Remote sensing image segmentation network based on tree structure

Abstract

The invention relates to a remote sensing image segmentation network based on a tree structure, and belongs to the technical field of computer vision. The remote sensing image segmentation network isa DeepLab V3 + structured tree network model, and the tree network model comprises a segmentation module and a tree processing module which are connected in sequence; the segmentation module is a DeepLab V3 + network model, and the DeepLab V3 + network model comprises an encoder part and a decoder part. The construction method of the tree-shaped processing module comprises the steps of constructing a confusion matrix, calculating a lower triangular matrix, establishing a confusion undirected complete graph, performing iterative edge cutting operation on the confusion undirected complete graph,and obtaining the tree-shaped processing module shown in the specification. The confusable pixels can be better distinguished, a more accurate segmentation result is obtained, the overall semantic segmentation precision of the high-resolution remote sensing image is effectively improved, and the segmentation accuracy of the confusable category data is remarkably improved.

Description

technical field [0001] The invention relates to a remote sensing image segmentation network based on a tree structure, specifically a remote sensing image segmentation network combining DeepLab V3+ and a tree structure, and belongs to the technical field of computer vision. Background technique [0002] Semantic segmentation of high-resolution remote sensing images refers to the task of assigning semantic labels to each pixel in the image. In recent years, with the rapid development of remote sensing mapping technology, we can easily obtain ultra-high resolution optical remote sensing images with a ground sampling interval (GSD) of 5 to 10 cm. Based on this, how to segment these images accurately and efficiently has become a research hotspot in the field of remote sensing image segmentation. For ultra-high resolution remote sensing image data, most traditional methods rely on supervised classifiers with artificially designed features for segmentation. While artificially de...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem of low accuracy of confusing category data segmentation, the purpose of the present invention is to provide a remote sensing image segmentation network that mixes DeepLab V3+ and tree structure, aiming at the segmentation problem of confusing category data in ultra-high resolution remote sensing images , to improve its segmentation accuracy

Images