Inverse quantization device and method used in video decoding device
An inverse quantization method is implemented by an inverse quantization device, the method configured for acquiring quantized coefficients, estimating a quantization parameter in quantization groups or quantization parameter prediction group units, generating an inverse quantization matrix for adaptive quantization, and generating transform coefficients from the quantized coefficients using the quantization parameter and the inverse quantization matrix.
1. An inverse quantization method used by an inverse quantization device, the method comprising:
decoding information about a current block, information about quantized coefficients, information about a quantization parameter, a delta index for an inverse quantization matrix, and delta coefficients of the inverse quantization matrix from a bitstream;
calculating a quantization parameter based on the information about the quantization parameter;
deriving an inference identifier (ID) from the information about the current block and selecting an inverse quantization matrix indicated by the inference ID, wherein a list for the inverse quantization matrix is pre-generated, and an inverse quantization matrix included in the list is generated using a predictive inverse quantization matrix calculated based on the delta index, and delta inverse quantized data generated from the delta coefficients; and
generating transform coefficients from the quantized coefficients using the quantization parameter and the selected inverse quantization matrix,
wherein delta coefficients corresponding to a part of the inverse quantization matrix included in the list are not decoded.
2. The method of claim 1 , further comprising: when a size of the QG and a size of a quantization parameter prediction group (QPG) are not equal to each other, decoding the size of the QPG.
3. The method of claim 2 , wherein, when the size of the QG is larger than the QPG and thereby the QG includes a plurality of QPGs, a predictive quantization parameter for each of the QPGs is generated using quantization parameters corresponding to positions above and on a left side of a top left position in each of the QPGs, and a quantization parameter for each of the QPGs is calculated using the predictive quantization parameter for each of the QPGs and a delta quantization parameter at an top left position of the QG.
4. The method of claim 2 , wherein, when the size of the QG is smaller than the QPG and thereby the QPG includes a plurality of QGs, a predictive quantization parameter for the QPG is generated by averaging quantization parameters corresponding to positions above and on a left side of a top left position of the QPG, and a quantization parameter for each of the QGs is calculated using the predictive quantization parameter for the QPG and a delta quantization parameter at an top left position of each of the QGs.
5. The method of claim 2 , wherein, when the QG and the QPG have the same size, a predictive quantization parameter is generated using quantization parameters corresponding positions above and on a left side of a top left position of the QG, and a quantization parameter for the QG is calculated using the predictive quantization parameter and a delta quantization parameter at the top left position.
6. A non-transitory computer readable medium storing a bitstream containing encoded data for a block of video data, transform coefficients of the block being generated from the bitstream by processes of:
decoding information about a current block, information about quantized coefficients, information about a quantization parameter, a delta index for an inverse quantization matrix, and delta coefficients of the inverse quantization matrix from a bitstream;
calculating a quantization parameter based on the information about the quantization parameter;
deriving an inference identifier (ID) from the information about the current block and selecting an inverse quantization matrix indicated by the inference ID, wherein a list for the inverse quantization matrix is pre-generated, and an inverse quantization matrix included in the list is generated using a predictive inverse quantization matrix calculated based on the delta index, and delta inverse quantized data generated from the delta coefficients; and
generating transform coefficients from the quantized coefficients using the quantization parameter and the selected inverse quantization matrix,
wherein delta coefficients corresponding to a part of the inverse quantization matrix included in the list are not decoded.