自适应分块优化的目标跟踪算法  

Target tracking based on adaptive block extraction

作　　者：杨波 王小虎 YANG Bo;WANG Xiao-hu(School of Finance and Management,Sichuan University of Arts and Science,Dazhou 635000,China;College of Mining Engineering,North China University of Science and Technology,Tangshan 063210,China)

机构地区：[1]四川文理学院财经管理学院,四川达州635000 [2]华北理工大学矿业工程学院,河北唐山063210

出　　处：《计算机工程与设计》2022年第6期1719-1724,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(41755072)。

摘　　要：为解决复杂环境下光照、遮挡等问题对目标跟踪算法的影响,提出基于自适应分块与异步更新的目标跟踪算法。构建光照不敏感特征和超像素自适应分块,通过双约束对分场中具有高置信度的子块进行自适应提取并用于目标跟踪,提高跟踪的精确性,避免全局搜索对运算效率的影响;通过自适应检测和异步更新特征子块,进一步提高算法效率和对背景的抗干扰能力。实验结果表明,与KCF、DSST等已知文献中的算法相比,文中算法具有更优的跟踪准确率,以及对遮挡、光照变化等复杂场景干扰的鲁棒性,验证了算法的有效性。To solve the problems of illumination changes,occlusion and similar interference faced by tracking algorithms in complex environments,a robust target tracking algorithm based on adaptive block segmentation and asynchronous update was proposed.The light insensitive features and super-pixel adaptive segmentation were used in the proposed method,and the double weight constraint was used to adaptively extract high confidence sub blocks for target tracking,so as to avoid the influence of global search and improve the tracking accuracy.Through occlusion detection and asynchronous update,the efficiency of the algorithm and the ability to filter background information were further improved.The experimental results show that,compared with the algorithms in known literatures such as KCF and DSST,the proposed algorithm has better tracking accuracy,as well as the robustness to complex scene interference such as occlusion,lighting changes,and so on,verifying the effectiveness of the proposed algorithm.

关 键 词：目标跟踪 局部敏感直方图 光照不敏感特征 超像素自适应分块 自适应遮挡检测 子块异步更新 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]