Abstract:Low-light images incur several complicated degradation factors such as poor brightness, low contrast, color degradation, and noise. Most previous deep learning-based approaches, however, only learn the mapping relationship of single channel between the input low-light images and the expected normal-light images, which is insufficient enough to deal with low-light images captured under uncertain imaging environment. Moreover, too deeper network architecture is not conducive to recover low-light images due to extremely low values in pixels. To surmount aforementioned issues, in this paper we propose a novel multi-branch and progressive network (MBPNet) for low-light image enhancement. To be more specific, the proposed MBPNet is comprised of four different branches which build the mapping relationship at different scales. The followed fusion is performed on the outputs obtained from four different branches for the final enhanced image. Furthermore, to better handle the difficulty of delivering structural information of low-light images with low values in pixels, a progressive enhancement strategy is applied in the proposed method, where four convolutional long short-term memory networks (LSTM) are embedded in four branches and an recurrent network architecture is developed to iteratively perform the enhancement process. In addition, a joint loss function consisting of the pixel loss, the multi-scale perceptual loss, the adversarial loss, the gradient loss, and the color loss is framed to optimize the model parameters. To evaluate the effectiveness of proposed MBPNet, three popularly used benchmark databases are used for both quantitative and qualitative assessments. The experimental results confirm that the proposed MBPNet obviously outperforms other state-of-the-art approaches in terms of quantitative and qualitative results. The code will be available at https://github.com/kbzhang0505/MBPNet.

SBC-Net: semantic-guided brightness curve estimation network for low-light image enhancement

Automatical Enhancement and Denoising of Extremely Low-light Images.

Low Light Enhancement by Unsupervised Network.

Low-Light Image Enhancement Based on Deep Convolutional Neural Network

Semantically-guided low-light image enhancement

Mutual Support and Promotion: Learning Structure Compensation and Context Completion for Low-Light Vision

Multi-Branch and Progressive Network for Low-Light Image Enhancement

Joint semantic-aware and noise suppression for low-light image enhancement without reference

Towards a Flexible Semantic Guided Model for Single Image Enhancement and Restoration

A Semantic-Aware Detail Adaptive Network for Image Enhancement

Multi-scale Deep Curve Estimation for Low-light Image Enhancement.

BLNet: A Fast Deep Learning Framework for Low-Light Image Enhancement with Noise Removal and Color Restoration

Brighten-and-Colorize: A Decoupled Network for Customized Low-Light Image Enhancement

Progressive Two-Stage Network for Low-Light Image Enhancement

Brightness Perceiving for Recursive Low-Light Image Enhancement

LEES-Net: Fast, Lightweight Unsupervised Curve Estimation Network for Low-Light Image Enhancement and Exposure Suppression.

Low-Illumination Image Enhancement Based on End-to-End Network Using Attention Module

Fusion-Based Low-Light Image Enhancement

Semantic-Guided Zero-Shot Learning for Low-Light Image/Video Enhancement

Division gets better: Learning brightness-aware and detail-sensitive representations for low-light image enhancement

TSN-CA: A Two-Stage Network with Channel Attention for Low-Light Image Enhancement