Adaptive filtering based upon boundary strength
Adjacent regions are identified in an image. Coding parameters for the adjacent regions are identified. Selective filtering is performed at the region between the identified adjacent regions.
1. An image decoding method filtering a boundary between two adjacent blocks in a reconstructed image selectively, comprising:
a motion compensation prediction step for conducting motion compensation prediction for each of blocks to be decoded by using the reconstructed image,
an inverse transformation step for conducting inverse orthogonal transformation for the data of the blocks to be decoded, and
a determination step for determining a filtering strength and whether or not to conduct filtering, with respect to each of the boundaries, wherein
the determining step determines:
(1) filtering is conducted with a first strength when at least one of the two adjacent blocks is intra-coded,
(2) filtering is conducted with a second strength that is weaker than the first strength when both of the two adjacent blocks are not intra-coded and a non-zero transformation coefficient is coded in at least one of the two adjacent blocks,
(3) filtering is not conducted when both of the two adjacent blocks are not intra-coded, a non-zero transformation coefficient is not coded in both of the two adjacent blocks, the two adjacent blocks are predicted by the same reference frame, and an absolute value of a difference between motion vectors of the two adjacent blocks is smaller than a specified threshold value.
2. An image decoding apparatus filtering a boundary between two adjacent blocks in a reconstructed image selectively, comprising:
a motion compensation prediction means for conducting motion compensation prediction for each of blocks to be decoded by using the reconstructed image,
an inverse transformation means for conducting inverse orthogonal transformation for the data of the blocks to be decoded, and
a determination means for determining a filtering strength and whether or not to conduct filtering, with respect to each of the boundaries, wherein
the determining means determines:
(1) filtering is conducted with a first strength when at least one of the two adjacent blocks is intra-coded,
(2) filtering is conducted with a second strength that is weaker than the first strength when both of the two adjacent blocks are not intra-coded and a non-zero transformation coefficient is coded in at least one of the two adjacent blocks,
(3) filtering is not conducted when both of the two adjacent blocks are not intra-coded, a non-zero transformation coefficient is not coded in both of the two adjacent blocks, the two adjacent blocks are predicted by the same reference frame, and an absolute value of a difference between motion vectors of the two adjacent blocks is smaller than a specified threshold value.