Transform coefficient coding method and device therefor
An image decoding method performed by a decoding device according to the present document comprises a step of receiving a bitstream containing residual information, and a step of deriving a quantized transform coefficient for a current block on the basis of the residual information contained in the bitstream, wherein the residual information includes a context syntax element coded on the basis of the context, and if the size of a transformation block is changed as much as a sub-block region having a predetermined size and the current block is the last sub-block in the sub-block region and the last quantized transform coefficient is in the current block, then the step of deriving a quantized transform coefficient includes context-based decoding of the context syntax element on the basis of a predetermined maximum value for the context syntax element, wherein the maximum value is determined on the basis of the size of the sub-block region.
1. An image decoding method performed by a decoding apparatus, the method comprising:
deriving quantized transform coefficients for a current block based on residual information included in a bitstream;
deriving dequantized transform coefficients for the current block based on the quantized transform coefficients for the current block;
deriving residual samples for the current block based on the dequantized transform coefficients; and
generating a reconstructed picture based on the residual samples for the current block, wherein the deriving of the quantized transform coefficients comprises:
deriving a maximum value in a context model of context-coded bins for the current block; and
decoding context syntax elements based on a context corresponding to the context model, wherein the context syntax elements are coded based on the context, and
wherein the context syntax elements are decoded based on the maximum value.
2. The image decoding method of claim 1 , wherein the context syntax element includes a significant coefficient flag related to whether or not the transform coefficient is a non-zero significant coefficient, a parity level flag related to a parity of a transform coefficient level for the transform coefficient, a first transform coefficient level flag related to whether or not the transform coefficient level is greater than a first threshold value, and a second transform coefficient level flag related to whether or not the transform coefficient level for the transform coefficient is greater than a second threshold value, and
wherein the maximum value is a maximum value of a sum of a number of the significant coefficient flags, a number of the parity level flags, a number of the first transform coefficient level flags, and a number of the second transform coefficient level flags.
3. The image decoding method of claim 2 , wherein the current block is a transform block, and
wherein a specific region of the image includes a plurality of complete sub-blocks.
4. The image decoding method of claim 3 , wherein the maximum value (CurrNumTBRegBin) is determined based on a following equation:
CurrNumTBRegBin
=
NumToBeCodedSBB
subblock
number
of
current
TB
×
NumTURegBin
[
width
]
[
height
]
where NumTURegBin[width][height] designates a maximum number of context-encoded bins that may be used in one transform block based on the size of the transform block, and NumToBeCodedSBB designates a number of the sub-blocks within the specific region.
5. The image decoding method of claim 2 , wherein when the sum of the number of the significant coefficient flags, the number of the first transform coefficient level flags, the number of the parity level flags, and the number of the second transform coefficient level flags derived based on a 0th transform coefficient to an nth transform coefficient determined by a coefficient scan order reaches the maximum value, then explicit signaling of a significant coefficient flag, a first transform coefficient level flag, a parity level flag, and a second transform coefficient level flag is omitted for a (n+l)th transform coefficient determined by the coefficient scan order, and a value of the (n+l)th transform coefficient is derived based on a value of coefficient level information included in the residual information.
6. The image decoding method of claim 5 , wherein the significant coefficient flags, the first transform coefficient level flags, the parity level flags, and the second transform coefficient level flags included in the residual information are context-based coded, and the coefficient level information is bypass-based coded.
7. The image decoding method of claim 2 , wherein the deriving the transform coefficient comprises:
decoding the first transform coefficient level flag and the parity level flag; and
deriving the transform coefficient based on a value of the decoded parity level flag and a
value of the decoded first transform coefficient level flag, and
wherein the decoding of the first transform coefficient level flag is performed prior to the decoding of the parity level flag.
8. An image encoding method performed by an encoding apparatus, the method comprising:
deriving residual samples for a current block;
deriving transform coefficients based on residual samples for a current block;
deriving quantized transform coefficients based on the transform coefficients for the current block; and
encoding residual information including information on the quantized transform coefficients, wherein the encoding of the residual information comprises:
deriving a maximum value in a context model of context-coded bins for the current block; and
encoding context syntax elements based on a context corresponding to the context model, wherein the context syntax elements are coded based on the context,
wherein the context syntax elements are encoded based on the maximum value.
9. The image encoding method of claim 8 , wherein the context syntax element includes a significant coefficient flag related to whether or not the transform coefficient is a non-zero significant coefficient, a parity level flag related to a parity of a transform coefficient level for the transform coefficient, a first transform coefficient level flag related to whether or not the transform coefficient level is greater than a first threshold value, and a second transform coefficient level flag related to whether or not the transform coefficient level for the transform coefficient is greater than a second threshold value, and
wherein the maximum value is a maximum value of a sum of a number of the significant coefficient flags, a number of the parity level flags, a number of the first transform coefficient level flags, and a number of the second transform coefficient level flags.
10. The image encoding method of claim 9 , wherein the current block is a transform block, and wherein a specific region of the image includes a plurality of complete sub-blocks.
11. The image encoding method of claim 10 , wherein the maximum value (CurrNumTBRegBin) is determined based on a following equation:
CurrNumTBRegBin
=
NumToBeCodedSBB
subblock
number
of
current
TB
×
NumTURegBin
[
width
]
[
height
]
where NumTURegBin[width][height] designates a maximum number of context-encoded bins that may be used in one transform block based on the size of the transform block, and NumToBeCodedSBB designates a number of the sub-blocks within the specific region.
12. The image encoding method of claim 9 , wherein when the sum of the number of the significant coefficient flags, the number of the first transform coefficient level flags, the number of the parity level flags, and the number of the second transform coefficient level flags derived based on a 0 th transform coefficient to an nth transform coefficient determined by a coefficient scan order reaches the maximum value, then explicit signaling of a significant coefficient flag, a first transform coefficient level flag, a parity level flag, and a second transform coefficient level flag is omitted for a (n+l)th transform coefficient determined by the coefficient scan order, and a value of the (n+l)th transform coefficient is derived based on a value of coefficient level information included in the residual information.
13. The image decoding method of claim 3 , wherein the maximum value determined based on the specific region in which a last significant transform coefficient of the current block exists is smaller than the maximum value of the context-coded bin based on the size of the currentblock.
14. The image encoding method of claim 10 , wherein the maximum value determined based on the specific region in which a last significant transform coefficient of the current block exists is smaller than the maximum value of the context-coded bin based on the size of the currentblock.
15. A non-transitory computer-readable storage medium storing a bitstream generated by a method, the method comprising:
deriving residual samples for a current block;
deriving transform coefficients based on residual samples for a current block;
deriving quantized transform coefficients based on the transform coefficients for the current block; and
encoding residual information including information on the quantized transform coefficients to generate the bitstream, wherein the encoding of the residual information comprises:
deriving a maximum value in a context model of context-coded bins for the current block; and
encoding context syntax elements based on a context corresponding to the context model, wherein the context syntax elements are coded based on the context,
wherein the context syntax elements are encoded based on the maximum value.