正态直方图匹配与非锐化掩蔽的红外图像增强  

Normal Histogram Matching and Non-sharpening Masking for Infrared Image Enhancement

作　　者：卢鹏 李锦丽 LU Peng;LI Jinli(School of Modern Information Industry,Guangzhou College of Commerce,Guangzhou 511363,China;School of Computer,Saint Paul University,Tuguegarao 3500,Philippines)

机构地区：[1]广州商学院现代信息产业学院,广东广州511363 [2]圣保罗大学计算机学院,卡加延土格加劳3500

出　　处：《无线电通信技术》2025年第2期400-406,共7页Radio Communications Technology

基　　金：国家自然科学基金(61907009);广东省普通高校特色创新类项目(2019KTSCX255)。

摘　　要：为了进一步提升红外图像的亮度、对比度和清晰度,提出了正态直方图匹配与非锐化掩蔽的红外图像增强方法。该方法借鉴理想效果图像的直方图分布,构造服从正态分布的直方图作为参照,对红外图像进行直方图匹配,以提升红外图像的亮度、对比度和清晰度;对于正态直方图匹配处理后的图像,利用具有各向同性的高斯滤波提取其差值图像,再将其与差值图像融合,实现图像的非锐化掩蔽处理,以丰富红外图像的纹理和细节信息。实验结果显示,所提方法对红外图像增强后的视觉效果优于参与实验比较的方法,对应增强图像的信息熵和平均梯度相对参与实验比较的方法分别提升0.5和7以上,而自然图像质量评价指标(Natural Image Quality Evaluator,NIQE)降低0.4以上,证明所提方法具有更好的红外图像增强性能。In order further to optimize the brightness,contrast,and clarity of infrared image,an infrared enhancement method based on normal histogram matching and non-sharpening masking is proposed.Enlightened by histogram distribution of ideal image,this method performs histogram matching on infrared image based on constructed histogram which obeys normal distribution,with the aim to optimize the brightness,contrast,and clarity of infrared image;for the image after normal histogram matching,isotropic Gaussian filter is used to extract its difference image,and is fused with the difference image,which achieves non-sharpening masking processing of image,so as to improve the texture and detail effect of infrared image.Experimental results shown that the proposed method outperforms the compared methods in the experiment in enhancing visual effect of infrared images.The information entropy and average gradient of images enhanced by the proposed method are increased by more than 0.5 and 7 respectively compared to the compared methods in the experiment,while the Natural Image Quality Evaluator(NIQE)is reduced by more than 0.4.Therefore,the proposed method behaves better in infrared image enhancement.

关 键 词：红外图像增强 直方图匹配 正态分布 非锐化掩蔽 平均梯度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]