一种新的基于特征直方图分解的红外焦平面阵列无效像元判别方法  

NEW APPROACH OF IRFPA NON-EFFECTIVE PIXEL DISCRIMINATION BASED ON PIXEL’S CHARACTERISTICS HISTOGRAM ANALYSIS

作　　者：石岩[1] 毛海岑[2] 张天序[1] 曹治国[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所,湖北武汉430074 [2]华中科技大学图像信息处理与智能控制国家教委开放实验室,湖北武汉430074

出　　处：《红外与毫米波学报》2005年第2期119-124,共6页Journal of Infrared and Millimeter Waves

基　　金：国家自然科学基金重点资助项目(60135020)

摘　　要：为有效地判别和剔除红外焦平面阵列(IRFPA)的无效像元,本文在分析了IRFPA有效像元响应模型的基础上,指出在均匀输入时IRFPA有效像元输出值的均值差(DOM)和标准差 (STD)特征的分布具有正态性,通过采用自适应投影匹配分解法对特征直方图进行高斯分解,进而估计出有效像元DOM和STD特征的分布区间,从而得到了判别无效像元的分类准则.针对实际IRFPA图像数据的实验表明,本文提出的无效像元的判别方法是有效的.A new approach of IRFPA non-effective pixel discrimination based on pixel's two kinds of characteristic histogram decomposition was presented. Based on the analysis of the model for the effective pixel response, the approach proposes that, if the incident flux is uniform, the standard deviation( STD) and the difference of two means(DOM) for all effective pixels are both normal distributed random variables. By decomposing STD and DOM histograms through adaptive projective decomposition, the DOM and STD distribution range of effective pixel then can be estimated, and the discrimination rule for non-effective pixel is got ultimately. The experiment result using real infrared image shows the proposed technique for IRFPA non-effective pixel discrimination is effective and feasible.

关 键 词：红外焦平面阵列 判别方法 像元 基于特征 图分解 IRFPA 响应模型 高斯分解 图像数据 分类准则 正态性 标准差 直方图 分解法 自适应 STD DOM 分布 匹配 投影 

分 类 号：TP391.72[自动化与计算机技术—计算机应用技术] TN215[自动化与计算机技术—计算机科学与技术]