Unformer: A Transformer-Based Approach for Adaptive Multi-Scale Feature Aggregation in Underwater Image Enhancement

Yuhao Qing,Yueying Wang,Huaicheng Yan,Xiangpeng Xie,Zhengguang Wu

DOI: https://doi.org/10.1109/tai.2024.3508667

2024-01-01

IEEE Transactions on Artificial Intelligence

Abstract:Underwater imaging is often compromised by light scattering and absorption, resulting in image degradation and distortion. This manifests as blurred details, color shifts, and diminished illumination and contrast, thereby hindering advancements in underwater research. To mitigate these issues, we propose Unformer, an innovative underwater image enhancement technique that leverages a transformer-based architecture for multi-scale adaptive feature aggregation. Our approach employs a multi-scale feature fusion strategy that adaptively restores illumination and detail features. We reevaluate the relationship between convolution and transformer to develop a novel encoder structure. This structure effectively integrates both long-range and short-range dependencies, dynamically combines local and global features, and constructs a comprehensive global context. Furthermore, we propose a unique multi-branch decoder architecture that enhances and efficiently extracts spatial context information through the transformer module. Extensive experiments on three datasets demonstrate that our proposed method outperforms other techniques in both subjective and objective evaluations. Compared with the latest methods, Unformer has improved the PSNR by 19.5% and 14.8% respectively on the LSUI and EUVP datasets. The code is available at: https://github.com/yhflq/Unformer .