Underwater images play a crucial role in various fields, including oceanographic engineering, marine exploitation, and marine environmental protection. However, the quality of underwater images is often severely degraded due to the complexities of the underwater environment and equipment limitations. This degradation hinders advancements in relevant research. Consequently, underwater image restoration has gained significant attention as a research area. With the growing interest in deep-sea exploration, deep-sea image restoration has emerged as a new focus, presenting unique challenges. This paper aims to conduct a systematic review of underwater image restoration technology, bridging the gap between shallow-sea and deep-sea image restoration fields through experimental analysis. This paper first categorizes shallow-sea image restoration methods into three types: physical model-based methods, prior-based methods, and deep learning-based methods that integrate physical models. The core concepts and characteristics of representative methods are analyzed. The research status and primary challenges in deep-sea image restoration are then summarized, including color cast and blur caused by underwater environmental characteristics, as well as insufficient and uneven lighting caused by artificial light sources. Potential solutions are explored, such as applying general shallow-sea restoration methods to address color cast and blur, and leveraging techniques from related fields like exposure image correction and low-light image enhancement to tackle lighting issues. Comprehensive experiments are conducted to examine the feasibility of shallow-sea image restoration methods and related image enhancement techniques for deep-sea image restoration. The experimental results provide valuable insights into existing methods for addressing the challenges of deep-sea image restoration. An in-depth discussion is presented, suggesting several future development directions in deep-sea image restoration. Three main points emerged from the research findings: i) Existing shallow-sea image restoration methods are insufficient to address the degradation issues in deep-sea environments, such as low-light and uneven illumination. ii) Combining imaging physical models with deep learning to restore deep-sea image quality may potentially yield desirable results. iii) The application potential of unsupervised and zero-shot learning methods in deep-sea image restoration warrants further investigation, given their ability to work with limited training data.

Sea ice segmentation based on Synthetic Aperture Radar (SAR) images has become an important technical means for polar climate change monitoring and navigation safety guarantee. However, the existing methods have limitations in the utilization of SAR polarization information and the modeling of local diversity details of sea ice, which leads to insufficient segmentation, especially in complex ice-water boundary regions. To address these issues, this paper proposes a novel Polarization-Fused Edge-Enhanced UNet (PFEE-UNet) designed specifically for sea ice segmentation from high-resolution SAR images. Specifically, we design the Cross-Polarization Channel Interaction (CPCI) module, which employs a dual interaction strategy of hierarchical inter-group cascading and symmetric cross-fusion. This approach effectively leverages the complementary features of the HH and HV polarization channels, significantly enhancing the distinction between sea ice and open water. Additionally, we present the Dense–Sparse Diversity Enhancement (DSDE) module, which combines a spatial-channel joint attention mechanism to strengthen the model’s ability to capture spatial relationships within complex ice–water structures, effectively alleviating misclassifications caused by abrupt local texture changes. Finally, we design the Selective Edge Fusion (SEF) module, which dynamically selects and integrates multi-level edge features, improving the continuity of sea ice boundaries and preserving its morphological integrity. The experimental results show that the proposed PFEE-UNet model outperforms mainstream segmentation methods on the AI4Arctic/ASIP sea ice dataset, achieving an average Intersection over Union (IoU) of 84.48%, which surpasses existing methods such as HRNet (82.52%) and DeepLabv3+ (82.40%). Additionally, PFEE-UNet was applied for end-to-end ice–water segmentation on real-world Sentinel-1 SAR scenes, demonstrating its effectiveness and robustness for practical sea ice monitoring.