基于同态滤波及多尺度Retinex的低照度图像增强算法  

Low-intensity Image Enhancement Algorithm Based on Homomorphic Filtering and Multi-scale Retinex

作　　者：董静薇[1] 徐博 马晓峰 韩闯 DONG Jing-wei;XU Bo;MA Xiao-feng;HAN Chuang(School of Measurement&Control Technology and Communications Engineering,The Higher Educational Key Laboratory for Measuring&Control Technology and Instrumentations of Heilongjiang Province, Harbin Uinversity of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学测控技术与通信工程学院,测控技术与仪器黑龙江省高校重点实验室,哈尔滨150080

出　　处：《科学技术与工程》2018年第22期238-242,共5页Science Technology and Engineering

基　　金：国家自然科学基金(61601149)、黑龙江省留学归国基金(LC201427)和黑龙江省科学基金(QC2017074)资助.

摘　　要：针对在彩色图像采集过程中,光源偏暗或曝光不足等因素,常导致图像亮度和对比度偏低问题,提出了一种改进的低照度图像增强算法。首先用改进的同态滤波增强低照度图像的RGB各分量;然后将RGB图像转换到HSV彩色空间,对饱和度分量进行自适应非线性拉伸;同时用改进的多尺度Retinex算法对亮度进行增强处理,对照射分量用伽马变换进行校正,对反射分量用Sigmoid函数进行处理,最后将图像再转换至RGB空间。用MATLAB对图像进行仿真处理。实验表明该算法提高了低照度图像的信息熵、峰值信噪比和对比度,提升了低照度图像的视觉效果。Aiming at the problems of low brightness and low contrast often caused by the dark or underexposed light source in the process of color image acquisition,an improved low illumination image enhancement algorithm is proposed.First,the improved homomorphic filter was used to enhance the RGB components of the low-illumination image.Then the RGB image was converted into the HSV color space to adaptively stretch the saturation component nonlinearly.At the same time,an improved multi-scale Retinex algorithm enhances the brightness,corrects the illumination component with gamma transform,processes the reflection component with Sigmoid function,and finally converts the image to RGB space again.Experimental results show that the proposed algorithm improves the entropy,peak signal-to-noise ratio and contrast of low-illumination images and enhances the visual effects of low-illumination images.

关 键 词：图像增强 HSV空间 同态滤波 多尺度RETINEX 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]