image enhancement for scattering media

Underwater visual enhancement (UVE) and 3D reconstruction face severe challenges due to inherent optical distortions—namely scattering, absorption, and chromatic aberration—while existing literature lacks a systematic survey on their synergistic development. This paper presents the first unified review of UVE and underwater 3D reconstruction, covering classical models, deep learning, Neural Radiance Fields (NeRF), 3D Gaussian Splatting, and physics-guided approaches. We establish a multi-dimensional evaluation framework and conduct quantitative and qualitative comparisons across mainstream benchmarks, revealing performance boundaries and application suitability under diverse degradation conditions. Furthermore, we identify three key future directions: physics-model integration, cross-domain generalization, and real-time robust reconstruction. This work provides both theoretical foundations and practical guidance for advancing underwater visual understanding and geometric reconstruction.

Underwater image degradation severely impairs object detection performance. This work systematically evaluates the impact of nine state-of-the-art underwater image enhancement methods—including physics-based, non-physics-based, and deep learning approaches—on modern detectors such as YOLOv8 and RT-DETR, correlating enhancement outcomes with both UIQM/UCIQE quality metrics and detection mAP. We uncover, for the first time, a dual paradox: while enhancement consistently degrades average detection accuracy across datasets (negative correlation with mAP), it significantly improves detection on specific individual images (positive correlation at the per-image level). Motivated by this finding, we propose a novel “image-wise adaptive enhancement” paradigm, identifying a subset of enhancement methods that reliably boost per-image detection performance. Extensive experiments confirm that enhancement yields no universal gain; instead, its benefit is highly image-dependent. To foster reproducibility and further research, we publicly release all source code, enhanced images, detection outputs, and analytical tools.

This work proposes a diffusion-based image enhancement method to address the challenges of underwater photography, including low contrast, spatial blur, and wavelength-dependent color distortion caused by light absorption and scattering. The approach uniquely integrates a physics-inspired synthetic underwater degradation pipeline with a diffusion generative model, enabling the learning of an inverse mapping of the degradation process. Trained from scratch on only approximately 2.5k high-quality images, the resulting 11-million-parameter model demonstrates exceptional perceptual fidelity and generalization capability at a resolution of 512×768, significantly improving the visual quality of underwater imagery.

Underwater images commonly suffer from low contrast, blurriness, and chromatic distortion. Existing methods often couple haze removal and color correction in a single model, neglecting their physical independence and synergistic interaction. This paper proposes WaterFormer, a decoupled Vision Transformer architecture: it employs dedicated dehazing and color restoration blocks to model these two degradation processes separately, and introduces a channel fusion block for dynamic inter-block coordination. A soft reconstruction layer, grounded in the underwater imaging physical model, is incorporated to enhance fidelity. Furthermore, we propose a joint optimization strategy combining chromatic consistency loss and Sobel-based color loss to simultaneously preserve color accuracy and structural details. Extensive experiments on multiple benchmark datasets demonstrate that WaterFormer achieves state-of-the-art performance in PSNR, SSIM, and human perceptual evaluation—significantly improving image contrast, sharpness, and color fidelity.

Underwater images suffer severe degradation due to wavelength-dependent light absorption and scattering, yet existing physics-guided methods are hindered by inaccurate estimation of depth and scattering parameters, resulting in poor generalization. To address this, we propose a physics-guided joint training framework featuring the novel Depth-Decoupled Degradation Model (DDM), which explicitly disentangles veiling light, degradation factors, and scene depth. We further design a three-branch subnetwork and a dual-branch UIEConv module to embed underwater imaging physical priors directly into end-to-end optimization. Our method achieves state-of-the-art PSNR/SSIM performance on real-world underwater scenes—including deep-sea environments with artificial illumination—while simultaneously producing high-fidelity depth maps. Notably, it is the first approach to jointly enhance image quality and depth estimation accuracy, thereby enabling physically consistent, dual-task support for underwater 3D perception.

Underwater images suffer severe degradation due to wavelength-dependent absorption and scattering in water, limiting the generalizability of prior-based methods and hindering purely data-driven deep learning approaches due to scarce annotated training data. To address these challenges, we propose the Physics-Aware Triple-Stream Network (PAT-Net), the first framework that tightly integrates a physics-guided triple-stream architecture—explicitly estimating direct transmission, backscattering, and ambient light—with an IFM (Implicit Feature Matching)-driven semi-supervised learning paradigm. PAT-Net jointly optimizes physical parameter estimation and image enhancement in an end-to-end manner, enhancing interpretability and physical consistency via improved IFM regularization. Its semi-supervised design significantly reduces reliance on labeled data. Extensive experiments across six benchmark datasets demonstrate that PAT-Net achieves state-of-the-art performance in both degradation parameter estimation and perceptual/quantitative image quality enhancement, consistently outperforming 16 representative baseline methods.

This work addresses the degradation of visual quality in underwater images captured in real oceanic environments, which is primarily caused by depth-dependent forward scattering blur and marine snow artifacts. To tackle this challenge, the authors propose a staged image enhancement framework that explicitly models the depth-dependent forward scattering effect for the first time and extracts realistic marine snow degradation patterns from authentic underwater imagery. These components are leveraged to generate high-fidelity synthetic data for fine-tuning a Joint-ID network, followed by a lightweight contrast enhancement post-processing step. The approach effectively bridges the domain gap between synthetic and real underwater images, yielding significant improvements in UIQM scores and perceptual clarity on a real-world dataset collected off the coast of Korea, thereby enhancing the usability of underwater imagery for robotic operations.

This work addresses the challenges of underwater image degradation—such as color distortion, low contrast, and haze—caused by complex aquatic environments. Existing learning-based methods often rely on noisy pseudo-labels, yielding unreliable supervision. To overcome this limitation, the authors propose a novel unsupervised enhancement framework that decomposes the task according to a physical imaging model into three stages: global color correction, dehazing, and background noise suppression. Crucially, the method leverages cross-task transfer priors from related visual domains as unpaired supervisory signals. This approach uniquely integrates physics-driven task decomposition with cross-domain transfer priors, achieving state-of-the-art performance across multiple quantitative and qualitative benchmarks. Moreover, it significantly improves the efficacy of downstream vision tasks, demonstrating both theoretical soundness and strong generalization capability.

This study addresses the lack of a systematic evaluation framework for underwater image reconstruction, which has hindered comprehensive assessment of methods in terms of accuracy, viewpoint consistency, and robustness to varying water conditions. To bridge this gap, the authors propose the first multidimensional evaluation framework that jointly considers reconstruction fidelity, camera motion consistency, and the impact of water quality, accompanied by a newly curated real-world underwater image dataset for benchmarking. Through extensive experiments comparing traditional physics-based scattering models with emerging vision-language models (VLMs), the results demonstrate that VLMs—despite eschewing explicit physical modeling—consistently outperform conventional approaches across all evaluated metrics, achieving superior reconstruction quality and generalization capability.

Underwater images suffer from color distortion and detail blurring due to light absorption and scattering, which severely degrades object detection performance. To address this, this work proposes a dual-branch collaborative enhancement framework: a detail enhancement branch improves brightness and local contrast in dark regions, while a color restoration branch reduces color casts through adaptive compensation. The decoupled design of the two branches balances visual quality with the requirements of downstream tasks. Integrated with YOLOv8, the method enables end-to-end optimization, achieving UIQM scores of 2.249 and 2.576 on the UIEB and EUVP datasets, respectively, and improving YOLOv8’s mAP50 by 2.1% on the URPC dataset, thereby significantly enhancing detection accuracy in complex underwater scenes.

Underwater images suffer from severe degradations—including low visibility, texture blurring, color distortion, and noise—yet existing enhancement methods exhibit poor generalizability and heavily rely on scarce, high-quality paired ground-truth data; moreover, prevailing benchmarks lack diversity and predominantly consist of single-view images. To address these limitations, this work introduces the first integration of denoising diffusion probabilistic models (DDPMs) with ControlNet for underwater image enhancement: DDPMs synthesize diverse, high-fidelity underwater images spanning multiple illumination conditions, viewpoints, and scenes, while ControlNet ensures structural fidelity and controllable enhancement. Crucially, our framework eliminates dependence on real clean reference images, substantially improving dataset diversity and model generalizability. Extensive experiments demonstrate superior performance over state-of-the-art methods in both quantitative metrics (PSNR/SSIM) and downstream tasks (e.g., marine species recognition), establishing a more robust and scalable data foundation for oceanic ecological monitoring.