基于置信度更新的目标跟踪算法  

Target Tracking Algorithm Based on Confidence Update

作　　者：岳阳 王国刚 YUE Yang;WANG Guogang(Shenyang University of Chemical Technology,Shenyang 110142,China)

机构地区：[1]沈阳化工大学信息工程学院,辽宁沈阳110142

出　　处：《沈阳化工大学学报》2024年第1期71-76,共6页Journal of Shenyang University of Chemical Technology

基　　金：国家重点研发计划项目(2018YFB1700200)。

摘　　要：针对相关滤波跟踪算法在目标跟踪过程中的目标遮挡、环境复杂等问题,基于SAMF跟踪器提出一种基于置信度更新的目标踪算法.在相关滤波跟踪框架下,采用置信度指标(APCE)判断目标遮挡与否,提出自适应模板更新以解决不同环境下视频场景中跟踪置信度阈值的适应性.在OTB-2013视频序列数据中集中选取33个遮挡和背景复杂属性视频序列进行实验验证,实验结果表明:该算法相对SAMF算法而言,在跟踪遮挡和背景复杂属性下的成功率分别提升1.6%和2.3%,识别精度较高,能够在目标遮挡和复杂背景等环境下稳定跟踪目标.To solve the problems of target occlusion and complex environment during target tracking,a new target tracking algorithm based on confidence updating was proposed based on SAMF tracker.Under the framework of correlation filtering and tracking,the confidence index(APCE)was proposed to judge whether the target is blocked or not,and the adaptive template updating was proposed to solve the adaptability of the tracking confidence threshold in video scenes under different environments.The OTB-2013 video sequence data set attribute selection 33 shade and background video sequences experiment.The experimental results showed that compared with the SAMF,algorithm in tracking shade and complex background attribute the success rate is increased by 1.6%and 2.3%respectively under the tracking occlusion and complex background attributes.The algorithm had higher identification accuracy,and track targets stably under the environment of target occlusion and complex background.

关 键 词：目标跟踪 置信度 相关滤波 目标遮挡 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]