深度摄像机和卡尔曼滤波器在人体目标跟踪中的应用研究  

Application of Depth Camera and Kalman Filter in Human Target Tracking

作　　者：陈锡锻 陈鹏[1] CHEN Xiduan;CHEN Peng(Zhejiang Industry&Trade Vocational College,Wenzhou 325003,China)

机构地区：[1]浙江工贸职业技术学院,浙江温州325003

出　　处：《浙江工贸职业技术学院学报》2022年第1期72-76,共5页Journal of Zhejiang Industry & Trade Vocational College

基　　金：2021年浙江工贸职业技术学院院级课题“基于结构特征匹配与渐进卡尔曼滤波的人体目标跟踪方法研究”(G210101)。

摘　　要：针对人体跟踪过程中因遮挡、姿态变化而导致的跟踪器鲁棒性变差的问题,提出一种基于深度相机和卡尔曼滤波的人体目标跟踪方法,在深度图像下,基于人体骨架关键点,将人体目标分解成若干个关键点的位置状态,分别建立各关键点运动的数学模型,通过卡尔曼滤波器的跟踪,增加了人体目标跟踪不确定性的描述,提高了对遮挡、识别错误等不利因素的鲁棒性。In view of the poor robustness of the tracker caused by occlusion and attitude change in the process of human tracking,a human target tracking method based on depth camera and Kalman filter is proposed.In the depth image,based on the key points of human skeleton,the human target is decomposed into the position state of each key point,and the mathematical models of the movement of each key point are established respectively,Through the tracking of Kalman filter,the description of human target tracking uncertainty is increased,and the robustness to adverse factors such as occlusion and recognition errors is improved.

关 键 词：人体目标 深度相机 卡尔曼滤波 鲁棒性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]