一种简单的单幅灰度图像高光检测与恢复方法  

A Simple Highlight Detection and Recovery Methods for Single Grayscale Image

作　　者：尹芳[1,2] 陈田田[1] 付自如 于晓洋[2] YIN Fang;CHEN Tian-tian;FU Zi-ru;YU Xiao-yang(School of Computer Science and Technology,Harbin University of Science and Technology,Harbin 150080,China;Instrument Science and Technology Postdoctoral Research Station,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学计算机科学与技术学院,黑龙江哈尔滨150080 [2]哈尔滨理工大学仪器科学与技术博士后科研流动站,黑龙江哈尔滨150080

出　　处：《哈尔滨理工大学学报》2018年第2期86-90,共5页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(61440025);黑龙江省教育厅科学技术研究项目(12541119)

摘　　要：高光的检测与去除一直是计算机视觉领域的一个热点问题,现有的大多数方法主要都是针对彩色图像,但是灰度图像的应用又很广泛,对于只包含亮度信息的灰度图像的高光检测和去除是一个难点问题,针对这一问题,提出了一种简单的单幅灰度图像高光检测方法。该方法对二维亮度饱和度直方图方法进行改进,并利用漫反射分量和镜面反射分量的分布获取高光亮度值范围,对可能存在的高光区域进行检测,最后,利用基于BSCB模型的图像修复方法去除高光。实验结果表明本文算法细节处理的较好,能够有效地检测出灰度图像中镜面反射区域,提高了图像高光区域的修复率。Specular detection and removal has been a hot topic in the field of computer vision.Most of the existing methods are for color images,but grayscale images are widely used,for a single grayscale image with only intensity information,specular detection and removal becomes a difficult issue.In this paper,a simple specular detection method of single grayscale image is proposed.Intensity-Saturation(MS)diagram method is improved and the distribution of diffuse component and specular component is used to obtain high brightness range.Possible specular area is detected with proportion.Finally,BSCB model-based the image restoration method removal highlights.Experimental results show that specular detection algorithm can better deal with the details of specular areas,find specular reflection area effectively in the single grayscale image and improve the repair rate specular areas of the image.

关 键 词：高光检测 反射模型 二维亮度饱和度直方图 BSCB模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]