基于Retinex的改进夜视高光抑制视频增强算法  

Improved Night Vision High Light Suppression Enhancement Algorithm Based on Retinex

作　　者：朱俊[1] ZHU Jun(School of Computer Science and Engineering,Nanjing University of Science and Technology,Nanjing 210094)

机构地区：[1]南京理工大学计算机科学与工程学院,南京210094

出　　处：《计算机与数字工程》2018年第6期1208-1211,共4页Computer & Digital Engineering

摘　　要：该算法是一种基于Retinex理论的具有高光抑制的夜间图像增强方法,该方法按照以下步骤实施:1)图像的色系转换,从RGB色系转换到HSV色系中,保留颜色信息的H和S通道信息,只对亮度通道V进行处理。2)图像的预增强处理。3)确定抑制参数,包括抑制模板和抑制范围;再进行高光区域的抑制。4)直方图展宽处理,利用高光抑制时产生的空灰度级,将图像像素灰度级进行重新分布,展宽直方图。5)图像归一化,再从HSV色系转换到RGB色系下进行显示,得到增强的夜间图像。该算法方法能够有效地增强夜间图像,并抑制图像中的高光,在不改变原本图像像素间关系的同时,最大限度地提高图像的对比度,使得图像能够表述最大信息量。This algorithm is a kind of high light inhibition of night image enhancing method which is based on the Retinex theo-ry,the method is implemented according to the following steps：1）image color conversion,the conversion from RGB to HSV colorcolor,color information is retained H and S channel information,only to the luminance channel V for processing. 2）pre image en-hancement processing steps. 3）determine the inhibition parameters,including inhibition of template and inhibiting range;then in-hibit the high light area. 4）histogram stretching,use gray level have high light inhibition,the pixels of the image gray level histo-gram for redistribution,broadening. 5）image normalization,conversion from HSV to RGB color system color display,enhancednight image. This algorithm can enhance the night image effectively,and suppress the high light in the image,without changing theoriginal image pixel relationship at the same time,the maximum increase the contrast of the image,the image can represent the max-imum amount of information.

关 键 词：车载夜视 夜间图像 图像归一化 直方图展宽 

分 类 号：TP391.91[自动化与计算机技术—计算机应用技术]