|国家预印本平台
首页|基于视觉预测的核相关滤波跟踪器设计

基于视觉预测的核相关滤波跟踪器设计

he design of nuclear related filter tracker based on visual prediction

中文摘要英文摘要

针对核相关滤波(KCF)跟踪器不能很好的追踪低帧率视频中高速运动目标的问题,在原算法的基础上,提出了一种据有目标位置预测机制的跟踪算法。在获得新的视频帧时,根据目标先前的位移数据,使用最小二乘法预测出目标在当前帧的位移,相应调整采样窗口位置,而非原算法中的在上一帧中的目标位置采样。与原方法相比,改进后的算法可以更好的追踪快速运动目标。

ccording to the problem that KCF tracker can\'t track of high-speed motion target well in low frame rate of video, a tracking algorithm which has target location prediction mechanism is proposed based on the original algorithm. When we get a new video frame, we use the least square method to predict the displacement of the target in the current frame according to the target displacement of previous data.Then we adjust sampling window position instead of using the sampling target location in the last frame of the original algorithm. Compared with the original method, the improved algorithm can track the fast moving target better.

李霖、黄扬帆、余然、石珂

电子技术应用

视觉目标跟踪视觉预测相关滤波器

visual object trackingvisual predictioncorrelation filter

李霖,黄扬帆,余然,石珂.基于视觉预测的核相关滤波跟踪器设计[EB/OL].(2018-06-08)[2025-08-23].http://www.paper.edu.cn/releasepaper/content/201806-46.点此复制

评论