期刊
INTERNATIONAL JOURNAL OF COMPUTER VISION
卷 130, 期 6, 页码 1440-1458出版社
SPRINGER
DOI: 10.1007/s11263-022-01583-y
关键词
Low-level vision; Image restoration; Image filtering; Image enhancement; Dual convolutional neural network
资金
- National Key Research and Development Program of China [2018AAA0102001]
- National Natural Science Foundation of China [61872421, 61922043, 61925204]
- Fundamental Research Funds for the Central Universities [30920041109]
- NSF CAREER [1149783]
This paper proposes a general dual parallel convolutional neural network (DualCNN) for addressing low-level vision problems by recovering the structures and details of target signals to generate desired signals. Experimental results demonstrate that DualCNN performs well in numerous low-level vision tasks, outperforming specially designed state-of-the-art methods.
We propose a general dual convolutional neural network (DualCNN) for low-level vision problems, e.g., super-resolution, edge-preserving filtering, deraining, and dehazing. These problems usually involve estimating two components of the target signals: structures and details. Motivated by this, we design the proposed DualCNN to have two parallel branches, which respectively recovers the structures and details in an end-to-end manner. The recovered structures and details can generate desired signals according to the formation model for each particular application. The DualCNN is a flexible framework for low-level vision tasks and can be easily incorporated into existing CNNs. Experimental results show that the DualCNN can be effectively applied to numerous low-level vision tasks with favorable performance against the state-of-the-art methods that have been specially designed for each individual task.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据