SAR image object classification method based on SDAE-SVM

Abstract

The invention discloses an SAR image object classification method on SDAE-SVM. The method mainly settles the problems of low characteristic robustness of the SAR image extracted from a common stack autoencoder and low classification accuracy of a softmax classifier fine-tuning deep network in an existing method. The SAR image object classification method comprises the steps of (1), reading in data; (2), performing dividing operation; (3), performing rotation operation; (4), determining an objective slice of the SAR image; (5), vectoring the objective slice; (6), constructing a four-layer initial stack type de-noising autoencoder (SDAE); (7), training the initial stack type de-noising autoencoder (SDAE); (8), performing fine-tuning; and (9), calculating testing accuracy. The SAR image object classification method has advantages of realizing high object characteristic robustness of the extracted SAR image and improving object classification accuracy of the SAR image.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a synthetic aperture radar SAR (Synthetic Aperture Radar) based on a stack denoising autoencoder and a support vector machine SDAE-SVM (Stacked Denoising Autoencoder Support Vector Machine) in the technical field of image classification Radar) image object classification method. The invention can be used for target classification and recognition in SAR images. Background technique [0002] Synthetic Aperture Radar (SAR) is a high-resolution imaging radar. As an imaging system, it has the advantage of not being affected by weather and light, and can monitor target scenes all day and all weather. Therefore, the target classification of SAR images has positive significance for both military and civilian applications. With the development and application of synthetic aperture radar SAR system and the improvement of synthetic aperture radar SAR data transmission level...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the feature extraction process is complicated, and the extracted features lack robustness, and the classification accuracy is low.

The disadvantage of this method is that the traditional stacked autoencoder and softmax classifier model lead to low classification accuracy

Images