基于帧间运动的视频分割  

Video segmentation based on inter-frame motion

作　　者：白雪生[1] 金国英[1] 徐光祐[1] 史元春[1] 

机构地区：[1]清华大学计算机科学与技术系,北京100084

出　　处：《清华大学学报（自然科学版）》2002年第1期76-79,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家"八六三"高技术项目 ( 86 3-30 6 -ZT0 4-0 2 -1及 86 3-30 6 -QN99-0 1)

摘　　要：视频分割是视频结构化和检索的重要技术 ,目前主要通过镜头分割得到。但许多应用需要对镜头根据摄像机运动作分割 ,现有技术不能满足应用实时需要。提出了一种对P帧运动向量稳健分析计算帧间运动并进行分割的方法。与传统光流法、M估计法比较 ,该算法只在压缩域上进行 ,不需迭代 ,具较强实时性。文中阐述了方法的基本思想、理论依据与实现 ,实验表明Video segmentation is an important technology for video structuring and retrieval, with most research work now focused on shot detection. However, many applications need the shots to be further segmented based on camera operational changes, which current algorithms can not do in real time. This paper presents a video segmentation method based on a robust analysis of PMV and inter frame motion. Compared with traditional optical flow and M estimation methods, the proposed algorithm only calculates in the DC domain and does not iterate, so it can segment almost in real time. The theoretical basis and the algorithm implementation are explained in detail, with experiments to show the satisfactory results from the algorithm.

关 键 词：视频分割 帧间运动模型 帧间位移向量 稳健分析 视频结构化 视频检索 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]