摘要
为了实现多视点视频编码的低时延随机访问,优化多视点视频随机访问性能,提出一种改进的多视点视频编码方法。该方法首先修改anchor帧图像视点间的预测关系,将视差补偿预测应用于anchor帧图像,而对non-anchor帧图像仅采用运动补偿预测;其次,根据视点间图像的运动信息具有高度相关性原理,对non-anchor帧图像应用基于自适应运动矢量精细化的Motion skip模式来优化其运动补偿预测性能。实验结果表明,文中所提的多视点视频编码方法在保证高压缩效率的前提下,具有较低的计算复杂度,并提高了视点的随机访问性能。
In order to achieve low-delay random access capability, a new multi-view coding scheme is proposed and evaluated. Firstly, inter-view correlation of anchor pictures is changed and only motion compensated prediction (MCP) is employed for non-anchor pictures. Secondly, according to high correlation of the inter-view motion information, a motion skip based adaptive fine-granular motion algorithm is exploited between non-anchor pictures, to optimize the performance of MCP. Experimental results show that the proposed algorithm can reduce computational complexity and improve random access performance, but preserving the compression performance at the same time.
出处
《南京邮电大学学报(自然科学版)》
2010年第3期74-78,共5页
Journal of Nanjing University of Posts and Telecommunications:Natural Science Edition
基金
江苏省自然科学基金(BK2007238)
江苏省高校重大基础研究计划(07KJA51006)
中兴基金(08外45)资助项目