多尺度自适应Gamma矫正的低照图像增强  

Low illumination image enhancement based on multi-scale adaptive Gamma correction

作　　者：侯利霞 聂丰英[1] 万里勇[1,2] HOU Li-xia;NIE Feng-ying;WAN Li-yong(School of Information and Artificial Intelligence,Nanchang Institute of Science and Technology,Nanchang 330108,Jiangxi,China;Management Science and Engineering Research Center,Jiangxi Normal University,Nanchang 330022,Jiangxi,China)

机构地区：[1]南昌工学院信息与人工智能学院,江西南昌330108 [2]江西师范大学管理科学与工程研究中心,江西南昌330022

出　　处：《云南大学学报（自然科学版）》2023年第1期57-66,共10页Journal of Yunnan University(Natural Sciences Edition)

基　　金：江西省教育厅科学技术研究项目(GJJ212517);教育部产学合作协同育人项目(202102450009)。

摘　　要：为了在提高低照图像的亮度和对比度的同时,保持图像的色彩自然度,提出了多尺度自适应Gamma矫正的低照图像增强方法.鉴于HSI颜色空间的明度分量Ⅰ决定着图像的明暗程度,以及明度分量的多尺度特性,首先,将图像转换到HSI颜色空间,对明度分量Ⅰ进行多尺度Retinex分解;然后,对分解得到的多尺度光照图像分别进行自适应的Gamma矫正,其中Gamma指数自适应于暗区像素的占比,鲁棒地改善光照图像的光照分布,经Retinex反变换得到增强的多尺度明度分量;最后,将增强的多尺度明度分量的线性融合作为增强的明度分量Ⅰ’,与色相分量H和饱和度分量S重组,得到最后的增强图像.实验结果表明,相对于部分最新提出的现有方法,新方法的PSNR和SSIM值分别高出1.19 dB和1.8%以上,具有更好的增强效果和鲁棒性,增强图像的视觉效果更适宜,对比度更高.In order to improve the brightness and contrast of the low illumination image while maintaining the color naturalness of the image,a low illumination image enhancement method based on multi-scale adaptive Gamma correction is proposed.In view of that the degree of light and shade of image is determined by the lightness componentⅠof HSI color space,and the multi-scale characteristic of lightness component.Firstly,the image is converted to HSI color space,and the brightness componentⅠis decomposed by multi-scale Retinex.Then,respectively,the decomposed multi-scale illumination images are subjected to adaptive Gamma correction,in which the Gamma index is adaptive to the proportion of pixels in the dark area,robustly improve the illumination distribution of the illumination image,the enhanced multi-scale brightness components are obtained by Retinex inverse transformation.Finally,the linear fusion of enhanced multi-scale lightness components is used as the enhanced brightness componentⅠ’,which is recombined with the hue component H and saturation component S to obtain the final enhanced image.The experimental results show that compared with some existing newly proposed methods,the proposed method are 1.19 dB and 1.8%higher than those of the some existing methods proposed recently,the proposed method shows better enhancement effect and robustness,the visual effect of its enhanced images is more suitable with higher contrast.

关 键 词：图像增强 多尺度RETINEX HSI颜色空间 自适应Gamma矫正 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]