【摘要】 针对H.264压缩域的运动对象分割问题,提出了一种基于Fisher准则的运动对象分割方法。首先将运动对象中每个宏块中子块的运动矢量进行加权,获取每个宏块的运动矢量,然后利用多帧的运动矢量进行累加,获取整个宏块的运动矢量,最后使用Fisher准则将运动对象的矢量分类,从而获取H.264压缩域中的运动对象。通过实验证明,在基本满足运动对象分割准确率的前提下,达到了较好的实时性。更多还原

【Abstract】 For the purpose of extracting moving object from H.264 compressed video,a extracting moving method based on Fisher criterion is presented.Firstly,the paper retrives the moving vectors(MV) of sub block in the Macro Block(MB) which was 16x16 pixels in P frame,and the moving vector of each sub block is weighted in each MB.The each moving vector of the same MB is also weighted in the inter frames.Finally,the weighted moving vector of each MB is retrived from the P frame and is judged by the Fisher criterion.The experimental results show that the proposed approach can quickly extract moving object from H.264 compressed video,which guarantees extracting quality.更多还原