基于动静背景下的视频对象自适应提取算法

Adaptive Algorithm of Video Object Segmentation Under Moving and Static Background

在MPEG-4标准中提出了VOP(Video Object Planes)的概念,可以通过VOP实现基于内容的操作。这样,从场景中自动地分割出视频对象就成为基于对象视频编码的先决条件。为此,提出一种基于动静背景下的视频对象自适应提取算法。该算法可以自动提取动态背景、静态背景,也可以在视频序列中出现背景和对象都停止变化时实现视频对象的提取。同时,该算法引入了视频前景和背景预处理,克服了由于视频图像对比度低造成的提取出的部分视频对象轮廓不完整的问题。

The VOP （Video Object Planes） concept was put forward in the MPEG-4 （Moving Picture Experts Group-4） standard. The content-based operation will be realized through VOP. Video object segmentation is a precondition that realizes video object coding. In this paper, a new algorithm of video object segmentation is presented. The algorithm can process video sequence automatically with moving background, static background or the video sequence that both object and background are all static. Using preprocess, the algorithm overcomes the problem of contour incompletion at the same time.