灰色预测和混沌PSO的红外小目标检测  

The detection of a small infrared target based on gray prediction and chaotic PSO

作　　者：尹丹艳[1] 吴一全[1,2] 

机构地区：[1]南京航空航天大学信息科学与技术学院,江苏南京210016 [2]光电控制技术重点实验室,河南洛阳471000

出　　处：《智能系统学报》2011年第2期126-131,共6页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金资助项目(60872065);航空科学基金资助项目(20105152026);南京大学计算机软件新技术国家重点实验室开放基金资助项目(KFKT2010B17)

摘　　要：在分析红外图像弱小目标和背景特征的基础上,提出了一种基于灰色预测和混沌PSO的红外小目标检测方法.该方法首先采用灰色系统理论中的GM(1.1)模型对红外图像中的背景进行时域预测,并用实际图像减去预测图像得到残差图像,在抑制背景的同时增强了目标;然后提出了混沌粒子群优化(particle swarm optimization,PSO)的模糊最大熵二维直方图斜分方法,采用此方法对所得残差图像进行分割即可将小目标检测出来.实验结果表明:该方法可显著提高红外目标的检测概率,实现较远距离小目标的检测.By analyzing the characteristics of a small target and the background of an infrared image,a detection method of an infrared small target based on gray prediction and chaotic particle swarm optimization（PSO） was proposed.First,the GM（1,1） model of gray system theory was adopted to predict the infrared image background in a time domain.The subtraction of the source image minus the predicted image gave the residual image.As a result,the background was suppressed and the target was enhanced.Then,a two-dimensional histogram oblique segmentation method based on chaotic PSO and fuzzy maximum entropy was presented.The residual image was segmented by this method,leading to the detection of the small target.The experimental results show that the proposed method can significantly increase the detection probability of an infrared target to achieve long-range small target detection.

关 键 词：红外小目标检测 背景抑制 灰色预测 模糊最大熵 混沌粒子群 

分 类 号：TP274.52[自动化与计算机技术—检测技术与自动化装置]