• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学 ›› 2020, Vol. 42 ›› Issue (07): 1244-1252.doi: 10.3969/j.issn.1007-130X.2020.07.013

• 图形与图像 • 上一篇    下一篇

基于L0范数的Retinex图像增强算法

陈茹霞1,强振平1,邵小锋1,何丽波2   

  1. (1.西南林业大学大数据与智能工程学院,云南 昆明 650224;2.云南警官学院信息网络安全学院,云南 昆明 650223)

  • 收稿日期:2019-10-28 修回日期:2020-01-13 接受日期:2020-07-25 出版日期:2020-07-25 发布日期:2020-07-27
  • 基金资助:
    国家自然科学基金(61941204);云南省应用基础研究计划(2020FB138);西南林业大学科研启动基金(111827)

A Retinex image enhancement algorithm based on  L0-norm

CHEN Ru-xia1,QIANG Zhen-ping1,SHAO Xiao-feng1,HE Li-bo2   

  1. (1.School of Big Data and Intelligent Engineering,Southwest Forestry University,Kunming 650224;

    2.Information Security College,Yunnan Police College,Kunming 650223,China)

  • Received:2019-10-28 Revised:2020-01-13 Accepted:2020-07-25 Online:2020-07-25 Published:2020-07-27

摘要: 图像在采集过程中会因为机械设备、天气状况等原因产生曝光不均等问题,使得图像的拍摄效果不佳,无法满足实际应用的需求。而传统的Retinex算法应用于图像增强时会导致图像边缘模糊与泛灰等问题。因此,针对传统的Retinex算法现存的问题,提出一种新颖的图像增强算法——基于L0范数的Retinex算法RIEALN。首先通过全局L0梯度最小化方法提取图像的轮廓成分,然后进行Retinex算法处理,再将提取的轮廓成分融合到原始图像,实现原始图像的增强。实现过程中还通过增加不同的L0梯度最小化因子确保不同尺度轮廓成分的均匀增强。实验结果表明,该算法不仅可以增强图像的对比度,而且还能够较好地保留边缘信息。

关键词: Retinex算法, L0范数, 提取轮廓, 图像增强

Abstract: In the image acquisition process, some issues such as uneven exposure caused by mechanical equipment, weather conditions and other reasons will make the quality of the taken picture poor and unable to meet the needs of practical applications. When the traditional Retinex algorithm is applied to enhance the image, it will cause the problems such as blurred image edges and graying. Therefore, aiming at the existing problems of the traditional Retinex algorithm, this paper proposes a novel image enhancement algorithm: Retinex image enhancement algorithm based on  L0-norm. Firstly, the contour components of the image are extracted by the global   L0 gradient minimization method and processed by the Retinex algorithm, and then the extracted contour components are fused to the original image to realize the enhancement of the original image. In the implementation of the algorithm, different global  L0  gradient minimization factors are used to ensure the uniform enhancement of contour components at different scales. Experimental results show that the proposed algorithm can better preserve the edge information of the image while enhancing the contrast of the image.




Key words: Retinex algorithm, L0-norm, contour extraction, image enhancement

中图分类号: