检测图像角点自适应确定跟踪模板的方法  

Method for Detecting Image Corners Using Adaptive Determining Tracking Template

作　　者：王德胜[1] 吴钟建[1] 姚秀娟[1] 金代中[1] 卢宏超 WANG Desheng;WU Zhongjian;YAO Xiujuan;JIN Daizhong;LU Hongchao(Southwest Institute of Technical Physics, Chengdu 610041, China;Military Representative Office of PLA in 209th Institute, Chengdu 610041, China)

机构地区：[1]西南技术物理研究所,四川成都610041 [2]驻209所军代室,四川成都610041

出　　处：《红外技术》2017年第7期638-641,共4页Infrared Technology

摘　　要：基于区域的跟踪方法,其跟踪模板的大小和位置对跟踪结果的影响十分显著,本文通过分析模板视觉显著性特征,提出采用AGAST快速角点检测方法,计算角点间相对距离,分析角点位置分布来自适应确定模板大小和位置,从而提高相关匹配跟踪算法的适应能力。仿真试验结果表明,该方法确定的模板能较好的包含目标显著性视觉特征,并且具有计算速度快,稳健性高,能有效提高采用区域模板匹配方法的跟踪性能。A tracking method based on region,is significantly affected by the size and position of the tracking template.Through the analysis of the template vision,which is a significant feature of the tracking template,the AGAST fast corner detection method is used,which calculates the angle between the relative distances and analyzes the angle point distribution to determine the template size and position.The simulation and test results show that this method can determine the template that contains the target with better significant visual features,has fast calculation speed with high robustness,and can effectively improve the tracking performance of the region template matching method.

关 键 词：视觉特征 角点检测 自适应模板 图像跟踪 

分 类 号：TN911[电子电信—通信与信息系统]