基于Camshift和SIFT线性融合的目标跟踪算法研究  

Object Tracking Algorithm Based on Linear Fusion of Camshift and SIFT

作　　者：叶家林[1] 宋建新[1] 

机构地区：[1]南京邮电大学江苏省图像处理与图像通信重点实验室,江苏南京210003

出　　处：《电视技术》2015年第17期135-138,共4页Video Engineering

摘　　要：Camshift算法是对Mean Shift算法的改进,它可以解决目标尺度缩放、持续跟踪等问题。但是当目标颜色与背景颜色接近或者目标遇到旋转问题时,Camshift算法容易失效。而SIFT算子对旋转、亮度变化保持不变性,对颜色相近也保持一定的稳定性。基于此,提出一种Camshift与SIFT算子线性融合的目标跟踪算法。首先利用Camshift算法来对目标进行初步的跟踪,得到跟踪区域,再利用SIFT特征向量来匹配目标区域与跟踪区域,得到SIFT的匹配和校正结果,再将两种算法的结果进行线性融合,得到最终的跟踪结果。实验结果表明,提出的方法可以有效地解决跟踪过程中出现的旋转、颜色相近等问题。Camshift algorithm is an improvement of the MeanShift algorithm, which can solve the scale change and continuous tracking. But when the color of the targets are similar to the color of background, or the targets rotate, Camshift algorithm may fail. Meanwhile, SIFT operator not only keeps invariant to rotation and illumination, but also keeps the stability of similar color. So a target tracking algorithm that fuses tile Camshift anti SIFT operator is proposed. Firstly, Camshift algorithm is used to track the moving target preliminary and get the tracking area. Secondly, the SIFT feature vector is used to match the target area and tracking area, so the SIFT matching and correction results are obtained. Lastly, the target tracking results are linear fusion of the two algo- rithm, which get the ultimate resnhs. Experimental results show that the proposed method can effectively solve the occlusion and the rotation in the tracking.

关 键 词：MEANSHIFT CAMSHIFT SIFT 跟踪 线性融合 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]