SIA OpenIR  > 光电信息技术研究室
基于前景感知的时空相关滤波跟踪算法
Alternative TitleForeground-aware based Spatio-temporal Correlation Filters for Visual Tracking
虞跃洋1,2,3,4,5; 史泽林1,2,3,4,5; 刘云鹏2,3,4,5
Department光电信息技术研究室
Source Publication激光与光电子学进展
ISSN1006-4125
2019
Volume56Issue:22Pages:1-12
Indexed ByCSCD
CSCD IDCSCD:6622175
Contribution Rank1
Funding Organization国家自然科学基金 ; 教育部新世纪优秀人才支持计划 ; 浙江大学引进人才科研启动项目
Keyword机器视觉 目标跟踪 相关滤波 时间一致性 重检测
Abstract

针对长时目标跟踪中目标背景混杂、遮挡、目标移出视野导致的跟踪失败问题,本文基于空间正则化相关滤波(SRDCF)跟踪算法,提出了一个基于前景感知的时空相关滤波算法。首先,提出了前景感知相关滤波方法,使得滤波器能够准确地把目标前景区域和背景区域有效区分;然后,把前景感知滤波器加入时间正则项中,使具有时空正则化功能的滤波器始终保持在一个低维的判别流形上;同时,采用交替方向乘子法(ADMM)求解,使得跟踪方法在传统特征的表达上能实现实时性;最后,确定目标重检测器的激活阈值,利用候选区域方法结合相关滤波方法实现重检测,达到长时跟踪的目的。在标准数据集OTB2013上分别利用传统特征和卷积特征进行实验,与SRDCF算法相比,跟踪平均成功率分别提高了5.6%和7%。该算法针对目标背景模糊,旋转,遮挡和移出视野等情况,具有较强的鲁棒性。

Other Abstract

Aiming at the problems of object tracking failure caused by background clutter, occlusion and out-of-view in long-term tracking, we propose a spatio-temporal regularized correlation filter with foreground aware algorithm for visual tracking based on the spatially regularized discriminative correlation filter (SRDCF). First, this work propose a foreground-aware function which can model how both the foreground and background of the object varies over time. Then, we regard it as a time consistency constraint, with which the new tracker enables joint spatio-temporal filter learning in a lower dimensional discriminative manifold. At the same time, it can be efficiently solved via the alternating direction method of multipliers (ADMM) to ensure real-time. Finally, the object re-detection threshold is determined. When the occlusion or out-of-view causes the failure, the region proposal can be extracted and the object position is re-detected by correlation filter. The experiments of hand-crafted features and deep features were completed on OTB2013 respectively. Compared with SRDCF, the average success rate is increased by 5.6% and 7% respectively. In addition, the proposed approach is robust for background blur, rotation, occlusion and out of view.

Language中文
Citation statistics
Document Type期刊论文
Identifierhttp://ir.sia.cn/handle/173321/24704
Collection光电信息技术研究室
Corresponding Author虞跃洋
Affiliation1.中国科学技术大学信息科学技术学院
2.中国科学院沈阳自动化研究所
3.中国科学院机器人与智能制造创新研究院
4.中国科学院光电信息处理重点实验室
5.辽宁省图像理解与视觉计算重点实验室
Recommended Citation
GB/T 7714
虞跃洋,史泽林,刘云鹏. 基于前景感知的时空相关滤波跟踪算法[J]. 激光与光电子学进展,2019,56(22):1-12.
APA 虞跃洋,史泽林,&刘云鹏.(2019).基于前景感知的时空相关滤波跟踪算法.激光与光电子学进展,56(22),1-12.
MLA 虞跃洋,et al."基于前景感知的时空相关滤波跟踪算法".激光与光电子学进展 56.22(2019):1-12.
Files in This Item: Download All
File Name/Size DocType Version Access License
基于前景感知的时空相关滤波跟踪算法.pd(1567KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[虞跃洋]'s Articles
[史泽林]'s Articles
[刘云鹏]'s Articles
Baidu academic
Similar articles in Baidu academic
[虞跃洋]'s Articles
[史泽林]'s Articles
[刘云鹏]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[虞跃洋]'s Articles
[史泽林]'s Articles
[刘云鹏]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 基于前景感知的时空相关滤波跟踪算法.pdf
Format: Adobe PDF
This file does not support browsing at this time
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.