Image coding method and image coding apparatus
Image coding systems in recent years support a plurality of partitions about a transform unit size, prediction unit size, and coding unit size, which are processing units of transform, prediction, and coding, respectively. That is, combination of respective unit sizes is diverse. The present disclosure provides a method and apparatus for reducing a candidate number of partition sizes in an image and for reducing complexity in image coding by determining transform unit sizes and by setting prediction unit sizes and coding unit sizes based on the determined transform unit sizes.
1. An image coding method for coding a block in an image on a block-by-block basis, comprising:
determining a coding unit size of a coding unit in the block, a transform unit size of a transform unit in the block, and a prediction unit size of a prediction unit in the block;
predicting the block, on each prediction unit, to generate a prediction block;
transforming a difference between the block and the prediction block to residual data, on each transform unit; and
coding the residual data on each coding unit to generate a bitstream,
wherein a minimum coding unit size is 8×8, and both a minimum transform unit size and a minimum prediction unit size are 4×4,
wherein when the transform unit size is set as the minimum coding unit size, both the transform unit size and the prediction unit size are the minimum coding unit size, the transform unit size is not the minimum transform unit size, and the prediction unit size is not the minimum prediction unit size, and
wherein when the transform unit size is not set as the minimum coding unit size, both the transform unit size and the prediction unit size are not always the same minimum coding unit size.
2. An image coding method for coding a block in an image on a block-by-block basis, comprising:
determining a first size of a first unit for coding the block, a second size of a second unit for transforming the block, and a third size of a third unit for predicting the block;
predicting the block on each third unit to generate a prediction block;
transforming a difference between the block and the prediction block to residual data, on each second unit; and
coding the residual data on each first unit to generate a bitstream,
wherein a minimum first unit size is 8×8, and both a minimum second unit size and a minimum third unit size are 4×4,
wherein when the second size is set as a same size as the minimum first unit size, both the second size and the third size are the minimum first unit size, the second size is not the minimum second unit size, and the third unit size is not the minimum third unit size, and
wherein when the transform unit size is not set as the minimum coding unit size, both the transform unit size and the prediction unit size are not always the same minimum coding unit size.