电子学报 ›› 2022, Vol. 50 ›› Issue (2): 415-425.DOI: 10.12263/DZXB.20210177

• 学术论文 • 上一篇    下一篇

结合多尺度循环卷积和多聚类空间的红外图像增强

路皓翔1, 刘振丙1, 张静2, 王子民1   

  1. 1.桂林电子科技大学计算机与信息安全学院,广西 桂林 541004
    2.桂林电子科技大学商学院,广西 桂林 541004
  • 收稿日期:2021-01-27 修回日期:2021-09-29 出版日期:2022-02-25 发布日期:2022-02-25
  • 通讯作者: 刘振丙
  • 作者简介:路皓翔 男,1991年10月生,河南安阳人.博士研究生.主要研究方向为图像处理、深度学习.E-mail: 646510477@qq.com
    刘振丙(通讯作者) 男,1980年6月生,广西桂林人.博士、教授,博士生导师.主要研究方向为医学图像处理、深度学习.
  • 基金资助:
    国家自然科学基金(61866009);广西重点研发项目(2017GXNSFDA198025);广西创新驱动重大专项(AA17202024)

Infrared Image Enhancement Based on Multi-Scale Cyclic Convolution and Multi-Clustering Space

LU Hao-xiang1, LIU Zhen-bing1, ZHANG Jing2, WANG Zi-min1   

  1. 1.School of Computer Science and Information Security,Guilin University of Electronic Technology,Guilin,Guangxi 541004,China
    2.School of Business,Guilin University of Electronic Technology,Guilin,Guangxi 541004,China
  • Received:2021-01-27 Revised:2021-09-29 Online:2022-02-25 Published:2022-02-25
  • Contact: LIU Zhen-bing

摘要:

红外图像低对比度、低亮度及轮廓细节模糊等问题限制了红外成像技术的应用.为了提高红外图像的质量,本文提出了一种结合多尺度循环卷积和多聚类空间红外图像增强方法,该方法分为五个阶段:多尺度循环卷积、多聚类空间混合均衡化、多角度组合滤波器、线性融合和图像层次化处理.其中,通过多尺度循环卷积和多角度组合滤波器增强图像细节及轮廓信息;多聚类空间混合均衡化通过改进的K-means对图像灰度值进行聚类分析,依据图像特性选取不同的均衡化方式提升图像对比度及亮度;图像层次化处理用于提升图像清晰度.结果表明:与对比方法相比,该方法能够显著增强红外图像细节及轮廓,提升红外图像对比度和亮度.

关键词: 红外图像, 图像增强, 聚类分析, 对比度拉伸

Abstract:

Infrared image has problems of low contrast, low brightness and blurring of contour details, limiting the application of the infrared imaging technology. To improve the quality of infrared image, an infrared image enhancement method based on multi-scale cyclic convolution and multi-clustering space is proposed. The method contains five stages: multi-scale cyclic convolution, hybrid equalization in the clustering space, multi-angle combination filter, linear fusion, and image hierarchical processing. The details and contour information of the image are enhanced by multi-scale cyclic convolution and multi-angle combination filter, and the hybrid equalization of the clustering space performs clustering analysis on the image gray value through the improved K-means. According to the image characteristics, different equalization methods are selected to improve the contrast and brightness of the image. The hierarchical processing of image is used to improve clarity of the image. Experimental results show that the proposed method not only can improve the details and contour of infrared image, but also can enhance the contrast and brightness of infrared image compared with the contrast method.

Key words: infrared image, image enhancement, cluster analysis, contrast stretch

中图分类号: