英文摘要 | In this paper, rate control schemes and video transcoding techniques were studied and the following outcomes were put forward in the research: 1.An integral rate control algorithm. In TM5 rate control algorithm, the conceptions and methods of complexity analysis, bit allocation, rate control, and adaptive quantization are introduced to control the bit rate of compressed video accurately. In this paper, the problem of normalized equation in adaptive quantization equation was discussed and the problem adaptive quantization doesn’t couple well with the rate control equation was analyzed. Through analyzing the equation of rate control, an integral rate control strategy was obtained. At last, an improvement with lower complexity was proposed, which made a big gain. 2.A low-complexity macroblock level rate control algorithm. The JVT-G012 rate control scheme of H.264 has the problem of high complexity. In this paper, to reduce complexity, the integral rate control was adopted, and a macroblock level low-complexity rate control scheme of H.264 was proposed. The PSNR of new rate control algorithm was similar to JVT-G012’s while being competent for both VBR case and CBR case. 3.Mixed MV mapping scheme for video transcoding. To utilize various MCP (Motion Compensated Prediction) modes of H.264 and speed up the transcoding while achieving good video quality in reducing resolution transcoding, a mixed motion vector mapping scheme for MPEG-2 to H.264 transcoder was proposed. Based on the variance of the input motion vectors, this scheme selected the best one from 4:1, 1:1, 2:1 mapping modes, and then mapped MVs into one MV by median value filter. 4.FME algorithm for video transcoding. In MPEG-2 to H.264 cascaded transcoder, the computational complexity of H.264 encoder is high, especially the ME module of encoder cost much time. In this paper, in order to reduce complexity and accelerate transcoding, a FME algorithm for video transcoding was suggested. Firstly, this algorithm utilized a new two-diamond MVR technique to refine on the input MV, which achieved fast transcoding speed. Then, considering the various MCP modes in H.264, this algorithm combined the MVR technique with UMHexagon FME algorithm and achieves good video quality. 5.Designing and implementing a video transcoding system. In this paper, the researcher utilized the techniques proposed above, adopted the cascaded transcoding architecture reusing input MV, designed a software-based MPEG-2 to H.264 video transcoding system, and realized the functions of reducing bit rate, reducing resolution, and reducing frame rate. Meanwhile, the researcher selected the right MPEG-2 decoding and H.264 encoding software, and then optimized the cost-time module of transcoder. This system has high efficiency and makes good video quality, can be used in areas such as IPTV. |
修改评论