融合SURF特征的压缩追踪算法  

Compressive Tracking Algorithm Based on SURF Feature Fusion

作　　者：方露[1] 韩超[1] 

机构地区：[1]安徽工程大学电气工程学院

出　　处：《四川理工学院学报（自然科学版）》2016年第4期39-43,共5页Journal of Sichuan University of Science & Engineering(Natural Science Edition)

基　　金：安徽省自然科学基金资助项目(1508085MF121);安徽高校自然科学研究项目(KJ2016A056);安徽检测技术与节能装置省级实验室开放研究基金(1506c085002);高校优秀中青年骨干人才国内外访学研修重点项目(gxfxZD2016100)

摘　　要：针对压缩追踪(Compressive Tracking,CT)算法在目标追踪中易受遮挡和扭曲变形干扰问题,结合该算法简单容易执行的追踪机制,提出一种融合SURF(Speeded-up robust features)和压缩特征的鲁棒性目标追踪算法。新算法有两方面的改进:一是在自适应更新目标外观模型的基础上,增加防止误更新外观模型机制,解决追踪过程中严重遮挡和扭曲变形问题;二是通过SURF特征点在前后两帧中的匹配关系,求解追踪目标尺寸变化,自适应调整目标模板大小。通过仿真实验表明:改进后的算法在公开的某些图像序列上的追踪效果良好,与CT算法及改进的CT算法相比正确性和鲁棒性上性能更优越。Aim at the compressive tracking（ CT） algorithm which easily to be occluded and distorted in target tracking,combining the simple tracking mechanism and easy implement,a robust target tracking algorithm which includes SURF（ Speeded-up robust features） and compression features is proposed. The new algorithm has two aspects of improvement as following： Firstly,based on the adaptive update of the target appearance model,the mechanism of the appearance model is added to prevent the false appearance,which solves the problem of serious occlusion and distortion in the tracking process.Secondly,through the matching relationship of the SURF feature points between adjacent frames,the solution of the target size change is completed and the adaptive adjustment of tracking template size is achieved. The simulation results show that the improved algorithm has a good tracking effect on some image sequences,and it is superior in accuracy and robustness in comparison with the CT algorithm and the improved CT algorithm.

关 键 词：压缩追踪算法 SURF算法 误更新机制 追踪模板 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]