Lingnan scholar Prof Sam Kwong’s research team wins 2025 IEEE SPS Best Paper Award

Prof Sam Kwong Tak-Wu, Associate Vice-President (Strategic Research), Dean of the School of Graduate Studies and J.K. Lee Chair Professor of Computational Intelligence at Lingnan University, together with scholars from the City University of Hong Kong, the Chinese Academy of Sciences, Beijing Jiaotong University, The University of Sydney, Tianjin University, and other institutions, has received the IEEE Signal Processing Society (IEEE SPS) 2025 Best Paper Award for the research paper titled “An Underwater Image Enhancement Benchmark Dataset and Beyond”.

This prestigious award is rigorously selected by the IEEE Signal Processing Society from outstanding papers published in its associated journals over the past six years. It represents one of the highest international academic recognitions of the research’s scholarly value and innovative contribution, and not only highlights the exceptional strength of Prof Kwong’s team, but also Lingnan’s leading position in the fields of computer vision and signal processing.

The influential IEEE Signal Processing Society is one of the largest and most authoritative professional academic bodies worldwide in the field. The Best Paper Award was established to recognise outstanding achievements that have made radical progress, exerted real academic impact, and demonstrated strong application - making it a major honour for researchers.

The award-winning paper was published in IEEE Transactions on Image Processing, the IEEE’s top-tier image processing journal. It is also in the highest ranked Q1 Journal Citation Reports (JCR) with the latest 2024 Impact Factor of 13.7, coming 8th in both Computer Science (Artificial Intelligence) and Engineering (Electrical & Electronic Engineering). The paper focuses on underwater image enhancement, a research area of great academic importance and broad application potential, and provides a crucial benchmark dataset and strong theoretical support.

Underwater environments are highly complex and challenging. Phenomena such as light attenuation, scattering, and “marine snow” frequently result in low visibility, poor contrast, and severe colour distortion in underwater images, seriously hindering progress in marine engineering and biology, underwater robotics, and related fields. The research team built the first large-scale, physical underwater image enhancement benchmark dataset for the study, comprising 950 authentic underwater images and 890 corresponding reference images. This work fills a critical gap in high-quality real-world datasets.

Building upon this, the team conducted a comprehensive and systematic qualitative and quantitative evaluation of current mainstream underwater image enhancement algorithms, providing essential benchmarks and references for future development. They also proposed the innovative Water-Net deep learning model, which removes colour distortion and improves image clarity, significantly outperforming traditional methods, and laying a solid foundation for the practical application of underwater image enhancement technology.

Lingnan University will continue to support its leading scholars’ valuable academic research, encourage interdisciplinary collaboration and innovation, and improve international exchange, thereby contributing to solving key technological challenges and advancing progress in relevant fields.

For details of the full paper, please visit: An Underwater Image Enhancement Benchmark Dataset and Beyond | IEEE Journals & Magazine | IEEE Xplore.