Method and apparatus for hierarchical data unit-based video encoding and decoding comprising quantization parameter prediction
A method of decoding a video includes determining an initial value of a quantization parameter (QP) used to perform inverse quantization on coding units included in a slice segment, based on syntax obtained from a bitstream; determining a slice-level initial QP for predicting the QP used to perform inverse quantization on the coding units included in the slice segment, based on the initial value of the QP; and determining a predicted QP of a first quantization group of a parallel-decodable data unit included in the slice segment, based on the slice-level initial QP.
1. A video decoding method comprising:
obtaining a quantization parameter of a slice;
generating a predicted quantization parameter of a current quantization group using the quantization parameter of the slice;
generating a predicted quantization parameter of a next quantization group using the quantization parameter of the slice; and
performing an inverse quantization on the current quantization group and the next quantization group,
wherein the current quantization group is a first quantization group in a current row included in the slice,
the next quantization group is a first quantization group in a next row included in the slice,
each of the current row and the next row comprises a plurality of largest coding units,
wherein a predicted quantization parameter of a quantization group in the current row which is not the current quantization group, is generated using a quantization parameter of a neighboring coding unit of the quantization group in the current row, and
wherein a predicted quantization parameter of a quantization group in the next row which is not the next quantization group, is generated using a quantization parameter of a neighboring coding unit of the quantization group in the next row.
2. The video decoding method of claim 1 ,
wherein the current quantization group is a set of at least one of coding units sharing the predicted quantization parameter of the current quantization group.