摘要
针对核相关滤波器在复杂光照条件下出现的跟踪不稳定的现象,提出一种基于LBP(local binary pattern)与核相关滤波器的运动目标跟踪算法。在传统算法上增加LBP处理方法,降低光照对特征提取的影响,进而提高核相关滤波器算法在跟踪过程中对目标信息的采集精准度。实验表明,与经典的核相关滤波器跟踪算法相比,基于LBP与核相关滤波器的运动目标跟踪算法在复杂光照的情况下的跟踪性能有明显提升,能较好应用于实时场景中去,是一种稳定的目标跟踪算法。
A moving target tracking algorithm based on LBP(Local binary pattern) and kernel correlation filter is proposed for kernel correlation filter when the tracking instable under complex illumination. In the traditional algorithm the LBP processing method is added to reduce the effect of light on the feature extract, and then the accuracy of the acquisition of the target information in the tracking process is improved. Experiments show that compared with the classical kernel correlation filter tracking algorithm, the moving target tracking algorithm based on LBP and kernel correlation filter can significantly improve the tracking performance when facing complex illumination. It can be used in real time scene and is a stable target tracking algorithm.
作者
齐永锋
王梦媛
QI Yongfeng;WANG Mengyuan(College of Computer Science and Engineering, Northwest Normal University, Lanzhou 730070, China)
出处
《红外技术》
CSCD
北大核心
2019年第6期572-576,共5页
Infrared Technology
基金
甘肃省高等学校科研项目(2016A-004)
关键词
核相关滤波器
LBP
特征提取
实时处理
机器学习
kernel correlation filter
LBP
feature extract
real-time processing
machine learning