IMAGE CODING METHOD AND IMAGE DECODING METHOD

• US 20120121012A1
• Filed: 01/26/2012
• Published: 05/17/2012
• Est. Priority Date: 07/27/2009
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Shiodera, Taichiro, Koto, Shinichiro
• Granted Patent: US 8,867,614 B2
• Time in Patent Office: Days
• US Class Current: 375/240.03
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/13 Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
• 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 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/18 the unit being a set of transform coefficients
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

According to an embodiment, an image coding method includes a transforming and quantizing step of orthogonally transforming and quantizing each of a plurality of blocks as a unit within a target region of an input image to obtain coefficients; and a variable-length coding step of coding a plurality of coefficient strings each having coefficients of same frequency component among the blocks within the target region by using a plurality of variable length coding schemes: The variable-length coding step includes switching among the variable-length coding schemes for each coefficient string.

Claims

1. An image coding method comprising:

a transforming and quantizing step of orthogonally transforming and quantizing a plurality of blocks within a target region of an input image as a unit to obtain coefficients;

a aligning step of aligning coefficients among the blocks in accordance with frequency to make a coefficient string, the coefficients in a same coefficient string has a same frequency component; and

a variable-length coding step of variable-length coding the coefficient strings by using different types of variable length coding schemes, wherein

the variable-length coding step includes switching among different types of the variable-length coding schemes for each coefficient string.

2. The method according to claim 1, wherein

the variable-length coding step includes:

a first variable-length coding step of variable-length coding the coefficient strings by the coefficient;

a second variable-length coding step of coding the coefficient strings by a variable-length partial data, wherein the variable-length partial data includes variable number of coefficients.

3. The method according to claim 2, further comprising:

a predicting step of predicting a predicted image of the block by using a reference image that has already been decoded, wherein

the transforming and quantizing step includes orthogonally transforming and quantizing a prediction error between the predicted image and the block of the input image to obtain the coefficients, and

the variable-length coding step includes switching between performing the first variable-length coding step and performing the second variable-length coding step for each coefficient string based on at least one of a quantization width representing a width of the quantization, a prediction direction representing a direction of prediction by the predicting process, and signal components of the input image.

4. The method according to claim 3, wherein

the variable-length coding step includes switching between performing the first variable-length coding step and performing the second variable-length coding step for each coefficient string so that the first variable-length coding step is performed on a coefficient string of low frequency components and the second variable-length coding step is performed on a coefficient string of high frequency components when the frequency components are separated into low frequency components including a direct-current component and high frequency components.

5. The method according to claim 4, wherein

the variable-length coding step includes:

widening a range of the low frequency components as the quantization width is larger; and

narrowing the range of the low frequency components as the quantization width is smaller.

6. The method according to claim 5, wherein

the first variable-length coding step includes performing variable-length coding of the coefficients by using Golomb-Rice coding while switching, for each coefficient string within the target region, a parameter indicating a length of a fixed length part of a Golomb-Rice code.

7. The method according to claim 6, further comprising:

a setting step of setting a compression rate;

a fixed-length coding step of coding the block with a fixed code length according to the compression rate set in the setting step;

a determining step of determining whether or not a generated code amount per unit time in coding in the variable-length coding step exceeds a code amount according to the compression rate set in the setting step;

a coding step of coding the block by the fixed-length coding step when it is determined in the determining step that the generated code amount exceeds the code amount depending on the compression rate, and coding the block by the variable-length coding step when it is determined in the determining step that the generated code amount does not exceed the code amount according to the compression rate; and

a compression information coding step of coding compression information indicating which of the variable-length coding step and the fixed-length coding step is used for coding in the coding step.

8. The method according to claim 7, wherein the variable-length coding step includes:

widening the range of the low frequency components as the compression rate is lower, and

narrowing the range of the low frequency components as the compression rate is higher.

9. The method according to claim 5, wherein

the first variable-length coding step includes performing variable-length coding of the coefficients using Huffman coding while switching, for each coefficient string, a probability model for determining a code table used for the Huffman coding.

10. The method according to claim 5, wherein

the first variable-length coding step and the second variable-length coding step include performing variable-length coding of the coefficients by using arithmetic coding with use of different contexts between the first variable-length coding step and the second variable-length coding step among contexts indicating transition states of the arithmetic coding.

11. The method according to claim 4, wherein

the variable-length coding step includes:

setting the range of the low frequency components so that a range of vertical components is wider than a range of horizontal components when the direction of prediction by the prediction process is horizontal prediction; and

setting the range of the low frequency components so that the range of horizontal components is wider than the range of vertical components when the direction of prediction by the prediction process is vertical prediction.

12. An image decoding method comprising:

a variable-length decoding step of variable-length decoding coded data of each of a plurality of coefficient string each having coefficients of same frequency component among a plurality of decoding target blocks within a decoding target region of a decoding target image by using different types of variable-length decoding schemes to obtain the coefficient strings; and

an inverse quantizing and inverse transforming step of inverse quantizing and inverse transforming coefficients of each decoding target block out of the coefficient strings, wherein

the variable-length decoding step includes switching among different types of the variable-length decoding schemes for each coefficient string.

13. The method according to claim 12, wherein

the variable-length decoding step includes:

a first variable-length decoding step of variable-length decoding the coded data by the coefficient;

a second variable-length decoding step of variable-length decoding the coded data by a variable-length partial data, wherein

the variable-length partial data includes variable number of coefficients.

14. The method according to claim 13, wherein

the coded data are generated by coding coefficients obtained by orthogonally transforming and quantizing an error between a predicted image predicted by using a reference image, which has already been obtained as a result of decoding the image, and the image in units of the block, and

the variable-length decoding step includes switching between performing the first variable-length decoding step and performing the second variable-length decoding step for each coefficient string based on at least one of a quantization width representing a width of the quantization, a prediction direction representing a direction of prediction by the predicting process, and signal components of the image.

15. The method according to claim 14, wherein

the variable-length decoding step includes switching between performing the first variable-length decoding step and performing the second variable-length decoding step for each coefficient string so that the first variable-length decoding step is performed on a coefficient string of low frequency components and the second variable-length decoding step is performed on a coefficient string of high frequency components when the frequency components are separated into low frequency components including a direct-current component and high frequency components.

16. The method according to claim 15, wherein

the variable-length decoding step includes:

widening a range of the low frequency components as the quantization width is larger; and

narrowing the range of the low frequency components as the quantization width is smaller.

17. The method according to claim 16, wherein

the first variable-length decoding step includes

performing variable-length decoding of a Golomb-Rice code to decode the coefficients from the coded data while switching, for each coefficient string within the target region, a parameter indicating a length of a fixed length part of the Golomb-Rice code.

18. The method according to claim 17, wherein

the coded data are generated by coding the image by switching between fixed-length coding of coding the block with a fixed code length and variable-length coding of coding the block using a variable-length code in units of a block according to a compression rate set so that a generated code amount per unit time does not exceed a predetermined code amount, and

the method further includes:

a first decoding step of decoding the coded data coded by the fixed-length coding;

a second decoding step of decoding the coded data coded by the variable-length coding by switching between performing the first variable-length decoding step and performing the second variable-length decoding step in the switching;

a coding method information decoding step of decoding coding method information supplied together with the coded data and indicating which of the fixed-length coding and the variable-length coding is used to code the coded data; and

a selecting step of selecting one of performing the first decoding step and performing the second decoding step according to the coding method information decoded in the coding method information decoding step.

19. The method according to claim 18, wherein

the variable-length decoding step includes:

widening the range of the low frequency components as the compression rate is lower; and

narrowing the range of the low frequency components as the compression rate is higher.

20. The method according to claim 16, wherein

the first variable-length decoding step includes decoding the coefficients from the coded data by performing variable-length decoding of a Huffman code while switching, for each coefficient string, a probability model for determining a code table used for generating the Huffman code.

21. The method according to claim 16, wherein

the first variable-length decoding step and the second variable-length decoding step include performing variable-length decoding of an arithmetic code to decode the coefficients from the coded data with use of different contexts between the first variable-length decoding step and the second variable-length decoding step among contexts indicating transition states of arithmetic coding that generates the arithmetic code.

22. The method according to claim 15, wherein

the switching step includes:

setting the range of the low frequency components so that a range of vertical components is wider than a range of horizontal components when the direction of prediction by the prediction process is horizontal prediction, and

setting the range of the low frequency components so that the range of horizontal components is wider than the range of vertical components when the direction of prediction by the prediction process is vertical prediction.