Remote sensing image semantic segmentation model and method based on double-layer global convolution

Abstract

The invention belongs to the technical field of image processing, and relates to a remote sensing image semantic segmentation model and method based on double-layer global convolution, and the model comprises a feature extraction network, an upper layer branch and a lower layer branch which are used for processing features of different layers, and a feature fusion network. The high-level features and the low-level features output by the feature extraction network are respectively enhanced by an upper-layer branch and a lower-layer branch, and then feature fusion is carried out; the upper layer branch comprises a patch attention module I and a global convolution module I; and the lower layer branch comprises a patch attention module II, a global convolution module II, an attention embedding module and a global convolution module III. According to the invention, an attention embedding module is adopted to embed local attention into low-level features from high-level features, so that context information can be embedded into the low-level features; a global convolution module is adopted to enlarge a receptive field in a combined convolution form, and the segmentation performance of a large-size ground object target is improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a remote sensing image semantic segmentation model and method based on double-layer global convolution. Background technique [0002] Remote sensing images have become the main data source for obtaining surface information. Semantic segmentation of remote sensing images is widely used in land monitoring, road detection, environmental monitoring and other fields. With the continuous development of satellite remote sensing technology, the resolution of remote sensing images has been greatly improved, small objects in high-resolution remote sensing images are presented, and the size difference of ground objects has become a new challenge for semantic segmentation of remote sensing images. Different types of ground objects in remote sensing images may share similar spectral features, so convolutional neural networks need to be used to improve the segmentation eff...

AI Technical Summary

Problems solved by technology

Although the classical semantic segmentation network is constantly exploring in obtaining richer spatial context information, due to the characteristics of complex remote sensing image backgrounds and large size differences between classes, it is difficult to extract rich spatial context information between ground objects, resulting in segmentation boundaries. Relatively rough, small targets are easy to be missed, and large targets are difficult to be completely segmented

Images