Picture decoding method for decoding coded picture data and performing distortion removal by comparing pixel difference values with threshold
A picture decoding method is provided for decoding coded picture data. The decoding method includes decoding coded picture data to obtain a first transform block and an adjacent second transform block in a reconstructed picture having a plurality of blocks forming a moving picture image. The decoding method also includes determining whether the first transform block is located inside a motion-compensation block in which the second transform block is located. The decoding method further includes performing a filtering on pixels in a block boundary between the first and second transform blocks when the first transform block is located outside the motion-compensation block in which the second transform block is located, and performing no filtering on the pixels in the block boundary between the first and second transform blocks when the first transform block is located inside the motion-compensation block in which the second transform block is located.
1. A picture decoding method for decoding coded picture data, which is included in a received video stream, said method comprising:
decoding coded picture data in the received video stream to obtain a first transform block and an adjacent second transform block in a reconstructed picture having a plurality of blocks forming a moving picture image, said first and second transform blocks being a unit for which a frequency transformation is applied and being motion-compensated, and the first and the second transform blocks having a quantization parameter and a coded motion compensation error equal to zero;
determining whether the first transform block is located inside a motion-compensation block in which the second transform block is located, the motion-compensation block being a unit for which a motion compensation is applied;
performing a filtering, using a filter, on pixels in a block boundary between the first and second transform blocks when the first transform block is located outside the motion-compensation block in which the second transform block is located; and
performing no filtering on the pixels in the block boundary between the first and second transform blocks when the first transform block is located inside the motion-compensation block in which the second transform block is located,
wherein, in a case of performing a filtering, using a filter, on pixels in a block boundary between the first and second transform blocks when the first transform block is located outside the motion compensation block in which the second transform block is located, said performing a filtering on pixels, using a filter, includes
obtaining a pixel difference value which is a difference between a pixel value of one pixel adjacent to the block boundary from the first transform block and a pixel value of one pixel adjacent to the block boundary from the second transform block;
comparing said pixel difference value with a threshold value which is determined based on said quantization parameters of the first and second transform blocks; and
not removing coding distortion in the pixels in the block boundary between the first and second transform blocks when said pixel difference value is greater than said threshold value,
wherein a block size of the motion-compensation block is smaller than a block size of the macroblock, and a block size of the motion-compensation block is greater than a block size of each of the first and second transform blocks, and said first and second transform blocks having first and second quantization parameters, respectively, and said quantization parameter is an average of said first and second quantization parameters, and said first and second quantization parameters have a same value when the first and second transform blocks are included in a same macroblock, and said first and second quantization parameters have a different value when the first and second transform blocks are included in a different macroblock, and
wherein a coded motion compensation error equal to zero means that transform coefficients in a 4×4 block that has been inter coded are all zero.