Deep learning-based multi-spectral satellite image segmentation for water body detection

Automated water body detection from satellite imagery is a fundamental stage for urban hydrological studies. In recent years, various deep convolutional neural network (DCNN)-based methods have been proposed to segment remote sensing data collected by conventional RGB or multi-spectral imagery for such studies. However, how to effectively explore the wider spectrum bands of multi-spectral sensors to achieve significantly better performance compared to the use of only RGB bands has been left underexplored. In this paper, we propose a novel deep convolutional neural network model – Multi-Channel Water Body Detection Network (MC-WBDN) – that incorporates three innovative components, a multi-channel fusion module, an Enhanced Atrous Spatial Pyramid Pooling (EASPP) module, and Space-to-Depth (S2D)/Depth-to-Space (D2S) operations, to outperform state-of-the-art DCNN-based water body detection methods. Experimental results convincingly show that our MCWBDN model achieves remarkable water body detection performance, is more robust to light and weather variations and can better distinguish tiny water bodies compared to other DCNN models.