Coding transform coefficients with throughput constraints

1. A method comprising:

receiving input data associated with a plurality of syntax elements representing transform coefficients of a current block; and

entropy coding the plurality of syntax elements selectively as either regular bins using context modeling or as bypass bins without context modeling, wherein a number of the regular bins used for entropy coding the plurality of syntax elements representing transform coefficients of a transform block is limited by a constraint,

wherein the constraint limiting the number of the regular bins is determined based on a size of the transform block and a basic termination threshold that is less than or equal to 32 per 4×4 sub-block, or equivalently is less than equal to 2 per sample,

wherein the current block is divided into one or more subblocks that are entropy coded according to a specified scan order,

wherein the plurality of syntax elements in each of the one or more subblocks are coded in a plurality of coding passes, and the syntax elements in a first coding pass are all coded using regular bins and a total number of the regular bins used for entropy coding the syntax elements in the first coding pass is limited by the constraint and the syntax elements in other coding passes are all coded using bypass bins, and

wherein when an accumulated number of used regular bins from the coded transform coefficients in the current block is greater than the constraint for a current transform coefficient, the first coding pass is skipped for a current transform coefficient of the current block, remaining transform coefficients of the current block, and all transform coefficients of remaining uncoded subblocks.

2. The method of claim 1 , wherein an absolute value for a transform coefficient is indicated by one or more of a first flag, a second flag, and a third flag that are entropy coded in the first coding pass, and wherein:

the first flag indicates whether an absolute value of the transform coefficient is greater than 0,

the second flag indicates whether the absolute value of the transform coefficient is greater than 1, and

the third flag indicates whether the absolute value of the transform coefficient is greater than 3.

3. The method of claim 2 , wherein:

a syntax flag for indicating a parity of the absolute value of the transform coefficient is entropy coded in the first coding pass,

a remainder of the transform coefficient is entropy coded in a second coding pass, and

a sign of the transform coefficient is entropy coded in a third coding pass.

4. The method of claim 3 , wherein the regular bins of the first coding pass are entropy coded by using one or more context models that are dependent on one or more indicators of absolute values of one or more neighboring transform coefficients.

5. The method of claim 4 , wherein at least one context model for entropy coding the first flag or the second flag is selected conditioned on a value of the third flag.

6. The method of claim 1 , wherein the constraint limiting the number of the regular bins is different for a first color component than for a second color component.

7. The method of claim 1 , wherein the constraint limiting the number of the regular bins is computed by multiplying a default constraint having a first value with a predefined factor that is derived from a ratio of a number of coded subblocks in the current block and a number of all subblocks in the current block.

8. The method of claim 1 , wherein the number of the regular bins used for individual syntax elements present in a transform subblock or a transform subblock coding pass is not limited as long as the accumulated number of the used regular bins from the coded transform coefficients in the current block is not greater than the constraint for the current transform coefficient.

9. The method of claim 1 , wherein the constraint limiting the number of the regular bins limits a number of regular bins used for a subblock of the current block.

10. An electronic apparatus comprising:

a decoder circuit capable of: receiving from a bitstream code bits that correspond to a plurality of syntax elements representing transform coefficients of a current block;

entropy decoding the code bits into the plurality of syntax elements selectively as either regular bins using context modeling or as bypass bins without context modeling, wherein a number of the regular bins used for entropy decoding the plurality of syntax elements representing transform coefficients of a transform block is limited by a constraint,

wherein the constraint limiting the number of the regular bins is determined based on a size of the transform block and a basic termination threshold that is less than or equal to 32 per 4×4 sub-block, or equivalently is less than equal to 2 per sample,

wherein the current block is divided into one or more subblocks that are entropy coded according to a specified scan order,

wherein the plurality of syntax elements in each of the one or more subblocks are coded in a plurality of coding passes, and the syntax elements in a first coding pass are all coded using regular bins and a total number of the regular bins used for entropy coding the syntax elements in the first coding pass is limited by the constraint and the syntax elements in other coding passes are all coded using bypass bins, and

wherein when an accumulated number of used regular bins from the coded transform coefficients in the current block is greater than the constraint for a current transform coefficient, the first coding pass is skipped for a current transform coefficient of the current block, remaining transform coefficients of the current block, and all transform coefficients of remaining encoded subblocks; and

reconstructing the current block using the entropy decoded syntax elements.

11. An electronic apparatus comprising:

an encoder circuit capable of receiving a plurality of syntax elements representing transform coefficients of a current block;

entropy encoding the plurality of syntax elements selectively as either regular bins using context modeling or as bypass bins without context modeling, wherein a number of the regular bins used for entropy encoding the plurality of syntax elements representing transform coefficients of a transform block is limited by a constraint,

wherein the constraint limiting the number of the regular bins is determined based on a size of the transform block and a basic termination threshold that is less than or equal to 32 per 4×4 sub-block, or equivalently is less than equal to 2 per sample,

wherein the current block is divided into one or more subblocks that are entropy coded according to a specified scan order,

wherein the plurality of syntax elements in each of the one or more subblocks are coded in a plurality of coding passes, and the syntax elements in a first coding pass are all coded using regular bins and a total number of the regular bins used for entropy coding the syntax elements in the first coding pass is limited by the constraint and the syntax elements in other coding passes are all coded using bypass bins, and

wherein when an accumulated number of used regular bins from the coded transform coefficients in the current block is greater than the constraint for a current transform coefficient, the first coding pass is skipped for a current transform coefficient of the current block, remaining transform coefficients of the current block, and all transform coefficients of remaining encoded subblocks; and

storing the entropy encoded syntax elements in a bitstream.

12. The method of claim 1 , wherein the regular bins limited by the constraint are used for entropy coding the plurality of syntax elements present in the transform subblock or the transform subblock coding pass.

13. The method of claim 12 , wherein the constraint limiting the number of the regular bins is a first constraint for a 2×2 transform subblock and a second constraint, which is different from the first constraint, for a 4×4 transform subblock.