The Optimization And Application Of Mutiview Stereoscopic Video Decoding Algrithm

In recent years, with the rapid development of computer science and video codec technology, the demand of natural reproduction of video scenes for human becomes possible. Multi-view stereo videos, which are of great adaptability to the viewing environment and with realistic visual effects, are getting more and more attentions and therefore have become one of the research focuses in the field of video research. Compared to flat videos, the amount of data of multi-view stereo videos increase rapidly with the number of viewpoints increases. Thus, it is difficult for multi-view stereo videos to store or transfer, which becomes a bottleneck of the multi-view stereo videos applications in industry. H.264 is a new generation video compression and depression standard. Despite of its higher compression ratio and better channel adaptation, it also has extensively application prospect.Although the H.264 standard has excellent performance in compression efficiency, but its test model’s decode speed is not very satisfactory. Therefore, it is playing a great impact on the playing fluency of Multi-view stereo videos. In this paper, we optimized the decoding part in the H.264 test model. As in the original test model of H.264, it goes through the whole codeword table to look up a codeword, which becomes a main reason of low efficiency. Based on the decoder principles of H.264/MVC, this paper provided a more efficient table look-up algorithm for CAVLC entropy decoding, and transplanted the optimized decoder to the player, using multi-thread to coordinate the relationship of decoding and synthetic display, making video playback more smoothly.Experimental results show that the proposed algorithm narrows the table look-up scope, improves the table look-up speed by about 70% and the overall decoding speed by 5.9%, and ultimately achieves the decoding and playing functions of 8-view stereo video in 264 format.