High-level feature fusion classification method for multi-modal remote sensing images based on deep learning

Abstract

The invention relates to a method for fusion and classification of high-level features of multi-modal remote sensing images based on deep learning, which comprises the following steps: preprocessing multi-modal remote sensing data sets, making training sets, verification sets, and test sets for high-level feature extractors, It is used for the construction of high-level feature extractor model parameters; the model construction of multi-modal high-level feature extractor is used to extract high-level features of multiple modal remote sensing images; the extraction and storage of high-level features of multi-modal remote sensing images is used for multi-modal Classifier training, verification and testing; high-level feature fusion algorithm, used to make training set, verification set and test set of classifier; construction of classifier model parameters, used for classification of newly generated data sets. The present invention can achieve remarkable results in the classification of remote sensing images and ground features through deep learning, and is suitable for complex and massive remote sensing image classifications. use value.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to the field of deep learning, and in particular to a method for fusion and classification of high-level features of multi-modal remote sensing images based on deep learning. Background technique [0002] With the development of remote sensing technology, it is of great research value to efficiently apply remote sensing technology to regional land surveys, quantitatively extract land use information, and monitor and evaluate the environment. Among them, remote sensing image classification technology, as an important support for remote sensing applications, has been widely used in thematic information extraction, change detection, thematic map production and the establishment of remote sensing databases. [0003] Remote sensing image classification is a key technology in remote sensing application systems. Its main task is to interpret and identify the attributes and distribution of g...

AI Technical Summary

Problems solved by technology

When dealing with multimodal remote sensing images from multiple sensors, different resolutions, and different physical representations, although machine learning has been widely used in remote sensing image classification, there are still four problems in this type of method: (1) The choice of multiple modalities is limited; (2) The spectrum, dimension, time phase, resolution and other characteristics of remote sensing images make data processing a severe challenge; (3) How to establish the interaction between multiple modalities It is difficult to improve the classification accuracy of remote sensing images; (4) With the development of remote sensing technology, massive remote sensing image processing is facing challenges

[0010] However, there is no method that can effectively overcome the four problems in the application of machine learning to remote sensing image classification as mentioned above, and can quickly, efficiently and safely classify complex and massive remote sensing data.

Images