A Review of Modern Low Light Image Enhancement Techniques

Abstract

Low-light image enhancement is a critical area in computer vision. It aimed to improve the quality of images captured under imperfect lighting and restore the image with a high-accuracy color distributed with normal lighting, leading to high similarity to the natural image conditions. The low-light images suffer from low contrast, unclear details, and intensive noise. This paper provides a review of the field of low-light image enhancement in terms of low-light image enhancement techniques, including traditional and deep learning methods. This paper presents a classification of LLIE traditional techniques divided into four categories, including histogram equalization (HE), Retinex theory, Dehazing, and gamma correction (GC) methods, as well as the methods based on deep learning.