Underwater scene prior inspired deep underwater image and video enhancement

By Chongyi Li, Saeed Anwar, Fatih Murat Porikli. Pattern Recognition, 2020

1,140 citations recorded by the source corpus.

Abstract In underwater scenes, wavelength-dependent light absorption and scattering degrade the visibility of images and videos. The degraded underwater images and videos affect the accuracy of pattern recognition, visual understanding, and key feature extraction in underwater scenes. In this paper, we propose an underwater image enhancement convolutional neural network (CNN) model based on underwater scene prior, called UWCNN. Instead of estimating the parameters of underwater imaging model, the proposed UWCNN model directly reconstructs the clear latent underwater image, which benefits from the underwater scene prior which can be used to synthesize underwater image training data. Besides, based on the light-weight network structure and effective training data, our UWCNN model can be easily extended to underwater videos for frame-by-frame enhancement. Specifically, combining an underwater imaging physical model with optical properties of underwater scenes, we first synthesize underwater image degradation datasets which cover a diverse set of water types and degradation levels. Then, a light-weight CNN model is designed for enhancing each underwater scene type, which is trained by the corresponding training data. At last, this UWCNN model is directly extended to underwater video enhancement. Experiments on real-world and synthetic underwater images and videos demonstrate that our method generalizes well to different underwater scenes.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/0de1a592912915a10fcd8637886203b688c03650

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

PUGAN: Physical Model-Guided Underwater Image Enhancement Using GAN With Dual-Discriminators (2023, methodology): “The comparison methods includes 6 non-learning-based methods ( i . e ., GDCP [4], ACDE [25], HLRP [67], MLLE [68], UNTV [69], and SPDF [70]) and 8 deep learning-based methods ( i . e ., deep-sesr [59], FUnIE-GAN [5], WaterNet [2], UWCNN [46], JI-Net [56], ACPAB [53], TOPAL [55], and Ucolor [6]).”
U-Shape Transformer for Underwater Image Enhancement (2021, methodology): “Besides Train-L, the second training set Train-U contains 800 pairs of underwater images from UIEB [26] and 1,250 synthetic underwater images from [25]; the third training set Train-E contains the paired training images in the EUVP [19] dataset.”
Using labeled data to evaluate change detectors in a multivariate streaming environment (2009, neutral): “Due to the need to divide ice into shore ice and drift ice, we also carried out experiments on NWPU_YRCC2 dataset.”
Underwater Image Enhancement With Hyper-Laplacian Reflectance Priors (2022, positive): “Lately, Li et al. [27] synthesized paired training data based on underwater scene prior and proposed an underwater image enhancement network (UWCNN).”
A Deep Journey into Super-resolution (2019, methodology): “Alongside SR, deep learning algorithms have shown promising results on other sub-ﬁelds in Artiﬁcial Intelligence [13] such as object classiﬁcation [14] and detection [15], natural language processing [16], [17], image processing [18], [19], and audio signal processing [20].”
Underwater Image Enhancement via Medium Transmission-Guided Multi-Color Space Embedding (2021, methodology): “For UWCNN [12], we used the original UWCNN models, in which each UWCNN model was trained using the underwater images synthesized by one type of water.”
An Underwater Image Enhancement Benchmark Dataset and Beyond (2019, methodology): “, UWCNNs) trained by ten types of underwater images was proposed in [52], where underwater images are synthesized based on a revised underwater image formation model [6] and the corresponding underwater scene parameters.”
Underwater Image Enhancement via Weighted Wavelet Visual Perception Fusion (2024, methodology): “Li et al. [51] proposed an underwater enhancement CNN inspired by underwater scenes prior, which utilized the water types and the degradation levels to construct synthesized underwater images.”