A Semantic Segmentation Method for High-resolution Remote Sensing Images Based on Encoder-Decoder

Abstract

Image segmentation is a key technology in remote sensing image interpretation, and it is widely used in many fields. Aiming at the common problems of low segmentation accuracy and blurred target boundary in the semantic segmentation of high-resolution remote sensing images, a semantic segmentation method of high-resolution remote sensing images based on encoder-decoder structure is proposed, in which an attention mechanism is introduced to highlight important features, and an optimized Pyramid pooling module is used to extract multi-scale features from different layers. Finally, a multi-level and multi-scale feature fusion strategy is adopted to achieve fine-grained segmentation of high-resolution remote sensing images. The method is also compared and tested on the ISPRS Vaihingen dataset to verify the effectiveness.