视频传输差错时域掩盖方法研究

Study on Temporal Error Concealment for Video Transmission

为了克服数字视频在铁路现场传输易出现信息丢失的问题,对视频差错掩盖技术进行了研究,提出一种对付视频传输差错的时域掩盖方法。将参考帧宏块(Macroblock)的运动向量依照其运动趋势外推,评估图像受损区域与这些外推预测宏块之间的相关性,估算出受损块的运动向量,实现受损图像区域的掩盖。为克服物体边界不正确替代的缺点,将所提出的算法与改进的边界匹配(BoundaryMatching)算法相结合实现了较为平滑的掩盖。基于H.263+的仿真实验表明,在间隔丢失GOB(GroupofBlock)或整帧图像完全丢失情况下,可以取得较为满意的效果。

To overcome the information lost when the compression video is applied in railroad field remote transmission, a novel temporal concealment method that combats the error effect is proposed. This technique extrapolates motion vectors of macroblocks from the reference frame to corrupted frame, and the motion extrapolation MBs are established. Base on the overlapped areas between damaged areas and the motion extrapolation areas, the best estimation motion vector of the corrupted block is obtained. Then, the corrupted block is concealed with the relative block in the reference frame. This method overcomes the disadvantage of temporal replacement, and the incorrect MB displacement at the object border when combined with the modified boundary matching algorithm. With one entire frame lost or interlace GOBs(Group of Block) lost and an H.263+ codec, simulation results, show the satisfactory concealment performance.This project is supported by the Science and Technology Development Plan Project under the Ministry of Railways.