In this paper, efficient one-dimensional (1-D) fast integer transform algorithms of the DCT matrix for the H.265 stan-dard is proposed. Based on the symmetric property of the integer transform matrix and the matrix op...In this paper, efficient one-dimensional (1-D) fast integer transform algorithms of the DCT matrix for the H.265 stan-dard is proposed. Based on the symmetric property of the integer transform matrix and the matrix operations, which denote the row/column permutations and the matrix decompositions, along with using the dyadic symmetry modification on the standard matrix, the efficient fast 1-D integer transform algorithms are developed. Therefore, the computational complexities of the proposed fast integer transform are smaller than those of the direct method. In addition to computational complexity reduction one of the proposed algorithms provides transformation quality improvement, while the other provides more computational complexity reduction while maintaining almost the same transformation quality. With lower complexity and better transformation quality, the first proposed fast algorithm is suitable to accelerate the quality-demanding video coding computations. On the other hand, with the significant lower complexity, the second proposed fast algorithm is suitable to accelerate the video coding computations.展开更多
文摘In this paper, efficient one-dimensional (1-D) fast integer transform algorithms of the DCT matrix for the H.265 stan-dard is proposed. Based on the symmetric property of the integer transform matrix and the matrix operations, which denote the row/column permutations and the matrix decompositions, along with using the dyadic symmetry modification on the standard matrix, the efficient fast 1-D integer transform algorithms are developed. Therefore, the computational complexities of the proposed fast integer transform are smaller than those of the direct method. In addition to computational complexity reduction one of the proposed algorithms provides transformation quality improvement, while the other provides more computational complexity reduction while maintaining almost the same transformation quality. With lower complexity and better transformation quality, the first proposed fast algorithm is suitable to accelerate the quality-demanding video coding computations. On the other hand, with the significant lower complexity, the second proposed fast algorithm is suitable to accelerate the video coding computations.
