Method and apparatus for image compression using block prediction mode
A method and apparatus of image coding performed at an encoder or decoder using block prediction are disclosed. The block vector bit-count information associated with one or more block vectors (BVs) for the current block is determined, where the BVs are used to locate one or more block predictors for the current block. Prediction residuals corresponding to differences between the current block and the block predictors are determined. A first cost comprising the BV and residual bit-count information associated with encoding the prediction residuals using the first coding tool is determined. The second cost comprising the BV bit-count information and without the residual bit-count information is also determined. The derivation of the BV bit-count information and the prediction residuals for the current block are performed only once, and the BV bit-count information and the prediction residuals are used by the first coding tool and the second coding tool.
1. A method of image decoding for an image to be performed at a decoder, wherein the decoder uses multiple coding tools including at least a first coding tool and a second coding tool associated with block prediction, the method comprising:
receiving a bitstream including compressed bits of a current block in a current slice or a current image;
if the first coding tool is used:
deriving one or more first block vectors for the current block from the bitstream according to first entropy decoding, wherein said one or more first block vectors are used to locate one or more first block predictors for the current block;
deriving decoded prediction residuals for prediction residuals from the bitstream, wherein the prediction residuals correspond to differences between the current block and said one or more first block predictors; and
reconstructing the current block based on the decoded prediction residuals and said one or more first block predictors for the current block according to first block-prediction reconstruction;
if the second coding tool is used:
deriving one or more second block vectors for the current block from the bitstream according to second entropy decoding, wherein said one or more second block vectors are used to locate one or more second block predictors for the current block; and
reconstructing the current block based on said one or more second block predictors for the current block according to second block-prediction reconstruction; and
providing reconstructed current block, wherein said first entropy decoding and said second entropy decoding share a same entropy decoder, the first block-prediction reconstruction and the second block-prediction reconstruction share a same block-prediction reconstruction unit to reconstruct one reconstructed block from one or more predictors and decoded prediction residuals, or both,
wherein the current block corresponding to a 2×8 block is partitioned into 2×2 blocks, each 2×2 block is coded according to 2×2 partition or 1×2 partition based on a partition type derived from the bitstream,
wherein one block vector is derived from the bitstream for each 1×2 block of one 2×2 block in the 1×2 partition, or one block vector is derived from the bitstream for each 2×2 block in the 2×2 partition,
wherein for the 2×2 partition, one 2×2 predictor is derived from previously reconstructed pixels in the current slice or the current image according to location of each 2×2 block and said one block vector associated with each 2×2 block, and for the 1×2 partition, each 1×2 predictor of one 2×2 block is derived from previously reconstructed pixels in the current slice or the current image according to location of each 1×2 block of one 2×2 block and said one block vector associated with each 1×2 block of one 2×2 block.
2. The method of claim 1 , wherein said one or more first block predictors for the current block are derived from previously reconstructed pixels in the current slice or the current image according to location of the current block and said one or more block vectors.
3. The method of claim 1 , wherein when the first coding tool is used, said deriving the decoded prediction residuals comprises deriving quantized prediction residuals for the current block from the bitstream.
4. The method of claim 3 , wherein when the first coding tool is used, said deriving the decoded prediction residuals further comprises applying inverse quantization to the quantized prediction residuals to recover the decoded prediction residuals.
5. The method of claim 1 , wherein when the first block-prediction reconstruction and the second block-prediction reconstruction share the same block-prediction reconstruction unit, the decoded prediction residuals are set to zero for the second block-prediction reconstruction.
6. The method of claim 1 , wherein when the second coding tool is used, said one or more predictors are used as a reconstructed current block.
7. An apparatus of image decoding for an image to be performed at a decoder, wherein the decoder uses multiple coding tools including at least a first coding tool and a second coding tool associated with block prediction, the apparatus comprising one or more electronic circuits or processors configured to:
receive a bitstream including compressed bits of a current block in a current slice or a current image;
if the first coding tool is used:
derive one or more first block vectors for the current block from the bitstream according to first entropy decoding, wherein said one or more first block vectors are used to locate one or more first block predictors for the current block;
derive decoded prediction residuals for prediction residuals from the bitstream, wherein the prediction residuals correspond to differences between the current block and said one or more first block predictors; and
reconstruct the current block based on the decoded prediction residuals and said one or more first block predictors for the current block according to first block-prediction reconstruction;
if the second coding tool is used:
derive one or more second block vectors for the current block from the bitstream according to second entropy decoding, wherein said one or more second block vectors are used to locate one or more second block predictors for the current block; and
reconstruct the current block based on said one or more second block predictors for the current block according to second block-prediction reconstruction; and
wherein said first entropy decoding and said second entropy decoding share a same entropy decoder, the first block-prediction reconstruction and the second block-prediction reconstruction share a same block-prediction reconstruction unit to reconstruct one reconstructed block from one or more predictors and decoded prediction residuals, or both,
wherein the current block corresponding to a 2×8 block is partitioned into 2×2 blocks, each 2×2 block is coded according to 2×2 partition or 1×2 partition based on a partition type derived from the bitstream,
wherein one block vector is derived from the bitstream for each 1×2 block of one 2×2 block in the 1×2 partition, or one block vector is derived from the bitstream for each 2×2 block in the 2×2 partition,
wherein for the 2×2 partition, one 2×2 predictor is derived from previously reconstructed pixels in the current slice or the current image according to location of each 2×2 block and said one block vector associated with each 2×2 block, and for the 1×2 partition, each 1×2 predictor of one 2×2 block is derived from previously reconstructed pixels in the current slice or the current image according to location of each 1×2 block of one 2×2 block and said one block vector associated with each 1×2 block of one 2×2 block.
8. The apparatus of claim 7 , wherein the one or more electronic circuits or processors are further configured to derive said one or more first block predictors for the current block from previously reconstructed pixels in the current slice or the current image according to location of the current block and said one or more block vectors.
9. The apparatus of claim 7 , wherein the one or more electronic circuits or processors are further configured to derive the decoded prediction residuals by deriving quantized prediction residuals for the current block from the bitstream when the first coding tool is used.
10. The apparatus of claim 9 , wherein the one or more electronic circuits or processors are further configured to derive the decoded prediction residuals by applying inverse quantization to the quantized prediction residuals to recover the decoded prediction residuals when the first coding tool is used.
11. The apparatus of claim 7 , wherein the one or more electronic circuits or processors are further configured to set to zero the decoded prediction residuals for the second block-prediction reconstruction when the first block-prediction reconstruction and the second block-prediction reconstruction share the same block-prediction reconstruction unit.
12. The apparatus of claim 7 , wherein the one or more electronic circuits or processors are further configured to use said one or more predictors as a reconstructed current block when the second coding tool is used.