基于残差融合的改进Retinex图像增强算法
IMPROVED RETINEX IMAGE ENHANCEMENT ALGORITHM BASED ON RESIDUAL FUSION
王奎 1黄福珍1
作者信息
- 1. 上海电力大学自动化工程学院 上海 200090
- 折叠
摘要
针对低照度图像边缘纹理模糊、亮度和对比度偏低等问题,提出一种基于残差融合的改进Retinex图像增强算法.该算法采用 自适应多尺度引导滤波AMGF(Adaptive Multi-scale Guided Filtering,AMGF)替代高斯核函数,根据Retinex理论获取反射图像;使用CLAHE拉伸反射图像的对比度;通过L0范数提取输入图像残差进行融合;进行颜色恢复处理.实验结果表明,所提算法有效地提升了低照度图像的边缘细节表达能力,提高了图像质量和视觉效果.
Abstract
In order to solve the problems of blurred edge texture,low brightness and low contrast of low-illumination image,an improved Retinex image enhancement algorithm based on residual fusion is proposed.Adaptive multi-scale guided filtering(AMGF)was used to replace Gaussian kernel function,and reflected image was obtained according to Retinex theory.The contrast of reflected image was stretched by CLAHE.The input image residual was extracted by L0 norm for fusion.Color recovery processing was carried out.Experimental results show that the edge detail expression ability of low illumination images is improved by the proposed algorithm effectively,and images quality and visual effect are improved.
关键词
Retinex/自适应多尺度引导滤波/CLAHE/L0范数/颜色恢复Key words
Retinex/AMGF/CLAHE/L0norm/Color recovery引用本文复制引用
基金项目
上海市电站自动化技术重点实验室项目(13DZ2273800)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据