LLNet: A deep autoencoder approach to natural low-light image enhancement

By Kin Gwn Lore, Adedotun Akintayo, Soumik Sarkar. Pattern Recognition, 2015

1,822 citations recorded by the source corpus.

Abstract In surveillance, monitoring and tactical reconnaissance, gathering visual information from a dynamic environment and accurately processing such data are essential to making informed decisions and ensuring the success of a mission. Camera sensors are often cost-limited to capture clear images or videos taken in a poorly-lit environment. Many applications aim to enhance brightness, contrast and reduce noise content from the images in an on-board real-time manner. We propose a deep autoencoder-based approach to identify signal features from low-light images and adaptively brighten images without over-amplifying/saturating the lighter parts in images with a high dynamic range. We show that a variant of the stacked-sparse denoising autoencoder can learn from synthetically darkened and noise-added training examples to adaptively enhance images taken from natural low-light environment and/or are hardware-degraded. Results show significant credibility of the approach both visually and by quantitative comparison with various techniques.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/9f29aad5bb5f511c4aac3ef414dc70ba1768059f

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

GLADNet: Low-Light Enhancement Network with Global Awareness (2018, methodology): “[8] used variant of the stacked-sparse denoising auto-encoder learning from synthetically darkened and noise-added training examples for simultaneously low-light enhancement and noise reduction (LLNet).”
URetinex-Net: Retinex-based Deep Unfolding Network for Low-light Image Enhancement (2022, background): “Due to these limitations existing in model-based methods, researchers take advantage of deep networks [3,11,15, 25,34,42] to restore low-light images in a data-driven manner.”
Low-Light Image Enhancement via a Deep Hybrid Network (2019, background): “[8] generate low-light images by darkening natural”
Attention Guided Low-Light Image Enhancement with a Large Scale Low-Light Simulation Dataset (2019, methodology): “Due to the MLP architecture, LLNet [50] needs to enhance large images one patch by one patch, which will limits its efficiency.”
Learning Enriched Features for Real Image Restoration and Enhancement (2020, neutral): “Notable works employ Retinex-inspired networks [88,106,122], encoder-decoder networks [17,67,80], and generative adversarial networks [18,50,24].”
Structure-Revealing Low-Light Image Enhancement Via Robust Retinex Model (2018, methodology): “[7] proposed a Low-Light Net (LLNet) using deep autoencoders to simultaneously (or sequentially) perform”
Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancement (2023, methodology): “[34] propose a stacked sparse denoising auto-encoder LLNet for joint low-light enhancement and noise suppression.”
Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement (2020, methodology): “For example, the LLNet [16] was trained on data simulated on random Gamma correction; the LOL dataset [27] of paired low/normal light images was collected through altering the exposure time and ISO during image acquisition; the MIT-Adobe FiveK dataset [3] comprises 5,000 raw images, each of which ha”
Underexposed Photo Enhancement Using Deep Illumination Estimation (2019, background): “[21] proposed a deep autoencoderbased approach for low-light image enhancement, while Gharbi et al.”
Learning a Simple Low-Light Image Enhancer from Paired Low-Light Instances (2023, positive): “[19] designed a stacked sparse denoising auto-encoder to enhance lowlight images.”
Getting to Know Low-light Images with The Exclusively Dark Dataset (2018, methodology): “In order to solve this noise issue, existing off-the-shelf denoising algorithms were commonly employed as a post-processing step ([17], [18], [49]) or by incorporating mechanisms in the proposed models ([13], [11]).”
Deep Joint Rain Detection and Removal from a Single Image (2016, background): “There are also some recent works on bad weather restoration or image enhancement, such as dehazing [8, 37], raindrop and dirt removal [13] and light enhancement [27].”
Low-Light Image Enhancement With Semi-Decoupled Decomposition (2020, background): “Lore et al. [19] proposed the seminal work based on the auto-encoder model.”
Ultra-High-Definition Low-Light Image Enhancement: A Benchmark and Transformer-Based Method (2022, methodology): “By applying random gamma correction and Gaussian noise, Lore et al. (Lore, Akintayo, and Sarkar 2017) synthesized 422 , 500 low-light image patches from 169 images.”
LR3M: Robust Low-Light Enhancement via Low-Rank Regularized Retinex Model (2020, methodology): “There are several kinds of methods to enhance low-light images, including histogram equalization [2], [3], inverse domain operation [4], [5], Retinex decomposition [6]–[9], and deep learning [10], [11].”
Low-Light Image Enhancement with Normalizing Flow (2021, methodology): “In addition, different losses are used to guide the training, e.g., MSE (Lore et al. 2017; Cai, Gu, and Zhang 2018), l 1 loss(Cai, Gu, and Zhang 2018), structural similarity (SSIM) (Cai, Gu, and Zhang 2018), smoothness loss (Wang et al. 2019a; Zhang et al. 2019) and color loss (Wang et al. 2019a;…”
EnlightenGAN: Deep Light Enhancement Without Paired Supervision (2019, methodology): “[4] proposed a stacked auto-encoder (LL-Net) to learn joint denoising and low-light enhancement on the patch level.”