摘要
多媒体通信的发展迫切需要面向对象的视频编解码技术 ,其关键是视频对象 (VO)的提取和视频对象平面(VOP)的分割 .现提出一种自动分割 VOP的新方法 .根据初始帧 ,使用形态运动滤波技术 ,初始运动对象首先被提取出 ,在后继帧中使用活动轮廓 (Snake)技术对运动对象进行拟合匹配 ,最后根据一系列精确的二值轮廓导出运动对象的提取 .实验结果表明 ,该算法可有效地提取视频对象和分割视频对象平面 .除了个别参数需要调节外 ,整个过程是自动完成的 .
Object-oriented video coding is important for the development of multimedia communication, in which extraction of video objects and segmentation of video object planes are key techniques. Initial moving objects are extracted from original frames by using a morphological motion filter. The subsequent frames are then matched using an active contour model(snake)method. Finally, moving objects are extracted with accurate binary contours. Experiments show that the described algorithm can extract video objects, leading to the segmentation of VOPs. This process can be completed with manual adjustment of only a few parameters.
出处
《上海大学学报（自然科学版）》
CAS
CSCD
2001年第1期1-6,共6页
Journal of Shanghai University:Natural Science Edition
基金
上海市自然科学基金!项目 (992 D14 0 35 )
上海市教委重点项目! (s990 12 3)
