Remote sensing image classification method and device based on priori geometric constraint and electronic equipment

Abstract

The invention provides a remote sensing image classification method and device based on priori geometric constraints and electronic equipment. The method comprises the steps: obtaining a remote sensing image and historical spatial distribution data of the remote sensing image, wherein the historical spatial distribution data is used for representing a spatial granularity condition of a historicalclassification result of the remote sensing image; inputting the remote sensing image and the historical spatial distribution data into a pre-trained deep learning classification model, and outputtinga classification result of the remote sensing image, wherein the deep learning classification model is obtained by training based on historical remote sensing images, classification labels of the historical remote sensing images and historical spatial distribution data polygons. In the mode, the historical spatial distribution data is used as the pattern spot granularity prior data of the remotesensing image, the spatial scale of the remote sensing image can be constrained, and the pattern spot granularity and boundary trend of the remote sensing image can be determined, so the accuracy, reliability and popularization applicability of the remote sensing image classification method based on prior geometric constraints are improved.

Description

technical field [0001] The invention relates to the technical field of remote sensing image classification, in particular to a remote sensing image classification method, device and electronic equipment based on prior geometric constraints. Background technique [0002] In recent years, deep learning technology has developed rapidly and has been widely used in the field of deep learning classification of remote sensing images. The traditional deep learning of remote sensing images takes image blocks as objects, and completes deep learning classification through two processes of sample training and sample prediction: (1) Sample training: using a certain amount of representative image blocks and corresponding classification results, under the deep learning engine Carry out training and learning, and establish a prediction model between image blocks and classification results; (2) Sample prediction: divide all images into image blocks of a specified size as required, and for ea...

AI Technical Summary

Problems solved by technology

The traditional image deep learning method is based entirely on image blocks for deep learning classification. Due to the limitation of insufficient image information, there is a bottleneck that the target scale is difficult to define: the boundary of the classification result is inaccurate, the result is unstable, and the generalization ability of the model is poor. The results of learning classification depend to a certain extent on the selection of training samples, and it is difficult to promote and apply across regions.

For example: the goal is to extract the building area, the result may be extracted as a residential area, or as a part of the house, resulting in a large difference between the classification result and the demand. Even if the building is partially extracted, the boundary becomes uneven and jagged due to trees and other occlusions. Meet actual needs

Images