DichroGAN: Towards Restoration of in-Air Colours of Seafloor from Satellite Imagery

2025-12-03

International Conference on Digital Image Computing: Techniques and Applications

AI Summary

This work addresses the challenging problem of recovering the true in-air color of the seafloor from satellite imagery, which is severely degraded by exponential light attenuation with depth in water. To this end, the authors propose DichroGAN, a novel conditional generative adversarial network featuring a unique four-generator collaborative architecture. The first two generators estimate diffuse and specular reflectance components from hyperspectral images to derive atmospheric radiance, while the latter two jointly model spectral characteristics and underwater light transmission, simultaneously compensating for absorption and scattering effects based on the underwater imaging equation. By integrating hyperspectral information with a physically grounded imaging model, DichroGAN enables end-to-end color reconstruction under limited satellite training data. Experiments on PRISMA satellite imagery and underwater datasets demonstrate that DichroGAN achieves or surpasses state-of-the-art performance in seafloor color recovery accuracy.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Recovering the in-air colours of seafloor from satellite imagery is a challenging task due to the exponential attenuation of light with depth in the water column. In this study, we present DichroGAN, a conditional generative adversarial network (cGAN) designed for this purpose. DichroGAN employs a two-steps simultaneous training: first, two generators utilise a hyperspectral image cube to estimate diffuse and specular reflections, thereby obtaining atmospheric scene radiance. Next, a third generator receives as input the generated scene radiance containing the features of each spectral band, while a fourth generator estimates the underwater light transmission. These generators work together to remove the effects of light absorption and scattering, restoring the in-air colours of seafloor based on the underwater image formation equation. DichroGAN is trained on a compact dataset derived from PRISMA satellite imagery, comprising RGB images paired with their corresponding spectral bands and masks. Extensive experiments on both satellite and underwater datasets demonstrate that DichroGAN achieves competitive performance compared to state-of-the-art underwater restoration techniques.

Problem

• underwater colour restoration
• satellite imagery
• in-air seafloor colours
• light attenuation
• image formation

Innovation

• DichroGAN
• conditional GAN
• underwater color restoration
• hyperspectral imaging
• two-stage training