Sea-Land Segmentation in Remote Sensing Images

Sea-land segmentation addresses the delineation of coastal boundaries and the separation of aquatic from terrestrial regions in remotely sensed imagery. Accurate segmentation underpins a wide range of applications, from coastline change detection and maritime navigation to coastal ecosystem monitoring and disaster response. The principal challenge arises from the high variability in spectral signatures, atmospheric effects, seasonal changes, and mixed-pixel phenomena at the water–land interface. Traditional thresholding and edge-detection methods have given way to data-driven approaches, most notably deep learning frameworks that combine hierarchical feature extraction with end-to-end training. Contemporary models aim to balance the capture of fine boundary detail with robust global context, often by integrating multi-scale feature fusion, attention mechanisms and hybrid encoder designs. Progress in this field has been catalysed by the growing availability of high-resolution multispectral and radar data, as well as benchmark datasets that characterise diverse coastal environments. As segmentation accuracy improves, downstream tasks such as automatic coastline extraction, change quantification and habitat mapping become more reliable, thereby supporting sustainable coastal management and hazard mitigation worldwide.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Sea-Land Segmentation in Remote Sensing Images publication trend

The graph below shows the total number of articles in sea-land segmentation in remote sensing images across all publications each year (not limited to Nature Index journals).

Technical terms

Semantic segmentation: The process of assigning a class label to each pixel in an image, enabling the separation of distinct regions such as sea and land.

Convolutional Neural Network (CNN): A deep learning architecture that applies convolutional filters to extract hierarchical features from images.

Transformer: A neural network model employing self-attention mechanisms to capture long-range dependencies and global context within data.

U-Net: A symmetric encoder-decoder CNN architecture with skip connections designed for precise image segmentation tasks.

Intersection over Union (IoU): A standard metric for evaluating segmentation accuracy, defined as the ratio of the overlap between predicted and ground-truth regions to their union.

Normalised Difference Water Index (NDWI): A spectral index derived from green and near-infrared bands, used to enhance water feature detection in optical imagery.

References

1. A Novel Deep Structure U-Net for Sea-Land Segmentation in Remote Sensing Images. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (2019).
2. Automatic coastline extraction through enhanced sea-land segmentation by modifying Standard U-Net. International Journal of Applied Earth Observation and Geoinformation (2022).
3. TCUNet: A Lightweight Dual-Branch Parallel Network for Sea–Land Segmentation in Remote Sensing Images. Remote Sensing (2023).