一种基于快速搜索的视频时域差错隐藏算法

为了克服视频传输中因传输差错引起的视频质量下降,提出一种基于快速搜索的边框匹配时域隐藏算法(CSBM)。该算法针对时域差错隐藏的运动矢量恢复问题,利用边框匹配算法改善被恢复的物体边缘模糊的情况;采用基于中心偏置的快速搜索样式得到最小边界匹配差值的候选运动矢量,减少了解码器差错恢复的计算复杂度。实验结果显示,针对不同性质的序列,该算法与边界匹配算法(SMA)、棱形搜索的边界匹配法(DSSM)等典型差错隐藏方法相比,平均搜索点数可减少12.5~19个点,亮度分量的峰值信噪比(PSNR)能改善0.93~1.55 dB,证明该算法能获得更好的差错隐藏效果,并减少了运算量。

A Temporal Error Concealment Algorithm Based on Fast Searching

Compressed video is vulnerable to transmission errors over unreliable channels.In this paper,a center-biased searching boundary-matching algorithm(CSBM) for video error concealment is proposed to combat the transmission errors.Aiming at motive vector resilience of temporal error concealment,boundary matching is used to improve object verge faintness,and the search pattern based on center-biased characteristics of motion vector(MV) is utilized to get the minimum boundary-matching differences of the MV candidate so as to reduce the computational complexity of the decoding error resilience.Compared with the traditional algorithms such as SMA,DSSM,the experiment results show that it can reduce 12.5 to 19 computing points and can improve 0.93 to 1.55 dB of the PSNR,which proves its better concealment performance and lower computation complexity.