融合颜色特征的尺度自适应相关跟踪

对于彩色视频的目标跟踪，传统基于相关滤波的跟踪方法没有利用视频帧的颜色信息，针对该问题本文在相关跟踪框架下提出了一种融合颜色特征的尺度自适应目标跟踪方法。该方法首先将HSV三个通道归一化后作为目标的颜色特征，并与灰度特征以及HOG特征进行融合，然后利用核相关滤波定位目标的中心位置，最后再利用尺度金字塔模型对目标尺度进行估计。在公开的彩色视频数据集上对所提方法进行测试，并和多种跟踪方法进行对比。实验结果表明，该方法的平均跟踪性能优于其它方法，验证了方法的有效性。

Scale adaptive correlation tracking combined with color features

Traditional correlation tracking methods failed in digging the color information of the image sequence in color video object tracking. To solve this problem, this paper proposed a robust scale adaptive tracker combined with color features in correlation filter framework. Firstly, the normalized HSV was extracted as color feature to integrate gray features including gray value and HOG. Secondly, the multi-channel kernelized correlation filter was employed to locate the center of the target. Finally, the target scale was estimated via scale pyramid. Experiments were performed on several benchmark challenging color sequences. The results show that our approach outperforms state-of-the-art tracking methods while operating in real-time.