基于模糊融合的红外弱小目标快速检测算法  

Fast detection algorithm for moving dim target in infrared image based on fuzzy fusion

作　　者：武斌[1] 姬红兵[1] 

机构地区：[1]西安电子科技大学电子工程学院,陕西西安710071

出　　处：《光电工程》2007年第12期6-11,共6页Opto-Electronic Engineering

基　　金：国家自然科学基金资助项目(60677040);国家部委预研基金资助(××××2030105DZ0177)

摘　　要：提出了一种基于模糊融合的红外弱小目标快速检测新算法。算法以差分图像为基础,根据差分图像的噪声特性引入衡量像素点灰度变化程度的隶属度函数,将经过隶属度函数"模糊化"的数帧差分图像进行模糊"与"操作实现融合,按照规则对融合后的图像进行两次模糊推理,实现了弱小目标的检测。仿真实验结果表明,该算法保留了差分法良好的实时性,克服了确定检测阈值难和"硬"判决带来检测概率低的缺点,能快速有效地检测出低信噪比红外图像序列中的像素个数不小于4的弱小运动目标。A new fast detection algorithm for moving dim target in infrared image based on fuzzy fusion is proposed in this paper. According to the noise characteristics of the differential image, the algorithm introduces the membership function that evaluates the intensity of gray-grade variation on each pixel point of the differential image. And then several differential images which are processed with the membership function are fused by fuzzy ＂and＂ operation. The dim target is found after the result image is dealt with through two operations of fuzzy reasoning. The experimental results show that our algorithm remains the excellent real time performance of image difference algorithm and avoids the shortcoming of low detection probability caused by difficult threshold determination and ＂hard＂ decision, and effectively detects dim moving target with more than 4 pixels

关 键 词：弱小目标检测 模糊隶属度 模糊融合 红外图像序列 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]