基于STRCF的改进HOG特征目标跟踪算法研究

Research on Target Tracking Algorithm Based on Improved HOG Feature of STRCF

视频序列中运动目标跟踪时常因旋转、形变、被遮挡等原因造成跟踪失败,无法达到较好预期。针对此情况提出一种基于STRCF的改进HOG特征目标跟踪算法。首先,通过搜索目标区域的Sobel特征局部最值,然后以局部最值所在区域为中心施加一个高斯权重,最后将权重与HOG特征相融合,并将其应用于目标跟踪。该方法的优点是扩大了目标与背景的差异化,使得所提取特征能更有效地描述目标物体。同时,为结合时间轴信息构建稳定模板,跟踪过程中模板更新采用多帧更新方式。实验结果表明,该方法可有效地解决跟踪过程中目标旋转以及被遮挡等问题,能实现目标的鲁棒性跟踪,具有一定应用价值。

When tracking moving objects in video sequences,it is often caused by rotation,deformation,occlusion and other reasons that the tracking fails to achieve better expectations.In view of this situation,a target tracking algorithm based on the improved HOG feature of STRCF is proposed.Firstly,the local maximum value of Sobel feature in the target region is searched,then a Gaussian weight is applied to the region where the local maximum value is located.Finally,the weight is fused with the HOG feature and applied to target tracking.The advantage of this method is to expand the difference between the target and the background,so that the extracted features can describe the target object more effectively.At the same time,in order to build a stable template combining with the time axis information,the template updating in the tracking process adopts the method of multi frame updating.The experimental results show that this method can effectively solve the problems of target rotation and occlusion in the tracking process,and it can achieve robust target tracking,which has a certain application value.