A Residual Complex Fourier for Road Segmentation in Remote Sensing Images

Abstract

Road segmentation from remote sensing images is crucial in various applications, including map navigation, urban planning, vegetation change analysis, and disaster assessment. However, problems such as minor interclass differences, shadow masking, and narrow or interrupted roads frequently occur, making it difficult for most methods relying on convolution or self-attention to quickly and comprehensively extract these road objects. In this article, we propose a residual complex Fourier network (RCFNet), which utilizes Fourier encoding operators to quickly and accurately extract roads. First, a Fourier encoder module is proposed to extract road information using complex-valued features in the frequency domain. Second, a road information mix module is introduced to align frequency-domain features across different scales. Finally, a multifeature fusion strategy is proposed to calibrate multilevel features from the decoder to enhance road segmentation results. Extensive experiments on two public datasets, the DeepGlobe road dataset and the Massachusetts road dataset, have demonstrated that RCFNet achieves notable performance with minimal computational cost (float point of operations).