首页 | 官方网站   微博 | 高级检索  
     

适用于遮挡问题的目标跟踪算法
引用本文:赵建伟,刘重庆. 适用于遮挡问题的目标跟踪算法[J]. 计算机辅助设计与图形学学报, 2003, 15(9): 1091-1095
作者姓名:赵建伟  刘重庆
作者单位:1. 上海广电(集团)有限公司中央研究院,上海,200032
2. 上海交通大学图像处理与模式识别研究所,上海,200030
摘    要:提出一种基于网格模型的目标跟踪算法.该算法首先进行遮挡区域检测,然后进行网格结点的运动估计和网格更新过程完成目标的多帧跟踪.改进的遮挡区域检测算法有效地提高了检测准确度,从而确保遮挡区域的准确跟踪;网格结点的运动估计是通过特征窗口运动补偿匹配完成,可以有效地克服块效应.实验证明,该算法解决了二维运动估计时网格模型在遮挡区域存在的问题,并可以有效地进行目标准确跟踪.

关 键 词:计算机视觉 图像处理 模式识别 遮挡问题 目标跟踪算法 网格模型
修稿时间:2002-07-09

Object Tracking Algorithm Applied to Occlusion Issue
Zhao Jianwei ) Liu Chongqing ) ). Object Tracking Algorithm Applied to Occlusion Issue[J]. Journal of Computer-Aided Design & Computer Graphics, 2003, 15(9): 1091-1095
Authors:Zhao Jianwei ) Liu Chongqing ) )
Affiliation:Zhao Jianwei 1) Liu Chongqing 2) 1)
Abstract:A novel object tracking algorithm based on mesh model is proposed to perform accurate object tracking while occlusion problem exists. Occlusion detection is done first, and then mesh nodes motion estimation and mesh update are carried to achieve object tracking in several frames. An improved occlusion detection method is applied to enhance the detection precision. Mesh nodes motion estimation is performed through feature window motion compensation matching, which can overcome block artifacts. Experimental results indicate that the proposed algorithm can solve occlusion problems existed in 2D motion estimation, and can accurate track object effectively.
Keywords:occlusion regions  mesh model  motion estimation  object tracking
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号