基于SURF和Camshift的运动目标跟踪

摘要

An improved algorithm based on SURF and Camshift was proposed to deal with the problem of tracking failures in complex backgrounds with color noises and occlusion.The position of object was located by matching the SURF feature of the template image and video sequence.The initial search window for Camshi[t tracking algorithm was initialized by extracting the moving object.The moving object of current frame was used to update template image after each tracking.Experimental results verify that the proposed algorithm can track the object accurately and has better robustness to color noises and occlusion.%为解决传统的Camshift算法在颜色干扰、遮挡等复杂场景下跟踪失效的问题,对SURF算法和Camshift算法进行深入研究,尝试将两种算法融合,把SURF算法融合到Camshift算法中.利用SURF算法对目标模板和视频序列进行特征匹配,将匹配成功的运动目标的跟踪窗口作为Camshift算法的初始搜索窗口,每次跟踪完成后,用当前帧跟踪到的运动目标更新目标模板.实验结果表明,加入了SURF特征的Camshift算法能够对颜色干扰和遮挡场景下的运动目标进行稳定跟踪,融合后的算法跟踪效果较好.