Method and apparatus for image coding
View Patent ↗A method of coding a moving picture includes dividing a picture into a plurality of blocks, and selectively performing one of global motion compensation processing and local motion compensation processing on each of the blocks to obtain predictive errors for the block, orthogonally transforming the predictive errors to obtain orthogonal transform coefficients, and quantizing the orthogonal transform coefficients using a quantizing parameter. Which type of motion compensation processing is performed for each block is determined based on a result of comparing evaluation values for the two types of motion compensation processings with each other. The evaluation values are calculated based on (1) a sum of absolute values of predictive errors for the block obtained when the type of compensation processing is performed on the block and (2) a value determined by the quantizing parameter.
1. An image encoding method comprising:
storing a reference image which is previously decoded;
dividing an input image into a plurality of blocks;
calculating evaluation values for global motion compensation and local motion compensation for the block,
selecting one of global motion compensation and local motion compensation by comparing said evaluation values for the block,
performing a selected one of global motion compensation and local motion compensation for the block using said reference image and said input image;
wherein said evaluation values are calculated based on a sum of absolute values of predictive errors for the block and a value determined by a quantizing parameter.
2. An image encoding method according to claim 1 , wherein a motion compensation is selected whose evaluation value is smaller relative to the other.
3. An image encoder comprising:
a memory for storing a reference image which is previously decoded;
a divider for dividing an input image into a plurality of blocks;
a motion compensator for calculating evaluation values for global motion compensation and local motion compensation for the block, selecting one of global motion compensation and local motion compensation by comparing said evaluation values for the block, and performing a selected one of global motion compensation and local motion compensation for the block using said reference image and said input image by estimating motion vectors;
wherein said evaluation values are calculated based on a sum of absolute value of predictive errors for the block and a value determined by quantizing parameter.
4. An image encoding method according to claim 3 , wherein a motion compensation is selected whose evaluation value is smaller relative to the other.
5. An image encoding method according to claim 1 , wherein the evaluation values are calculated based on a kind of the local motion compensation processing performed on the block.
6. An image encoding method according to claim 1 , wherein the local motion compensation processing performed on the block is one of an INTER mode local motion compensation processing and an INTRA mode local motion compensation processing.
7. An image encoding method according to claim 1 , wherein an evaluation value is decided that the global motion compensation is easier to be selected when the value determined by the quantizing parameter is bigger.
8. An image encoding method according to claim 3 , an evaluation value is decided that the global motion compensation is easier to be selected when the value determined by the quantizing parameter is bigger.
9. An image encoding method according to claim 1 , wherein the value determined by the quantizing parameter is different depending upon whether a motion vector is zero or not zero.
10. An image encoding method according to claim 7 , wherein the value determined by the quantizing parameter is different depending upon whether a motion vector is zero or not zero.