A research team led by Professor QIN Hongde has achieved a major advance in underwater imaging technology, with their findings published in Information Fusion, a top-tier international journal in the field of artificial intelligence.
The paper, titled Diffusion-based frequency degradation prior fusion with hierarchical wavelet decompositions for underwater image enhancement, lists CHEN Haojie, a PhD candidate at the College of Shipbuilding Engineering, as first author, and Professor WANG Zhuo as corresponding author.
Addressing long-standing technical limitations in underwater image processing, the team developed an innovative FPG-Diff framework for underwater image enhancement and fusion. Combining the advantages of frequency-domain diffusion and wavelet decomposition techniques, the approach fundamentally reduces model computational complexity, overcoming the traditional industry trade-off between image quality and processing efficiency.
Testing on authoritative public datasets confirms that the new technology outperforms all existing mainstream algorithms in overall performance. It achieves leading results in core evaluation metrics including Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM), as well as in subjective visual quality. Even in challenging underwater conditions such as low illumination and high turbidity, the algorithm accurately restores image colour and fine detail, with strong robustness and generalisation capability.
The breakthrough will significantly enhance the visual perception capabilities of underwater equipment, with applications across a range of marine engineering scenarios including autonomous underwater vehicle (AUV) exploration, underwater emergency rescue and subsea facility inspection. It provides a new technical solution for underwater visual perception systems in smart ocean development.
Information Fusion is a leading international journal in artificial intelligence, classified as a Top journal in Zone 1 of the Chinese Academy of Sciences (CAS) journal ranking. It focuses on cutting-edge research including multimodal information fusion, intelligent perception and distributed sensing, with a latest impact factor of 17.4.
