基于多尺度Retinex的雾天降质图象增强算法  

Enhancement Algorithm of Fog-degraded Image Based on Multiscale Retinex

作　　者：赵晓霞[1,2] 王汝琳[1] 李雪艳[1] 

机构地区：[1]中国矿业大学(北京)机电与信息工程学院,北京100083 [2]中北大学信息与通信工程学院,山西太原030051

出　　处：《工矿自动化》2009年第10期62-66,共5页Journal Of Mine Automation

基　　金：国家自然科学基金(50534060)

摘　　要：雾天图象受到严重退化,大大降低了图象的使用价值。利用Retinex理论的颜色恒常性,通过提取照度分量来去除雾天对图象的退化作用。指出MSR算法处理雾天图象的不足,提出将MSR算法处理结果与原图象结合的改进方法:引入原图象对数分量和颜色信息,以保留原图象亮度的全局对比度和恢复平坦区域的颜色;提出进一步增强雾天图象细节信息的方法:将MSR处理结果与边缘增强的原始图象相结合,然后对输出图象各颜色分量先进行正态截取,再统一拉伸到显示设备的动态范围。实验结果表明,该方法能取得较好的图象去雾效果。The image is often significantly degraded by fog and mist and its value is greatly reduced.According to color constancy of Retinex theory,the fog degradation can be eliminated by extracting illumination component.Shortages of fog images proccessed by MSR algorithm were pointed out and an improved algorithm combining the resulting image by MSR algorithm with the original image,which uses logarithm component and color information of the original image to keep global contrast and recover colors of flat areas.A method to enhance details of fog images further was proposed: Firstly,the resulting image is obtained by recombination of the output of MSR with the original image with enhanced edges.Secondly,color components of the resulting image are respectively clipped,then all pixels are scaled up to the dynamic range of display device.The experiment result showed that the algorithm can effectively remove fog degradation from images.

关 键 词：图象处理 雾天降质图象 图象增强 多尺度RETINEX MSR 改进 

分 类 号：TD679[矿业工程—矿山机电] TN941.1[电子电信—信号与信息处理]