DEANet: Decomposition Enhancement and Adjustment Network for Low-Light Image Enhancement
This addresses the problem of poor image quality in low-light conditions for computer vision applications, though it appears incremental as it builds on Retinex theory.
The paper tackles low-light image enhancement by proposing DEANet, a Retinex-based network that decomposes, enhances, and adjusts images, achieving better visual and quality results than existing state-of-the-art methods on the LOL dataset.
Images obtained under low-light conditions will seriously affect the quality of the images. Solving the problem of poor low-light image quality can effectively improve the visual quality of images and better improve the usability of computer vision. In addition, it has very important applications in many fields. This paper proposes a DEANet based on Retinex for low-light image enhancement. It combines the frequency information and content information of the image into three sub-networks: decomposition network, enhancement network and adjustment network. These three sub-networks are respectively used for decomposition, denoising, contrast enhancement and detail preservation, adjustment, and image generation. Our model has good robust results for all low-light images. The model is trained on the public data set LOL, and the experimental results show that our method is better than the existing state-of-the-art methods in terms of vision and quality.