A Remote Sensing Image Classification Method Based on Deep Contourlet Network of Attention Mechanism

Abstract

The invention discloses a remote sensing image classification method based on attention mechanism depth Contourlet network, establishes a remote sensing image library, obtains a training sample set and a test sample set; The two-by-two stacked layer constitutes a convolution module, and the attention mechanism is used to enhance the data of the merged feature map through the channel attention module; perform iterative training; perform global contrast normalization processing on the remote sensing images to be classified, Obtain the average intensity of the entire remote sensing image, and then perform normalization to obtain the remote sensing image to be classified after normalization processing; input the normalized unknown remote sensing image into the trained convolutional neural network model, The unknown remote sensing pictures are classified, and the network output classification results are obtained. The present invention combines the method of Contourlet decomposition and deep convolutional network, introduces the channel attention mechanism, and can simultaneously play the advantages of deep learning and Contourlet transformation.

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 based on an attention mechanism deep Contourlet network. Background technique [0002] Remote sensing image scene classification is the basic task of remote sensing image understanding, which can be applied to military and civilian fields. With the development of remote sensing technology, the quality of the captured remote sensing images has also been improved. The features contained in the images are more detailed and the spectral features are more complex. This has led to many early remote sensing image recognition methods in the current high-quality The accuracy of the image classification task is not as good as expected. [0003] In recent years, deep learning theory has developed rapidly in the field of image processing, and its performance has been superior to traditional image classification algorithms. Ma...

AI Technical Summary

Problems solved by technology

However, the deep learning model currently applied to remote sensing image classification has many limitations, ignoring the spectral information of the image, but directly allowing the model to fit parameters from the input pixel-level image, and the scattered search space also brings a large number of unknown parameters. and training speed limitations

Images