Abstract: Image sensors imaging in low-light environments can cause problems such as high image noise, low contrast, and failure to represent large amounts of detailed information, which affect not only human eye observation but also applications related to computer vision and low-light image processing. The use of image enhancement methods and deep learning methods can improve the contrast of low-light images and improve the image quality. The article first introduces the traditional low-illumination image enhancement algorithms categorized and summarizes the improvement process of these algorithms in recent years, then introduces the low-illumination image enhancement methods based on deep learning, and finally introduces the existing low-illumination image datasets and the evaluation criteria of the enhanced images.
Paper
Full text
A Survey of Low-Light Image Enhancement
Semantic Scholar · Computer Science · 2022
Abstract
Abstract: Image sensors imaging in low-light environments can cause problems such as high image noise, low contrast, and failure to represent large amounts of detailed information, which affect not only human eye observation but also applications related to computer vision and low-light image processing. The use of image enhancement methods and deep learning methods can improve the contrast of low-light images and improve the image quality. The article first introduces the traditional low-illumination image enhancement algorithms categorized and summarizes the improvement process of these algorithms in recent years, then introduces the low-illumination image enhancement methods based on deep learning, and finally introduces the existing low-illumination image datasets and the evaluation criteria of the enhanced images.