Multi-source remote sensing image classification method based on two-way attention fusion neural network

Abstract

The invention discloses a multi-source remote sensing image classification method based on a two-way attention fusion neural network, and mainly solves the problem of low classification precision of multi-source remote sensing images in the prior art. The implementation scheme comprises the following steps: 1) preprocessing and dividing hyperspectral data and laser radar data to obtain a trainingsample and a test sample; 2) designing an attention fusion layer based on an attention mechanism to carry out weighted screening and fusion on spectral data and laser radar data, andestablishing a two-way interconnection convolutional neural network, (3) training the interconnection convolutional neural network by taking multiple types of cross entropies as a loss function to obtain a trained network model, and (4) predicting a test sample by using the trained model to obtain a final classification result. The method can extract the features of the multi-source remote sensing data and effectively fuse and classify the features, improves the problem of overhigh dimension in fusion, improves the average classification precision, and can be used for fusing remote sensing images obtained by two different sensors.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a remote sensing image classification method, which can be used to fuse and classify remote sensing images obtained by two different sensors. Background technique [0002] In recent years, the number of remote sensing sensors has grown rapidly, and people can simultaneously obtain multi-source datasets of the same scene, which makes it possible to integrate different information captured by different sensors. For example, multispectral image MSI or hyperspectral image HSI usually consists of multiple spectral channels of the same scene, containing detailed spectral and spatial information, providing the ability to accurately distinguish materials of interest. On the other hand, lidar LiDAR data can represent the elevation and target height information of the scene, which helps to distinguish objects made of similar materials but with different heights. There...

AI Technical Summary

Problems solved by technology

This type of method can take advantage of the advantages of convolutional neural networks to extract effective features from hyperspectral images and LiDAR images, and apply these features to classification problems. However, this model still has certain shortcomings when performing fusion classification. Advantages: firstly, the hyperspectral image network and LiDAR data network are separated from each other, and the circulation and fusion of feature information cannot be performed well; secondly, the features of the two images extracted by the convolutional neural network are only input to the The fully connected layer classifies, and does not filter and fuse features, which will lead to excessively high dimensionality of features and feature redundancy.

Images