基于帧间运动估计的视频背景缺失信息恢复  

Lost Video Background Information Recovering Based on Motion Estimation

作　　者：罗三定[1] 孙薇嘉[1] 沙莎[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083

出　　处：《计算机工程》2006年第8期235-237,251,共4页Computer Engineering

摘　　要：针对视频图像背景受物体遮挡和视野局限造成图像信息丢失的问题,根据序列图像的运动特性,提出了一种基于帧间运动估计的动态视频缺损信息恢复方法,6参数仿射模型的建立可克服因拍摄角度变化产生的视频图像变形,进行准确的补偿,与MPEG-4中的Sprite图像相比,既恢复了缺失信息,又保证了视频的动态特性,具有较好的实用价值。Aimed at the problem of the lost information caused by the covering on the video image or the confining of the visual field, the paper proposes a method of the lost video information recovering based on motion estimation. It can conquer the image distortion caused by the changing shooting angle by the affine model with 6 parameters. Compared with the Sprite image in MPEG-4, it can not only recover the lost information but also maintain the dynamic characteristic of video. The result shows that it is provided with significant value of application.

关 键 词：视频处理 运动估计 信息恢复 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]