基于人类视觉机制和粒子滤波的红外小目标跟踪  

Infrared target tracking based on human vision mechanism and particle filter

作　　者：王瑞 朱志宇[1] 张冰[1] WANG Rui;ZHU Zhiyu;ZHANG Bing(School of Electronics and Information,Jiangsu University of Science and Technology,Zhenjiang 212003,China)

机构地区：[1]江苏科技大学电子信息学院,镇江212003

出　　处：《江苏科技大学学报（自然科学版）》2018年第4期528-532,共5页Journal of Jiangsu University of Science and Technology:Natural Science Edition

基　　金：国家自然科学基金资助项目(61671222);江苏省自然科学基金资助项目(SBK2015021788)

摘　　要：针对红外小目标图像信噪比低,背景复杂等特点,提出一种基于人类视觉系统对比机制和粒子滤波的红外小目标跟踪方法.该方法充分模拟人眼视觉系统对比机制,计算图像局部视觉对比度获得显著图,以局部视觉对比度显著图为粒子滤波跟踪的目标特征,锁定初始帧感兴趣目标区域,提取"九宫格"式目标区域建立跟踪模版,实现对红外图像序列小目标的精确跟踪,并与基于灰度特征的传统粒子滤波跟踪方法进行了对比.实验结果显示,提出的方法能够在低信噪比条件下更稳健地跟踪红外弱小目标.Since signal to noise ratio(SNR)of infrared small target image is low,and the background is complex and so on,an infrared small target tracking method based on the contrast mechanism of human visual system and particle filter is proposed.The method fully simulated the contrast mechanism of human visual system.Saliency map is obtained by calculating the local image visual contrast degree,and the local visual contrast is the characteristic of the particle filter tracking.Target region of interest in the initial frame is locked,tracking template is established by target region of the squares type extraction,and accurate tracking of small targets in infrared image sequences is achieved.Compared with the traditional particle filter tracking method based on gray feature,experimental results show that the proposed method can track the infrared dim target more robustly in low SNR condition.

关 键 词：红外弱小目标 粒子滤波 目标跟踪 视觉机制 

分 类 号：TN751[电子电信—电路与系统]