一种改进的长时间压缩感知跟踪算法

压缩感知跟踪(CT)算法具有简单、高效、实时的优点,但是却存在着跟踪窗口尺寸不能自适应变化,无法有效处理遮挡以及跟踪失败后的目标再发现等问题.为了解决上述问题,提出了一种改进的长时间压缩感知跟踪算法.所提出的算法采用多尺度的目标外观再匹配方法,使得跟踪窗口大小能够适应目标尺寸变化.此外,通过分析滑动窗口内跟踪窗口图像的整体特征变化来判定目标是否发生遮挡.为了解决跟踪器漂移问题,采用Haar特征在线生成检测器,实现目标的再发现.实验结果表明提出的算法相比原CT算法具有更好的鲁棒性和准确性.

An Improved Long-time Compressive Tracker

Compressive tracker is famous for its simplification and efficiency. However, there still exists a few problems to be solved. For example, only fixed-size tracking windows can be generated, besides, the problem of occlusion as well as re-identification of the target after a track failure is not well handled. To address the issues mentioned above, an improved long-time compressive tracker is proposed. Through multi-scale appearance matching of the target, tracking windows adaptive to the object size are generated. Besides, the change of the global feature of consecutive frames in a sliding window is analyzed to see if an occlusion occurs. Finally, to solve the problem of drift, an online detector using Haar features is learned to re-identify the lost object. The experimental results demonstrate that our algorithm performs better than the CT algorithm in robustness and precision.