decoding means, and outputting the reference image to the addition means;

wherein the coded image data is obtained from an image coding apparatus that receives an input image per coding unit, which is obtained by dividing an original digital image into coding units, and codes the input image to obtain coded data, said apparatus comprising

subtraction means for providing a difference image between the input image per coding unit and a reference image generated from a coded image which has already been coded;

image transform means for receiving, as an input image to be transformed, one of the input image per coding unit and the difference image, and performing a prescribed transformation process to the input image to be transformed, thereby to obtain a transformed image;

quantization means for quantizing the transformed image by one of a prescribed quantization scale and a modified quantization scale which is given from a quantization scale decision means, thereby to obtain transformed and quantized coefficients;

coding means for coding the transformed and quantized coefficients to obtain coded data, and outputting the coded data,

local characteristic value calculation means for obtaining a complexity index value from the input image to be transformed and, using the complexity index value, calculating local characteristic values which are used for modification of the prescribed quantization scale;

control information set means for setting control information which is used for modification of the prescribed quantization scale; and

quantization scale decision means for modifying the prescribed quantization scale using the quantity of the coded data output from the coding means, the local characteristic values calculated by the local characteristic value calculation means, and the control information set by the control information set means, and outputting the modified quantization scale toward the quantization means.

8. An image coding method wherein an input image per coding unit, which is obtained by dividing an original digital image into coding units, is coded to obtain coded data, said method comprising:

subtraction step of providing a difference image between the input image per coding unit and a reference image generated from a coded image which has already been coded;

image transform step of receiving, as an input image to be transformed, one of the input image per coding unit and the difference image, and performing a prescribed transformation process to the input image to be transformed, thereby to obtain a transformed image;

quantization step of quantizing the transformed image by one of a prescribed quantization scale and a modified quantization scale which is decided in a quantization scale decision step, thereby to obtain transformed and quantized coefficients;

coding step of coding the transformed and quantized coefficients to generate coded data;

local characteristic value calculation step of obtaining a complexity index value from the input image to be transformed and, using the complexity index value, calculating local characteristic values which are used for modification of the prescribed quantization scale;

control information set step of setting control information which is used for modification of the prescribed quantization scale; and

quantization scale decision step of modifying the prescribed quantization scale using the quantity of the coded data generated in the coding step, the local characteristic values calculated in the local characteristic value calculation step, and the control information set in the control information set step, thereby to decide a modified quantization scale.

9. The image coding method of claim 8 wherein:

the local characteristic value calculation step is a parameter calculation step in which quantization scale modification parameters are calculated on the basis of the complexity index values, an image significance decision result is obtained as a result of comparison of the complexity index value of each input image to be transformed with a prescribed value, and a significant image area ratio is obtained, which ratio shows the number of the input images to be transformed whose image significance decision results are "significant";

in the control information set step, a target value of the quantity of the coded data is set as a target data quantity; and

in the quantization scale decision step, the modified quantization scale is decided on the basis of the target data quantity set in the control information set step, and the quantization scale parameters, the significant image area ratio and the image significance decision result which are calculated in the parameter calculation step.

10. The image coding method of claim 8 wherein:

the local characteristic value calculation step is a complexity calculation step in which a complexity mean value is obtained as an average of the complexity index values, an image significance decision result is obtained as a result of comparison of the complexity index value of each input image to be transformed with a prescribed value, and a significant image area ratio is obtained, which ratio shows the number of the input images to be transformed whose image significance decision results are "significant";

in the control information set step, a target value of the quantity of the coded data is set as a target data quantity, and a complexity mean value and a significant image area ratio are set on the basis of control information including the complexity mean value and the significant image area ratio; and

in the quantization scale decision step, the modified quantization scale is decided on the basis of quantization scale modification parameters calculated from the complexity index values, and the complexity mean value, the significant image area ratio and the target data quantity which are set in the control information set step.

11. The image coding method of claim 8 wherein:

the local characteristic value calculation step is a parameter calculation step in which a complexity mean value is obtained as an average of the complexity index values with respect to the original image, modulation parameters are obtained from the statistical characteristics of the complexity index values, weighting parameters are obtained from the complexity index values using a prescribed function, quantization scale modification parameters are calculated using the complexity index values, the complexity index mean value, and the modulation parameters, and the weighting parameters and the quantization scale modification parameters are generated so that these parameters can be used in the quantization scale decision step;

in the control information set step, a target value of the quantity of the coded data is set as a target data quantity, quantization characteristic information showing quantization characteristics for decision of the quantization scale is set, and the target data quantity and the quantization characteristic information are output so that these can be used in the quantization scale decision step; and

in the quantization scale decision step, the modified quantization scale is decided on the basis of the quantization scale modification parameters calculated in the parameter calculation step, and the quantization characteristic information and the target data quantity which are set in the control information set step.

12. The image coding method of claim 8 wherein:

in the control information set step, on the basis of the complexity index values being the control information, a complexity mean value is obtained as an average of the complexity index values with respect to the original image, modulation parameters are obtained from the statistical characteristics of the complexity index values, quantization scale modification parameters are calculated using the complexity index values, the complexity mean value, and the modulation parameters, weighting parameters are obtained from the complexity index values using a prescribed function, a target value of the quantity of the coded data is set as a target data quantity, and quantization characteristic information showing characteristics of quantization for decision of the modified quantization scale is set; and

in the quantization scale decision step, the modified quantization scale is decided on the basis of the quantization scale modification parameters, the weighting parameters, the quantization characteristic information, and the target data quantity which are set in the control information set step.

13. An image decoding method for decoding coded image data to obtain an image, comprising:

variable-length decoding step of decoding coded image data to generate a transformed and quantized image, reference image information, and a modified quantization scale;

inverse quantization step of performing inverse quantization to the transformed and quantized image using the modified quantization scale, thereby to obtain an inversely quantized image;

inverse image transform step of performing inverse transform to the inversely quantized image, thereby to obtain an inversely transformed image;

addition step of generating a decoded image using the inversely transformed image and a reference image which is generated in an image storage step; and

image storage step of storing the decoded image for a prescribed period of time, and generating a reference image from the stored decoded image and the reference image information generated in the variable-length decoding step;

wherein the coded data obtained by the image coding method according to claim 8 is processed as the coded image data.

14. A method for transmitting coded image data, wherein the coded image data are coded data processed in the image coding method according to claim 8.

15. A recording medium containing an image coding program for coding an input image per coding unit, which is obtained by dividing an original digital image into coding units, thereby to obtain coded data, said image coding program comprising:

subtraction step of providing a difference image between the input image per coding unit and a reference image generated from a coded image which has already been coded;

image transform step of receiving, as an input image to be transformed, one of the input image per coding unit and the difference image, and performing a prescribed transformation process to the input image to be transformed, thereby to obtain a transformed image;

quantization step of quantizing the transformed image by one of a prescribed quantization scale and a modified quantization scale which is decided in a quantization scale decision step, thereby to obtain transformed and quantized coefficients;

coding step of coding the transformed and quantized coefficients to generate coded data;

local characteristic value calculation step of obtaining a complexity index value from the input image to be transformed and, using the complexity index value, calculating local characteristic values which are used for modification of the prescribed quantization scale;

control information set step of setting control information which is used for modification of the prescribed quantization scale; and

quantization scale decision step of modifying the prescribed quantization scale using the quantity of the coded data generated in the coding step, the local characteristic values calculated in the local characteristic value calculation step, and the control information set in the control information set step, thereby to decide a modified quantization scale.

16. The recording medium of claim 15 wherein:

the local characteristic value calculation step of the image coding program is a parameter calculation step in which quantization scale modification parameters are calculated on the basis of the complexity index values, an image significance decision result is obtained as a result of comparison of the complexity index value of each input image to be transformed with a prescribed value, and a significant image area ratio is obtained, which ratio shows the number of the input images to be transformed whose image significance decision results are "significant";

in the control information set step of the image coding program, a target value of the quantity of the coded data is set as a target data quantity; and

in the quantization scale decision step of the image coding program, the modified quantization scale is decided on the basis of the target data quantity set in the control information set step, and the quantization scale parameters, the significant image area ratio and the image significance decision result which are calculated in the parameter calculation step.

17. The recording medium of claim 15 wherein:

the local characteristic value calculation step of the image coding program is a complexity calculation step in which a complexity mean value is obtained as an average of the complexity index values, an image significance decision result is obtained as a result of comparison of the complexity index value of each input image to be transformed with a prescribed value, and a significant image area ratio is obtained, which ratio shows the number of the input images to be transformed whose image significance decision results are "significant";

in the control information set step of the image coding program, a target value of the quantity of the coded data is set as a target data quantity, and a complexity mean value and a significant image area ratio are set on the basis of control information including the complexity mean value and the significant image area ratio; and

in the quantization scale decision step of the image coding program, a modified quantization scale is decided on the basis of quantization scale modification parameters calculated from the complexity index values, and the complexity mean value, the significant image area ratio and the target data quantity which are set in the control information set step.

18. The recording medium of claim 15 wherein:

the local characteristic value calculation step of the image coding program is a parameter calculation step in which a complexity mean value is obtained as an average of the complexity index values with respect to the original image, modulation parameters are obtained from the statistical characteristics of the complexity index values, weighting parameters are obtained from the complexity index values using a prescribed function, quantization scale modification parameters are calculated using the complexity index values, the complexity index mean value, and the modulation parameters, and the weighting parameters and the quantization scale modification parameters are generated so that these parameters can be used in the quantization scale decision step;

in the control information set step of the image coding program, a target value of the quantity of the coded data is set as a target data quantity, quantization characteristic information showing quantization characteristics for decision of the modified quantization scale is set, and the target data quantity and the quantization characteristic information are output so that these can be used in the quantization scale decision step; and

in the quantization scale decision step of the image coding program, the modified quantization scale is decided on the basis of the quantization scale modification parameters calculated in the parameter calculation step, and the quantization characteristic information and the target data quantity which are set in the control information set step.

19. The recording medium of claim 15 wherein:

in the control information set step of the image coding program, on the basis of the complexity index values being the control information, a complexity mean value is obtained as an average of the complexity index values with respect to the original image, modulation parameters are obtained from the statistical characteristics of the complexity index values, quantization scale modification parameters are calculated using the complexity index values, the complexity mean value, and the modulation parameters, weighting parameters are obtained from the complexity index values using a prescribed function, a target value of the quantity of the coded data is set as a target data quantity, and quantization characteristic information showing characteristics of quantization for decision of the modified quantization scale is set; and

in the quantization scale decision step of the image coding program, the modified quantization scale is decided on the basis of the quantization scale modification parameters, the weighting parameters, the quantization characteristic information, and the target data quantity which are set in the control information set step.

20. A recording medium containing an image decoding program for decoding coded image data to obtain an image, said image decoding program comprising:

variable-length decoding step of decoding coded image data to generate a transformed and quantized image, reference image information, and a modified quantization scale;

inverse quantization step of performing inverse quantization to the transformed and quantized image using the modified quantization scale, thereby to obtain an inversely quantized image;

inverse image transform step of performing inverse transform to the inversely quantized image, thereby to obtain an inversely transformed image;

addition step of generating a decoded image using the inversely transformed image and a reference image which is generated in an image storage step; and

image storage step of storing the decoded image for a prescribed period of time, and generating a reference image from the stored decoded image and the reference image information generated in the variable-length decoding step;

wherein the coded image data is obtained from an image coding method wherein an input image per coding unit, which is obtained by dividing an original digital image into coding units, is coded to obtain coded data, said method comprising

subtraction step of providing a difference image between the input image per coding unit and a reference image generated from a coded image which has already been coded;

image transform step of receiving, as an input image to be transformed, one of the input image per coding unit and the difference image, and performing a prescribed transformation process to the input image to be transformed, thereby to obtain a transformed image;

quantization step of quantizing the transformed image by one of a prescribed quantization scale and a modified quantization scale which is decided in a quantization scale decision step, thereby to obtain transformed and quantized coefficients;

coding step of coding the transformed and quantized coefficients to generate coded data;

local characteristic value calculation step of obtaining a complexity index value from the input image to be transformed and, using the complexity index value, calculating local characteristic values which are used for modification of the prescribed quantization scale; and

quantization scale decision step of modifying prescribed quantization scale using the quantity of the coded data generated in the coding step, the local characteristic values calculated in the local characteristic value calculation step, and the control information set in the control information set step, thereby to decide a modified quantization scale.