基于改进CamShift的目标跟踪系统算法  

Target Tracking System Based on Improved Camshift Algorithm

作　　者：敖邦乾 杨莎 刘小雍 叶振环 AO Bangqian;YANG Sha;LIU Xiaoyong;YE Zhenhuan(College of Engineering,Zunyi Normal University,Zunyi 563006,China;College of Physics and Electronics,Zunyi Normal University,Zunyi 563006,China)

机构地区：[1]遵义师范学院工学院,贵州遵义563006 [2]遵义师范学院物理与电子科学学院,贵州遵义563006

出　　处：《探测与控制学报》2021年第6期92-100,共9页Journal of Detection & Control

基　　金：遵义市校联合科技研发资金项目资助(HZ字[2018]16号,HZ字[2020]10号,HZ字[2020]16号)。

摘　　要：针对RGB对光照敏感以及运动目标被遮挡时跟踪效果较差等问题,提出基于Canny边缘检测的改进Camshift跟踪算法。该算法首先将跟踪目标的色度直方图和边缘信息进行特征匹配,通过不断迭代计算目标跟踪窗口的大小和位置,然后使用卡尔曼滤波算法对运动目标进行更新预测,动态地搜索运动目标窗口的中心位置。仿真实验结果表明,该算法在跟踪目标与环境颜色相近的情况下具有较强的鲁棒性,同时对于快速运动或被遮挡的跟踪目标也具有较大的性能改进,可用于水面无人艇等运动目标的实时检测跟踪。To solve the problems that the RGB color was sensitive to light and the tracking effect was poor when moving targets were occluded,an improved Camshift target tracking algorithm was proposed based on Canny edge detection,which combined the chromaticity histogram of the tracking target and its edge information to perform feature matching.The window was searched through continuous iteration to recalculate the target size and its position.Then Kalman filter algorithm was used to update and predict the center position of the moving target.Simulation results showed the method was of strong robustness for tracking situations where the tracking target and the environment color were similar,and the tracking performance had an improvement even if the objective was moving fast or was occluded by some other obstacles,which meant it could be adopted for real-time tracking of small targets such as surface unmanned vessels.

关 键 词：特征匹配 搜索窗口 目标跟踪 卡尔曼滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]