视频传输中基于Gibbs随机场模型的错误隐藏方法  

Error Concealment Method Based on Gibbs Random Field Model in Video Communications

作　　者：余志雄[1] 孙洪[1] 

机构地区：[1]武汉大学电子信息学院,湖北武汉430079

出　　处：《小型微型计算机系统》2006年第10期1948-1951,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(60372057)资助.

摘　　要：恶劣的信道传输质量将会引起数据丢包和误码传输等现象,严重影响视频图像质量.为了提高错误图像质量,本文先利用帧间错误隐藏方法恢复部分变化量小的数据块,再利用差错块的相邻区域方向梯度分布,建立一个Gibbs随机场模型,并通过迭代求出剩余差错块.作为一种优质的图像建模方法,Gibbs随机场利用差错块周围的已知像素信息来模拟丢失的图像区域,该区域与邻域图像有同样的分布参数,并能在边界处与邻域图像和谐的融合在一起.实验结果表明,与同类算法比较,该算法视频恢复的效果良好,并且计算复杂度比较简单.Bad signal channel transmission quality leads to data loss and bit transmission errors, which greatly affect video image quality. In order to improve error image quality, an inter-frame method is firstly used to recover some data blocks with small movement; then according to the directional gradient distribution of neighboring blocks, this paper establish a Gibbs Random Field （GRF） model and recover the remain blocks by iteration. As a hiLgh quality image model method, GRF uses the pixels around the missing image blocks to estimate the missing blocks. The estimated blocks have same distribution parameters with the neighboring blocks, and can be well connected with other blocks along the edges. Experiments show that video quality is improved and the computational complexity of this method is quite acceptable.

关 键 词：视频通信 错误隐藏 双线性插值 GIBBS随机场 马尔可夫随机场 

分 类 号：TN911.73[电子电信—通信与信息系统]