一种改进的CamShift运动目标跟踪算法  

An Improved Cam Shift Moving Target Tracking Algorithm

作　　者：薛桐[1] 任世卿[1] 

机构地区：[1]沈阳理工大学信息科学与工程学院,辽宁沈阳110159

出　　处：《沈阳理工大学学报》2015年第2期28-34,共7页Journal of Shenyang Ligong University

摘　　要：Cam Shift算法具有复杂度低,实时性和鲁棒性好等优点,被广泛地应用于目标跟踪领域。但是Cam Shift算法在运动目标接近与其颜色相近的固定遮挡物时,跟踪会出现明显的偏移。针对这一问题,提出了一种解决与运动目标颜色相近的固定遮挡物对目标跟踪的干扰的方法。首先,运用单高斯背景模型对运动目标进行背景建模;其次,通过背景模型与当前帧的运算,确定与运动目标颜色相近的固定遮挡物所在的位置;最后,利用H分量灰度图计算运动目标所在位置。实验结果显示,该方法在运动目标接近相似颜色固定遮挡物时,能很好地跟踪运动目标。CamShift algorithm has low computing complexity, good real - time performance and robustness, so it is widely applied in target tracking. But the tracking appears shift when moving target is closing the fixed occlusion which has similar color with it. In order to solve this problem, a solution is proposed to solve this kind of occlusion. Firstly, single Gaussian algorithm is adopted to build the background model;Then, the position of similar color fixed occlusion is located by the background model calculation and the current frame;Finally, H component grayscale is applied to calculate the moving object position. The experimental re- suits show that this solution can track the target better while the fixed occlusion has similar color with the closing target.

关 键 词：目标跟踪 CAM SHIFT算法 单高斯背景模型 HSV颜色空间模型 

分 类 号：TN911[电子电信—通信与信息系统]