SCR-Net: A Dual-Channel Water Body Extraction Model Based on Multi-Spectral Remote Sensing Imagery-A Case Study of...

Area of Science:

• Remote Sensing
• Artificial Intelligence
• Ecological Monitoring

Background:

• Current semantic segmentation models for lake monitoring often neglect multi-spectral data, leading to inaccurate boundary detection and loss of small water bodies.
• Existing algorithms have limited practical applicability and require further validation in real-world scenarios.

Purpose of the Study:

• To develop an advanced water body identification model for multi-spectral remote sensing images.
• To enhance the accuracy of lake segmentation and monitoring for ecological assessments.

Main Methods:

• Introduction of SCR-Net, a dual-channel encoding-decoding model designed for multi-spectral remote sensing.
• Altering the number of image data channels to improve feature learning for lakes and target location extraction.
• Training and validation on multi-spectral remote sensing datasets.

Main Results:

• SCR-Net demonstrates superior segmentation accuracy compared to state-of-the-art models on multiple datasets.
• The model effectively leverages spectral properties for improved water body identification.
• Case study on Daihai Lake validates its practical application in lake area calculation.

Conclusions:

• SCR-Net offers a significant advancement in water body identification using multi-spectral remote sensing.
• The model provides valuable insights for ecological environment monitoring and remote sensing image processing.
• Accurate lake area monitoring is crucial for regional ecological balance assessment.