中国邮电高校学报(英文版) ›› 2022, Vol. 29 ›› Issue (2): 63-72.doi: 10.19682/j.cnki.1005-8885.2022.0004

• Image Processing • 上一篇    下一篇

Multi-scale fusion residual encoder-decoder approach for low illumination image enhancement

潘晓英1,魏苗1,王昊2,贾丰竹1   

  1. 1. 西安邮电大学
    2.
  • 收稿日期:2021-01-13 修回日期:2021-07-09 出版日期:2022-04-26 发布日期:2022-04-26
  • 通讯作者: 魏苗 E-mail:weim2018@163.com
  • 基金资助:
    国家重点研发计划项目;西安邮电大学研究生创新基金项目;陕西省网络数据分析与智能处理重点实验室专项基金资助课题

Multi-scale fusion residual encoder-decoder approach for low illumination image enhancement

  • Received:2021-01-13 Revised:2021-07-09 Online:2022-04-26 Published:2022-04-26
  • Supported by:
    This work was supported by the National Key Research and Development Project ( 2019YFC0121502 ), the Graduate Innovation Fund Project of Xi'an University of Posts and Telecommunications ( CXJJLY2019062 ), and the Key Laboratory of Network Data Analysis and Intelligent Processing of Shaanxi Province.

摘要:

The sensing light source of the line scan camera cannot be fully exposed in a low light environment due to the extremely small number of photons and high noise, which leads to a reduction in image quality. A multi-scale fusion residual encoder-decoder (FRED) was proposed to solve the problem. By directly learning the end-to-end mapping between light and dark images, FRED can enhance the image's brightness with the details and colors of the original image fully restored. A residual block (RB) was added to the network structure to increase feature diversity and speed up network training. Moreover, the addition of a dense context feature aggregation module (DCFAM) made up for the deficiency of spatial information in the deep network by aggregating the context's global multi-scale features. The experimental results show that the FRED is superior to most other algorithms in visual effect and  quantitative evaluation of peak signa-to-noise ratio (PSNR) and structural similarity index measure (SSIM). For the factor that FRED can restore the brightness of images while representing the edge and color of the image effectively, a satisfactory visual quality is obtained under the enhancement of low-light.

关键词: 图像增强|低照度|特征融合|残差网络

Abstract:

The sensing light source of the line scan camera cannot be fully exposed in a low light environment due to the extremely small number of photons and high noise, which leads to a reduction in image quality. A multi-scale fusion residual encoder-decoder (FRED) was proposed to solve the problem. By directly learning the end-to-end mapping between light and dark images, FRED can enhance the image's brightness with the details and colors of the original image fully restored. A residual block (RB) was added to the network structure to increase feature diversity and speed up network training. Moreover, the addition of a dense context feature aggregation module (DCFAM) made up for the deficiency of spatial information in the deep network by aggregating the context's global multi-scale features. The experimental results show that the FRED is superior to most other algorithms in visual effect and  quantitative evaluation of peak signa-to-noise ratio (PSNR) and structural similarity index measure (SSIM). For the factor that FRED can restore the brightness of images while representing the edge and color of the image effectively, a satisfactory visual quality is obtained under the enhancement of low-light.

Key words: image enhancement| low illumination|feature fusion| residual network