尺度补偿的相关核滤波器跟踪

相关核滤波器跟踪是工程实际中非常实用的跟踪算法,它的算法简单,只需要在下一帧进行一个样本的密集采样就能对目标进行跟踪,但是对于有尺度变化的目标跟踪的适用性不足。采用了尺度补偿的相关核滤波器跟踪算法,对相关核滤波器跟踪进行了改进。首先使用了点跟踪补偿机制对相关核滤波器尺度和位移进行补偿；其次采用了压缩感知提取的特征建立模板对目标进行建模,在关键帧进行目标的重检测来防止尺度估计带来的跟踪误差。通过实验对提出的算法进行了标准视频库的测试,并在中心点误差和实际跟踪覆盖率两个指标上与原算法进行了对比分析。实验结果表明,提出的具有尺度补偿的跟踪算法提高了相关核滤波器跟踪在有尺度属性变化视频序列中的准确率和实用性。

Kernelized Correlation Filters Tracking with Scale Compensation

Kernelized correlation filters is a very useful tracking algorithm in practice,whose algorithm is simple and requires only one dense sampling in the next frame for tracking,however the applicability is insufficient when tracking object with scale changes.In this paper,we used kernelized correlation filter tracking algorithm with scale compensation to improve the original algorithm.Firstly,we used point tracking compensation mechanism for this algorithm to compensate for scale changes and shift change.Secondly,we modeled the target with features extracted from compressed sen-sing and detected the tracking object in key frame for the sake of error in the scale estimation.The algorithm is tested in standard tracking library and compared with original tracking tracking algorthm in center error and the actual overlapped rate.The results show that the tracking algorithm with scale compensation in this paper improves accuracy and practicality when used in vedio with scale attributes.