基于多特征自适应融合的粒子滤波目标跟踪算法  

Target Tracking Algorithm Based on Particle Filter with Adaptive Fusion of Multiple Feature

作　　者：董娜 刘军[1] DONG Na;LIU Jun(College of Mechanical and Electronical Engineering,Lanzhou University of Technology,Lanzhou 730050)

机构地区：[1]兰州理工大学机电工程学院,兰州730050

出　　处：《计算机与数字工程》2020年第8期1919-1923,共5页Computer & Digital Engineering

基　　金：国家自然科学基金项目“大规模非相似混杂制造系统缓冲区容量优化分配技术研究”(编号:71861025)资助。

摘　　要：针对传统粒子滤波算法在应对光照变化、尺度变化、较大形变时存在的跟踪稳定性差的缺陷,提出了多特征自适应融合的粒子滤波目标跟踪算法。该算法在进行模型匹配时,采用了动态分层融合策略将各特征的匹配权值自适应地进行融合;另外该算法还利用帧差法的实时检测性检测出运动物体的运动区域来充盈粒子的多样性;最后,利用模板更新策略来实时更新目标模板。实验结果表明,论文算法对光照变化、尺度变化、部分遮挡有着较高的鲁棒性。Aiming at the defect of poor tracking stability of traditional particle filter algorithm in dealing with illumination change,scale change and partial occlusion,a particle filter target tracking algorithm with adaptive fusion of multiple feature is pro⁃posed.In the process of model matching,dynamic hierarchical fusion strategy is adopted to adaptively fuse the matching weights of each feature.In addition,the real-time detect ability of frame difference method is used to detect the moving areas of moving objects to fill the diversity of particles.Finally,the template update strategy is used to update the target template in real time.The experi⁃mental results show that the proposed algorithm is robust to the phenomenon with illumination change,scale change and partial oc⁃clusion.

关 键 词：多特征自适应融合 动态分层融合策略 目标跟踪 粒子滤波 帧差法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]