复杂场景下灰度图像的运动目标跟踪

提出了一种对复杂场景下灰度图像序列中运动目标分割和跟踪的新方法:首先,利用光流法分割出运动目标;然后,以mean shift算法为核心跟踪感兴趣的目标.跟踪过程中以目标灰度直方图为特征进行帧与帧之间的目标匹配,其匹配的相似度以Bhattacharyya系数来测量.算法中利用Kalman滤波器对运动目标在图像中的位置进行预测,不仅可以有效解决目标的暂时遮挡问题,而且可以缩小模式匹配的搜索范围,提高处理速度.实验结果和对实验相关数据的分析验证了该跟踪算法的有效性和实时性.

Moving Targets Tracking Under Complex Scene in Gray Image Sequences

An efficient approach for moving target segmentation and tracking under complex scene in gray image sequences is presented.Firstly,moving targets are segmented with optical flow and the one which we want can be selected in a cluttered environment.Then,mean shift analysis is employed to track the target.During tracking,gray histogram is used to match the target frame by frame,the similarity is measured with Bhattacharyya coefficient.To solve the temporal occlusion problem,a Kalman filter is adopted to estimate the next location of the target.Using such a strategy,the searching scope of model match in each frame is reduced,and the tracking algorithm is more efficient.The experiment results show that our method is effective and real-time.