基于Canny边缘检测的Mean-Shift跟踪核窗宽确定方法

Method of Determinating the Mean-Shift Tracking Kernel-Bandwidth Based on Canny Edge Detection

摘要:

传统核窗宽固定的Mean-Shift跟踪算法不能很好地对做复杂运动、尺寸变化的目标进行有效地跟踪,变核窗宽是解决这一问题的途径。图像的边缘表征了目标的重要特征,可以通过提取目标的边缘来确定核窗宽并用于跟踪。基于Canny边缘检测算法确定跟踪窗宽加快了Mean-Shift跟踪算法的运算速度,仿真结果验证了该跟踪算法的有效性。

Abstract:

Classic Mean-Shift based tracking algorithm uses fixed kernel-bandwidth, which limits the performance when the object does complex movement and scale exceeds the size of the tracking window. Using changeable kernel-bandwidth can solve this question. Image edge represents the important tincture of object, so can determinate the kernel-bandwidth by extracting the object edge and then track. The method of determinating the Mean-Shift tracking kernel-bandwidth based on Canny edge detection expedites the operation of Mean-Shift tracking, the simulation result verifies the affectivity of the proposed method.