摘要
提出一种利用图像差分、局部光流法和目标匹配,实现视频对象目标提取和跟踪的方法.首先,对预处理后的图像进行差分;在差分的基础上,运用局部光流场对目标进行运动估计;对估计结果依据相似度进行聚类分析,消除小的非目标区域:当目标未发生剧烈变化时,更新目标模板,否则,利用已有目标模板进行匹配跟踪.实验结果验证了该算法的有效性,并实现了目标变形和部分遮挡时的稳定跟踪.
A method on detection and tracking of moving object is proposed based on the method of image difference,partial optical flow and objectives matching.As preprocessing,the image difference is calculated.By the optical flow computation of the difference image the moving target is primitively estimated to get a binary image. With the clustering method the small blocks in the binary image,which do not belong to target are eliminated.If the shape of the detected target doesn't change largely,the detected target in the last frame is taken as a template.Then the shape target is corrected by a matching technology.By it the disturb of the target blocking,deformation or other moving objects on the target can be eliminated.The experiment results validate the effect of our method.
出处
《红外与激光工程》
EI
CSCD
北大核心
2006年第z4期312-315,共4页
Infrared and Laser Engineering
关键词
图像差分
光流法
目标跟踪
聚类分析
Image differential
Optical flow
Object tracking
Clustering analysis
