基于尺度自适应的背景感知目标跟踪算法  

A Background-Aware Target Tracking Algorithm Based on Scale Adaptation

作　　者：刘珊[1,2] 杨卫 邵星灵[1,2] LIU Shan;YANG Wei;SHAO Xingling(North University of China,Key Laboratory of Electronic Testing Technology,Taiyuan 030051 China;North University of China,Key Laboratory of Instrument Science and Dynamic Testing Ministry of Education,Taiyuan 030051 China)

机构地区：[1]中北大学电子测试技术重点实验室,太原030051 [2]中北大学仪器科学与动态测试教育部重点实验室,太原030051

出　　处：《电光与控制》2020年第9期19-23,共5页Electronics Optics & Control

基　　金：国家自然科学基金(61803348)。

摘　　要：针对目标跟踪过程中出现的尺度变化、光照变化、运动模糊及背景杂乱等现象,提出一种基于背景感知相关滤波跟踪器(BACF)的多特征融合与尺度自适应相结合的目标跟踪算法。提取目标和背景信息的多种特征信息,根据响应图对特征分配权重进行加权融合完成目标位置预测;利用方向梯度直方图(HOG)特征和颜色名称(CN)两种互补特征替换判别尺度空间跟踪算法中单一的HOG特征,对目标进行尺度估计;选用自适应变化的阈值更新滤波器模型。仿真对比结果表明所提方法具有较高成功率和精确度。Aiming at the scale changes,illumination changes,motion blur and background clutter in the target tracking process,a target tracking algorithm combining multi-feature fusion with scale adaption and based on Background-Aware Correlation Filters(BACF)is proposed.Firstly,various feature information of the target and the background are extracted,and weighting fusion is made to the feature distribution weights according to the response graph,to complete the target position prediction.Then,a single HOG feature in the discriminant scale space tracking algorithm is replaced by two complementary features of the Histogram of Oriented Gradient(HOG)feature and the Color Names(CN)to estimate the target scale.Finally,the adaptively updated threshold is selected to update filter model.The simulation results show that the proposed method has higher success rate and accuracy.

关 键 词：机器视觉 目标跟踪 相关滤波 特征融合 尺度估计 背景感知相关滤波跟踪器(BACF) 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]