【作者】 黄杰华； 李德强；

【Author】 HUANG Jie-hua~(1,2)LI De-qiang~1 (1.Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyaag 110016,China;2.Graduate University of Chinese Academy of Sciences,Beijing 100080,China)

【机构】 中国科学院 沈阳自动化研究所；

【摘要】 目标跟踪是计算机视觉中的一个重要课题。文中提出一种方法,用于解决跟踪过程中由于目标的运动、变形和外貌变化所引起的跟踪困难问题。该方法基于灰度差平方和,通过优化迭代, 求解出当前图像中目标的位置,且考虑利用了目标的先验知识,用初始模板和更新模板同时跟踪, 再通过选择误差小的结果作为最终的跟踪目标。仿真实验的结果表明,该方法对于目标的运动、变形和外貌变化以及遮挡,能更稳定地实现跟踪。更多还原

【Abstract】 Target tracking is a very important issue in computer vision.Due to the object motion, shape deformation and object appearance change,target tracking is very complex.A method is proposed to solve the problem.Based on sum of image gray square difference,the method iteratively found the object position in the image.Considered the object knowledge,it tracked the object by both initial template and update template,and then regarded the result with the minor error as the object.The experiments show the proposed method is more robust to target tracking.更多还原