基于Retinex模型的彩色图像全局增强算法  

Global Color Image Enhancement Algorithm Based on Retinex Model

作　　者：李福文[1] 金伟其[1] 陈伟力[1] 曹扬[1] 王霞[1] 王岭雪[1] 

机构地区：[1]北京理工大学光电学院

出　　处：《北京理工大学学报》2010年第8期947-951,共5页Transactions of Beijing Institute of Technology

基　　金：国家“八六三”计划项目(2006AA03A151)

摘　　要：介绍了基于颜色恒常性理论的亮度感知Retinex模型,通过对HSV空间亮度图像的快速高斯滤波和全局γ变换,估计场景的光照信息,获得仅包含物体本身特性的反射信息图像,进而实现图像增强.主观视觉观察以及客观定量计算结果均表明,对于受光照等因素影响的可视性较差的图像,基于Retinex模型的彩色图像全局增强算法效果十分显著.Retinex as a luminance perceptual model based on color consistency is introduced. Illumination image is evaluated using a fast Gaussian filtering and global γtransform to original image value of HSV space.Then the reflection image including intrinsic properties of objects is obtained and image enhancement is achieved.Visual observation and quantitative analysis show that the global enhancement algorithm has significant effect on images affected by real illumination factor.

关 键 词：Retinex模型 彩色图像增强 图像质量评价 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]