《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (2): 552-559.DOI: 10.11772/j.issn.1001-9081.2022010093

• 多媒体计算与计算机仿真 • 上一篇    

注意力机制下的多阶段低照度图像增强网络

谌贵辉1, 林瑾瑜1(), 李跃华2, 李忠兵1, 魏钰力1, 卢凯1   

  1. 1.西南石油大学 电气信息学院,成都 610500
    2.南充职业技术学院 机电工程系,四川 南充 637131
  • 收稿日期:2022-01-25 修回日期:2022-04-13 接受日期:2022-04-18 发布日期:2022-04-21 出版日期:2023-02-10
  • 通讯作者: 林瑾瑜
  • 作者简介:谌贵辉(1971—),男,四川成都人,教授,硕士,主要研究方向:数字图像处理
    李跃华(1971—),男,四川蓬安人,讲师,硕士,主要研究方向:物理、电气自动化
    李忠兵(1985—),男,湖北武汉人,讲师,博士,主要研究方向:数字图像处理
    魏钰力(1999—),男,四川绵阳人,硕士研究生,主要研究方向:大数据分析
    卢凯(1997—),男,四川成都人,硕士研究生,主要研究方向:数字图像处理。
  • 基金资助:
    南充市市校科技战略合作项目(21SXHZ020)

Multi-stage low-illuminance image enhancement network based on attention mechanism

Guihui CHEN1, Jinyu LIN1(), Yuehua LI2, Zhongbing LI1, Yuli WEI1, Kai LU1   

  1. 1.School of Electrical Engineering and Information,Southwest Petroleum University,Chengdu Sichuan 610500,China
    2.Department of Mechanical and Electrical Engineering,Nanchong Vocational and Technical College,Nanchong Sichuan 637131,China
  • Received:2022-01-25 Revised:2022-04-13 Accepted:2022-04-18 Online:2022-04-21 Published:2023-02-10
  • Contact: Jinyu LIN
  • About author:CHEN Guihui, born in 1971, M. S., professor. His research interests include digital image processing.
    LI Yuehua, born in 1971, M. S., lecturer. His research interests include physics, electrical automatization.
    LI Zhongbing, born in 1985, Ph. D., lecturer. His research interests include digital image processing.
    WEI Yuli, born in 1999, M. S. candidate. His research interests include big data analysis.
    LU Kai, born in 1997, M. S. candidate. His research interests include digital image processing.
  • Supported by:
    Nanchong City-University Science and Technology Strategic Cooperation Project(21SXHZ020)

摘要:

对于低照度图像增强过程中,因图像内容重叠且部分区域亮度差异较大导致的图像细节丢失的问题,提出一个注意力机制下的多阶段低照度图像增强网络。第一阶段利用改进的多尺度融合模块对图像进行初步增强;第二阶段利用第一阶段增强后的图像信息与本阶段的输入进行级联,并将其结果作为该阶段多尺度融合模块的输入;第三阶段利用第二阶段增强后的图像信息与该阶段的输入级联,并将其结果作为该阶段多尺度融合模块的输入。这样利用多阶段的方式完成自适应的亮度提升和细节的保留。在公开数据集LOL和SICE上的实验结果表明,相较于MSR算法、灰度直方图均衡化(HE)算法和RetinexNet等算法和网络,所提网络的峰值信噪比(PSNR)的数值提高了11.0%~28.9%,结构相似性(SSIM)的数值提高了6.8%~46.5%。所提网络利用多阶段和注意力机制实现低照度图像增强,有效解决了图像内容重叠和亮度差异大的问题,得到的图像细节更丰富,纹理更清晰,主观辨识度更高。

关键词: 低照度图像, 注意力机制, 多阶段, 多尺度, 自适应

Abstract:

A multi-stage low-illuminance image enhancement network based on attention mechanism was proposed to solve the problem that the details of low-illuminance images are lost due to the overlapping of image contents and large brightness differences in some regions during the enhancement process of low-illuminance images. At the first stage, an improved multi-scale fusion module was used to perform preliminary image enhancement. At the second stage, the enhanced image information of the first stage was cascaded with the input of this stage, and the result was used as the input of the multi-scale fusion module in this stage. At the third stage, the enhanced image information of the second stage was cascaded with the input of the this stage, and the result was used as the input of the multi-scale fusion module in this stage. In this way, with the use of multi-stage fusion, not only the brightness of the image was improved adaptively, but also the details were retained adaptively. Experimental results on open datasets LOL and SICE show that compared to the algorithms and networks such as MSR (Multi-Scale Retinex) algorithm, gray Histogram Equalization (HE) algorithm and RetinexNet (Retina cortex Network), the proposed network has the value of Peak Signal-to-Noise Ratio (PSNR) 11.0% to 28.9% higher, and the value of Structural SIMilarity (SSIM) increased by 6.8% to 46.5%. By using multi-stage method and attention mechanism to realize low-illuminance image enhancement, the proposed network effectively solves the problems of image content overlapping and large brightness difference, and the images obtained by this network are more detailed and subjective recognizable with clearer textures.

Key words: low-illumination image, attention mechanism, multi-stage, multi-scale, adaptive

中图分类号: