IMPROVEMENT FOR SCALABLE MOTION VECTOR CODING
Nowadays, a lot of wavelet-based video codecs can achieve performance comparable with the sate-of-art h. 264 and mainly provide three scalabilities concerning spatial resolution, temporal resolution and video quality (also known as SNR). Therefore, in order to achieve the best codec performance, the motion information should be coded in a scalable way corresponding to the texture information given. This paper proposes a scheme to improve the coding performance of the scalable motion vectors (MVs) generated from HVSBM, in which the redundancy of the block splitting information is exploited and the motion correlation within the same temporal layer is also considered. At the same time, the characteristic of motion vector itself is exploited, and each hit- plane is further separated into four channels according to the block size of the MVs. The scheme proposed is able to achieve more efficient scalable motion vector bitstreams saving bits up to 33％, and the video coding performance at low bit rate and low resolution is consequently improved.
motion vector coding scalable video coding MC-EZBC
Xiangfu Liu Janru Xue Xing Xing Shugang Guo
Institute of Artificial Intelligence and Robotics, Xian Jiaotong University, Xian, Shaanxi China, 710049
国际会议
开封
英文
387-393
2006-10-15（万方平台首次上网日期，不代表论文的发表时间）