改进Retinex在红外图像增强中的应用  

Application of Improved Retinex in Infrared Image Enhancement

作　　者：任荣[1] 张小伍[1] 

机构地区：[1]北方民族大学计算机科学与工程学院,银川750021

出　　处：《激光杂志》2016年第8期93-96,共4页Laser Journal

基　　金：宁夏自治区重点学科项目(NZD1501)

摘　　要：增强是改善红外图像质量的关键环节,针对标准Retinex算法的红外图像"光晕"现象,提出一种改进Retinex的红外图像增强算法。首先对红外图像进行直方图均衡化处理,改善红外图像灰度分布,然后采用小波变换对对红外图像分解,采用Retinex算法低频子图像进行对比度增加,并通过阈值法去除高频子图像噪声。结果表明,相对于其它红外图像增算法,改进Retinex算法避免了"光晕"现象,增强红外图像细节,红外图像视觉效果与人的感官感应更加相符。enhancing is a key point to improve the quality of infrared image, Retinex algorithm for infrared image has defect of ＂halo＂ phenomenon, this paper put forward an infrared image enhancement method based on improved Retinex. At first, the infrared image is histogram equalization, and the infrared image quality is improved, and then wavelet transform is used to decompose infrared image, and Retinex algorithm is used to increase the contrast of low frequency sub image while the noise in the high-frequency sub images is removed by the threshold method. The exper- imental results show that compared with standard Retinex algorithm, the proposed algorithm does not exist ＂halo＂ phe- nomenon, enhance detail of infrared image, the visual effect of infrared image is more consistent with human＇ s sensory induction.

关 键 词：红外图像 直方图均衡化 色彩恒常理论 视觉效果 图像增强 

分 类 号：TN911[电子电信—通信与信息系统]