An efficient architecture of multiple 8×8 transforms for H.264/AVC and VC-1 decoders

Abstract

This paper proposes an efficient architecture, which can perform multiple 8x8 transforms for both H.264/AVC and VC-I decoders. The hardware design which supports multiple standards becomes more and more important. By designing a unique data flow for VC-I 8x8 inverse transform, the H.264/AVC and VC-I 8x8 inverse transforms are realized in a hardware sharing architecture. The proposed multiple transforms architecture contains fast one-dimensional (I-D) transforms and rounding operations. Simulation results show the proposed architecture takes 6,702 gates which are much less than the individual designs for the H.264/AVC and VC-I 8x8 inverse transforms.I. IN1RODUCTIONThe H.264/AVC standard [1] was jointly proposed by the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Expert Group (MPEG). H.264/ A VC, which attains a higher compression performance with a lower bit rate and higher quality, adopts a number of new special features to compress video much more effectively than past standards. Experiments show that H.264/AVC can achieve over 50% bit rate savings while achieving the similar quality compared to other existing standards such as MPEG-2 and MPEG-4. VC-l [2] is a video codec specification that has been standardized by the Society of Motion Picture and Television Engineers (SMPTE) and implemented by Microsoft as Microsoft Windows Media Video (WMV) 9. For computational complexity, VC-l decoder is twice more efficient than H.264/ A VC one.Various standards consist of unique coding features, including compression ratio, computational complexity, video quality, and so on. To expand its applications, the hardware codec design, which supports several coding standards become the trend in consumer markets. H.264/AVC and VC-l both provide 8-point integer transforms to attain higher compression ratio. The transform matrices of H.264/ A VC and VC-l 8-point transforms are much distinct. In order to achieve low-cost designs, the hardware sharing architecture for both H.264/AVC and VC-l 8-point transforms is necessary rather than individual transform designs.There are many systems which designed the transform architecture for various applications. For the H.264/AVC encoder, Amer et al.[3] investigated the fast 8-point transform

Low Cost 1D DCT Core for Multiple Video Codec

“…Based on the similarity in matrix operation, Ref. [8][9][10][11][12][13][14][15][16] developed their own algorithms and architectures to compute 8×8 DCT. However, most of them were designed to realize specified standards and did not take different standards compatibility into their own consideration.…”