Hierarchical guided network for low‐light image enhancement

Abstract

Due to insufficient illumination in low‐light conditions, the brightness and contrast of the captured images are low, which affect the processing of other computer vision tasks. Low‐light enhancement is a challenging task that requires simultaneous processing of colour, brightness, contrast, artefacts and noise. To solve this problem, the authors apply the deep residual network to the low‐light enhancement task, and propose a hierarchical guided low‐light enhancement network. The key of this method is recombined hierarchical guided features through the feature aggregation module to realize low‐light enhancement. The network is based on the U‐Net network, and then hierarchically guided with the input pyramid branch in the encoding and decoding network. The input pyramid structure realizes multi‐level receptive fields and generates a hierarchical representation. The encoding and decoding structure concatenates the hierarchical features of the input pyramid and generates a set of hierarchical features. Finally, the feature aggregation module is used to fuse different features to achieve low‐light enhancement tasks. The effectiveness of the components is proved through ablation experiments. In addition, the authors are also evaluating on different data sets, and the experimental results show that the method proposed is superior to other methods in subjective and objective evaluation.

Unpaired strong-light image augmentation network with illumination curve

“…L = 𝑤 1 * 𝐿 𝑐 + 𝑤 2 * 𝐿 𝑒 + 𝑤 3 * 𝐿 𝑠 (7) This paper contains three loss functions, namely color constancy loss function, exposure control loss function and illumination smoothing loss function. These three loss functions are indispensable to our network.…”

Research on Unsupervised Low-Light Railway Fastener Image Enhancement Method Based on Contrastive Learning GAN

“…To better handle the details and noise of low-light images, Feng et al [ 17 ] proposed a layered guided low-light image enhancement network. Using dual-tree complex wavelet transform (DT-CWT) for brightness guidance and multiple branches for weak light enhancement tasks, this method can generate more realistic low-light images and improve image detail and clarity.…”

“…This paper introduces an illumination loss function to constrain the illumination distribution of the generated images, making railway fastener images appear more natural and consistent under various lighting conditions. For example, compared to the method by Feng et al [ 17 ], the illumination loss function in this paper more effectively adjusts the illumination distribution of the images, improving their naturalness and consistency.…”