摘要
实用的跟踪系统应该具备尺度自适应机制。针对传统跟踪方法难以实时准确适应目标尺度变化这一问题,提出一种基于对数极坐标变换域下目标匹配的尺度变化目标跟踪算法,算法结合粒子滤波和改进的Mean Shift进行空间定位,确定目标的形心,通过对数极坐标变换域中目标和候选的最大相关匹配系数来确定目标的尺度参数。实验结果表明:与传统方法相比较,该算法可以自适应选择合适的跟踪窗大小,具有较好的鲁棒性。
A practical tracking system should be equipped with scale adaptation mechanism. To solve the problem of the traditional methods which could not keep up with the scale changing timely and accurately, the particle filter and an improved Mean Shift method were introduced to locate the centroid of the target. Then, the correlation coefficient was adopted to map the object and its candidate in the Log-Polar Transformation domain to fit the target scale. Experimental results demonstrated that the composite algorithm can select a proper size of tracking window adaptively. Compared with the traditional method, it has a better robustness.
出处
《电光与控制》
北大核心
2013年第12期42-45,87,共5页
Electronics Optics & Control
基金
国家自然科学基金项目(61175029)
