一种基于角点的Mean Shift目标跟踪算法*

为了提高经典的Mean Shift算法在复杂场景中的跟踪性能，提出了一种基于角点的目标表示方法。首先，利用Harris角点检测算法提取表示目标主要特征的角点；其次，基于提取的角点，建立目标模型，将其嵌入Mean Shift算法进行跟踪。该方法仅用少量的关键点表示目标，能够自动去除目标和背景中的次要特征，有效地抑制背景成分对目标定位的影响，从而改进Mean Shift目标跟踪算法的性能。通过测试两个复杂环境下的视频，实验结果表明，相对于传统的目标跟踪算法，提出的方法取得了更好的性能。

Mean Shift tracking algorithm based on corner points

To improve the performance of the classical Mean Shift tracking algorithm in the complex scene, developed a target representation method based on corner points. First, detected the corner points which contain the main features of the object by using Harris corner detector. Then used the obtained corner points to calculate the target model, which was furthermore embedded into the Mean Shift tracking algorithm. Because the proposed method tracked the target only with few key points, it could remove automatically the minor features of the object and background and suppress the interference of backgrounds for target localization. Experimental results on two challenged videos in cluttering environments show that Mean Shift with the proposed method gets better results than the standard method.