压扩多尺度Retinex图像增强算法研究  

Research on Compressed Multiscale Retinex Image Enhancement Algorithms

作　　者：刘明珠[1] 苏桃宇 王依宁 LIU Ming-zhu;SU Tao-yu;WANG Yi-ning(School of Measure-control Technology and Communication Engineering,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学测控技术与通信工程学院,哈尔滨150080

出　　处：《哈尔滨理工大学学报》2020年第5期93-99,共7页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(61401126);黑龙江省自然科学基金(F2016021)。

摘　　要：在低照度条件下对人脸图像进行光照补偿和利用带色彩恢复的Retinex算法是传统图像增强算法中较为有效的一种。针对传统的MSRCR算法在处理图像后仍会产生清晰度低、细节模糊和色彩不饱满等问题。因此本文提出带色彩恢复的压扩多尺度Retinex算法—CMSRCR,利用多彩色通道变曲变换函数及因子修正法,对低照度人脸图像进行图像增强和色彩恢复。结果表明,CMSRCR算法能够使低照度图像在清晰度、细节和颜色恢复等方面都有明显的改善。The Retinex algorithm for face image illumination compensation and color restoration is an effective one in the traditional image enhancement algorithm under the condition of low illumination. The traditional MSRCR algorithm can still produce some problems such as low sharpness, blurred details and incomplete color after image processing. Therefore, this paper proposes a compression expansion multi-scale Retinex algorithm with color recovery-CMSRCR, which uses multi-color channel bending transformation function and factor correction method to enhance and restore low-illumination face images. The results show that the CMSRCR algorithm can improve the sharpness, detail and color restoration of low illumination image.

关 键 词：图像增强 低照度图像 压扩变换 RETINEX算法 

分 类 号：TN911.73[电子电信—通信与信息系统]