摘要
针对由于信道误码导致帧内编码图像中出现的错误块,提出频域和空域相结合的掩盖算法.该算法基于图像局部平滑性和边界连续性,首先以相邻正确图像块的低频系数估计该错误块的直流分量并进行离散余弦反变换,实现对错误块的初步恢复.同时利用Sobel梯度算子检测相邻正确块中边界信息,进而估计错误块内的边界方向及其幅度.以马尔可夫随机场作为图像的先验模型,并基于最大后验估计实现对初步处理后的错误块进行空域优化.实验结果表明,和同类方法相比,该算法不仅能得到更好的掩盖图像质量,而且其计算量大幅度地减少.
An algorithm was presented to conceal the errors in I frame for video communications. Firstly, the algorithm estimates DC value of the error block based on its neighborhoods, sets all AC values to 0, and then does IDCT. So the error block can be primarily concealed. Secondly, the algorithm optimizes the block using statistical methods in spatial domain.Compared to traditional methods, the experimental results show that the image quality concealed by the proposed algorithm is significantly improved and the computational complexity is greatly decreased.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2004年第4期606-609,共4页
Journal of Shanghai Jiaotong University
基金
上海市教委青年基金资助项目(03GQ22)
关键词
视频通信
错误掩盖
马尔可夫随机场
video communication
error concealment
Markov random field (MRF)