一种基于像素最小距离的视频对象分割算法

随着MPEG-4的发展，视频对象平面分割日渐成为人们所感兴趣的研究课题，为了使所需的视频对象具有更清晰的可视性，视频对象的自动分割就成为关键技术。本文给出了基于统计的自动视频对象分割方法，该方法首先利用统计的方法建立背景模型，然后按照减背景法分割视频对象。同时提出像素最小距离和总体像素最小距离准则来克服由于背景的微小变化而引起的前景对象的错误检测。

An Automatic Video Object Segmentation Algorithm Based on Statistics

With the development of MPEG-4,video object segmentation gradually becomes a study task which people are interested in. In order to make the visibility of video object more clearly, automatic video object segmentation has become a key technology. Based on the statistical algorithm, we set up the model of the background using the statistical method, then segment the video object by the algorithm of abstract background. At the same time, the minimum distance between pixels and the whole MDBP rule are proposed to remove error detection of video object caused by small change of background.