| 引用本文: | 余映,徐超越,李淼,等.金字塔渐进融合低照度图像增强网络.[J].国防科技大学学报,2024,46(2):224-237.[点击复制] |
| YU Ying,XU Chaoyue,LI Miao,et al.Pyramid asymptotic fusion low-illumination image enhancement network[J].Journal of National University of Defense Technology,2024,46(2):224-237[点击复制] |
|
| |
|
|
| 本文已被：浏览 3182次 下载 1736次 |
| 金字塔渐进融合低照度图像增强网络 |
| 余映,徐超越,李淼,何鹏浩,杨昊 |
|
(云南大学 信息学院, 云南 昆明 650500)
|
| 摘要: |
| 针对现有低照度图像增强网络对不同尺度特征信息存在感知与表达能力不足的问题,提出金字塔渐进融合低照度图像增强网络模型。网络对图像进行多次下采样操作以组成特征金字塔,通过在特征金字塔的三个不同分支上加入跳跃连接,将不同尺度的特征图进行相互融合。通过精细恢复模块进一步提取精炼信息,将特征图恢复到正常的光照图像。结果表明,该网络模型不但能有效地提升低照度图像的整体亮度,而且能很好地保持图像中的细节信息和清晰的物体边缘轮廓,同时能够有效地抑制图像中的暗部噪声,使增强后的图像整体画面真实自然。 |
| 关键词: 低照度图像增强 深度学习 特征金字塔 多尺度特征 跳跃连接 |
| DOI：10.11887/j.cn.202402023 |
| 投稿日期：2022-01-06 |
| 基金项目:国家自然科学基金资助项目(62166048,61263048)；云南省应用基础研究计划资助项目(2018FB102)；云南大学“中青年骨干教师培养计划”资助项目(XT412003) |
|
| Pyramid asymptotic fusion low-illumination image enhancement network |
| YU Ying, XU Chaoyue, LI Miao, HE Penghao, YANG Hao |
|
(School of Information Science and Engineering, Yunnan University, Kunming 650500, China)
|
| Abstract: |
| Since existing low-illumination image enhancement networks have insufficient ability to perceive and express feature information of different scales, a low-illumination image enhancement network model based on pyramid asymptotic fusion was proposed. The network performed multiple down-sampling operations on the image to form a feature pyramid. It fused the feature maps at different scales by adding skip connections to three different branches of the feature pyramid. Fine recovery module further extracted the refined information, and restored the feature map to a normal light image. Results indicate that, the network model not only effectively enhances the brightness of the overall low-illumination image, but also maintains the detailed information and clear edge contours of the objects in the image. Moreover, it can effectively suppress the dark noise, and make the overall enhanced image realistic and natural. |
| Keywords: low-illumination image enhancement deep learning feature pyramid multi-scale features skip connection |
|
|
|
|
|
