###
上海电力大学学报:2019,35(5):442-448
本文二维码信息
码上扫一扫!
基于改进核相关滤波算法的目标跟踪
(上海电力学院 自动化工程学院)
Target Tracking Based on Improved Kernel Correlation Filtering
(School of Automation Engineering, Shanghai University of Electric Power, Shanghai 200090, China)
摘要
图/表
参考文献
本刊相似文献
All Journals 相似文献
All Journals 引证文献
本文已被:浏览 1497次   下载 537
投稿时间:2019-04-02    
中文摘要: 针对核相关滤波算法仅使用一种特征表达进行目标追踪,使其在一些场景中跟踪效果不佳的问题,提出了一种多特征融合的核相关滤波跟踪方法。采用31维的方向梯度直方图特征、58维的局部二值模式特征和1维的灰度特征进行融合。该算法选择在特征层进行特征融合,先将方向梯度特征和局部二值模式特征并联融合,再将融合后的特征串联融合灰度特征,形成新的特征表达。在OTB(Object Tracking Benchmark)数据集上进行了测试,结果表明,该算法具有更好的跟踪效果。
中文关键词: 核相关滤波  特征融合  目标跟踪
Abstract:A kernel correlation filtering method based on multi-feature fusion is proposed for the kernel correlation filtering algorithm using only one feature expression for target tracking,which makes the tracking effect of the algorithm in some scenes poor.The 31-dimensional histogram of oriented gradient feature,the 58-dimensional local binary patterns feature and the 1-dimensional gray feature are used for fusion.The algorithm selects the feature fusion in the feature layer,combining the histogram of oriented gradient feature and the local binary mode feature.Then,the fused features are fusing into the gray feature to form a new feature expression.The improved algorithm is tested on the Object Tracking Benchmark(OTB) dataset.Experiments show that the algorithm has better effect on tracking.
文章编号:20195007     中图分类号:TP391.4    文献标志码:
基金项目:
引用文本:
孙紫君,黄福珍.基于改进核相关滤波算法的目标跟踪[J].上海电力大学学报,2019,35(5):442-448.
SUN Zijun,HUANG Fuzhen.Target Tracking Based on Improved Kernel Correlation Filtering[J].Journal of Shanghai University of Electric Power,2019,35(5):442-448.