Method and apparatus of transform coefficient coding with TB-level constraint
A method and apparatus for encoding or decoding transform coefficients in a video coding system are disclosed. According to this method, a region after zero-outis determined. A remaining number of context-coded bins is determined for the current TB based on a size of the region after zero-out. The current TB is encoded or decoded using context-based coding with a constraint based on the remaining number of context-coded bins. According to another method, a remaining number of context-coded bins is determined for the current TB based on the last significant CG index.
1. A method for encoding prediction residues in a video coding system, the method comprising:
receiving input data corresponding to transform coefficients associated with a current transformed block (TB) to be coded at an encoder side;
determining initially a maximum allowed number of context-coded bins of one or more coefficient level related syntax elements for the current TB based on a size of a region after zero-out, wherein a width or height of the region after zero-out is clipped to a pre-defined value before said determining initially the maximum allowed number, wherein the pre-defined value corresponds to 32, wherein the pre-defined value corresponds to 16 if MTS (multiple transform set) is enabled and the SBT (sub-block transform) is applied to the current TB, wherein dependency of determining initially the maximum allowed number comprises an area of the region after zero-out, wherein the dependency of said determining initially the maximum allowed number comprises the area of the region after zero-out multiplied by a factor equal to 1.75;
tracking each time a transform coefficient of the current TB is encoded using context-based coding;
based on the maximum allowed number being reached, switching to bypass mode for transform level syntax coding for all remaining coefficients of a currently processed sub-block of the TB, and all remaining sub-blocks of the TB; and
providing the coded current TB.
2. The method of claim 1 , wherein a target transform coefficient of the current TB outside the region after zero-out is not coded.
3. The method of claim 1 , wherein the pre-defined value corresponds to 16 if MTS (multiple transform set) and a sub-block transform is allowed for a current block containing the current TB.
4. A method for decoding prediction residues in a video coding system, the method comprising: receiving a coded current TB (transform block) corresponding to coded transform coefficients associated with a current TB to be decoded at a decoder side; determining initially a maximum allowed number of context-coded bins of one or more coefficient level related syntax elements for the current TB based on a size of a region after zero-out, wherein a width or height of the region after zero-out is clipped to a pre-defined value before said determining initially the maximum allowed number, wherein the pre-defined value corresponds to 32, wherein the pre-defined value corresponds to 16 if MTS (multiple transform set) is allowed for the current TB, wherein dependency of determining initially the maximum allowed number comprises an area of the region after zero-out, wherein the dependency of said determining initially the maximum allowed number comprises the area of the region after zero-out multiplied by a factor equal to 1.75; tracking each time a transform coefficient of the current TB is decoded to generate a reconstructed current TB using context-based coding; based on the maximum allowed number being reached, switching to bypass mode for transform level syntax coding for all remaining coefficients of a currently processed sub-block of the TB, and all remaining sub-blocks of the TB; and providing the reconstructed current TB.
5. The method of claim 4 , wherein a target transform coefficient of the current TB outside the region after zero-out is not parsed.
6. The method of claim 4 , wherein the pre-defined value corresponds to 16 if MTS (multiple transform set) and a sub-block transform is allowed for a current block containing the current TB.
7. The method of claim 4 , wherein the dependency of said determining the maximum allowed number further comprises the area of the region after zero-out and then right-shifted by 2.
8. The method of claim 7 , wherein the factor can be different for luma TB and chroma TB.
9. An apparatus for decoding prediction residues in a video coding system, the apparatus comprising one or more electronic circuits or processors arranged to:
receive a coded current TB (transform block) corresponding to coded transform coefficients associated with a current TB to be decoded at a decoder side;
determine initially a maximum allowed number of context-coded bins of one or more coefficient level related syntax elements for the current TB based on a size of a region after zero-out, wherein a width or height of the region after zero-out is clipped to a pre-predefined value before said determining initially the maximum allowed number, wherein the pre-defined value corresponds to 32, wherein the pre-defined value corresponds to 16 if MTS (multiple transform set) is enabled and the SBT (sub-block transform) is applied to the current TB, wherein dependency of determining initially the maximum allowed number comprises an area of the region after zero-out, wherein the dependency of said determining initially the maximum allowed number comprises the area of the region after zero-out multiplied by a factor equal to 1.75;
track each time a transform coefficient of the current TB is decoded to generate a reconstructed current TB using context-based coding;
based on the maximum allowed number being reached, switch to bypass mode for transform level syntax coding for all remaining coefficients of a currently processed sub-block of the TB, and all remaining sub-blocks of the TB; and
provide the reconstructed current TB.