光学与光电技术, 2017, 15 (2): 94, 网络出版: 2017-05-09
一种基于中心距离加权的快速相关跟踪算法
A Rapid Correlation Tracking Algorithm Based on Weight Distance
目标跟踪 Kalman滤波 相似性度量 快速搜索 目标遮挡 target tracking Kalman filter similarity measurement rapid search strategy object occlusion
摘要
针对跟踪过程中目标被遮挡的问题,提出了一种基于中心距离加权的快速相关跟踪算法,定义了基于灰度差异和距离加权的相似性度量,并根据相关系数自适应更新目标模板,提高了目标的跟踪稳定性和定位精度,结合多尺度的高斯金字塔快速搜索策略,加快了目标匹配速度,在目标遮挡过程中采用Kalman滤波算法预测目标位置,目标退出遮挡时可以重新捕获。实验结果表明,该方法能较好地抑制噪声和目标遮挡的影响,目标遮挡重新捕获成功率高于85%,目标跟踪精度优于10个像素。该算法可有效地跟踪复杂地面背景的运动目标。
Abstract
To solve the occlusion problem of object tracking, a rapid correlation tracking algorithm based on weight distance is proposed in this paper. A similarity measurement is defined based on weight distance and grayscale differences between object template and the reference image area. Adaptive template update strategy which can adjust the correlation coefficient is used to improve the stability and tracking accuracy. Multi-scale Gaussian pyramid strategy is used to speed target up matching. Kalman filter is applied to predict the objects trajectory under occlusions. The experimental result show that the target can be tracked steadily using method presented in this paper. The target acquisition rate is higher than 85% and target tracking accuracy is better than 10 pixels. This algorithm can effectively track the moving targets in complex background.
郭利生. 一种基于中心距离加权的快速相关跟踪算法[J]. 光学与光电技术, 2017, 15(2): 94. GUO Li-sheng. A Rapid Correlation Tracking Algorithm Based on Weight Distance[J]. OPTICS & OPTOELECTRONIC TECHNOLOGY, 2017, 15(2): 94.