基于色度矢量的图像错误隐藏内插算法

MPEG- 2等图像编码算法对信道干扰是非常敏感的 ,由于错误传播现象 ,1bit的错误将可能使图像质量有明显的下降 .因此 ,接收端常常提供错误隐藏技术 ,来改善图像效果 .该文以 MPEG- 2的视频信号为例 ,根据图像像素间的强相关性 ,对以条为单位出现的图像错误 ,提出了一种空间色差内插算法 ,并推导出了内插像素色差的显示表达式 ,从而可以实现快速实时处理 .另外 ,还对传统的像素内插算法进行了修改 .最后 ,对时间内插算法中的运动矢量进行了推导 .仿真结果表明 ,应用该算法进行丢失像素的内插 ,效果是满意的 .

Chromatism-Vector-Based Error Concealment Techniques for Video Streams

Picture coding algorithms such as MPEG-2 are very sensitive to channel disturbance. A single bit error may cause severe degradation of picture quality due to error propagation, therefore error concealment techniques are required at the receiver. Due to close correlation between pixels, an interpolation chromatism algorithm, which is used to conceal lost information as slices, is presented for MPEG-2. An explicit formula for interpolated pixels is derived in order to be used effectively in real time processing. In addition, a traditional interpolation algorithm is modified. Simulation test shows that this algorithm is satisfactory in interpolating lost pixels. Finally, motion vector estimation is discussed and a formula is derived.