多尺度Retinex灰度图像增强算法  

An Enhancing Algorithm of the Grey Image Based on MultiScale Retinex

作　　者：宋凯[1] 沈红[1] 刘昶[1] 

机构地区：[1]沈阳理工大学信息科学与工程学院,辽宁沈阳110168

出　　处：《辽宁大学学报（自然科学版）》2008年第1期46-48,共3页Journal of Liaoning University：Natural Sciences Edition

基　　金：863计划项目(2003AA41250);辽宁省教育厅A类基金(20243303)

摘　　要：为提高图像识别在分类时的质量,必须在图像的预处理阶段对噪声进行滤除,对图像中的目标对象加以增强.从研究图像增强的空域法入手,利用多尺度Retinex灰度图像增强算法完成图像的增强.做到改善图像颜色恒常性,压缩图像动态范围,提高对比度,有效显示淹没在阴影、光照等区域中的细节.在仿真实验中,对图像进行高斯滤波,确定了高斯滤波系数.仿真结果证明该方法可行,在完成图像增强的同时,对噪声有较好的抑制作用.To improve the quality of classification in image Recognition, the noise must be removed to enhance the object during the pretreatment of color image. Began with image enhance in airspace, MultiScale Retinex was used to realize enhancing of the grey image firstly. The invariance of color was improved, image dynamic range was compressed to improve balance, the details in shadow and light area were effectively showed. In emulational example, the Gauss filter parameters were determined, and the image was enhanced. The result was given to illustrate the efficiency of the method, At the same time the image was enhanced , the noise was controlled.

关 键 词：多尺度RETINEX 图像增强 高斯滤波器 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]