Polarized SAR Image Classification Method Based on Wavelet Sparse Autoencoder

Abstract

The invention discloses a polarization SAR image classification method based on a wavelet sparse self-encoder, which mainly solves the problem of the classification accuracy decline caused by the irrelevance and redundancy of extracted feature data and unreasonable feature extraction. Its main steps are: (1), input image; (2), preprocessing; (3), extracting image features; (4), selecting training samples and test samples; (5), training wavelet sparse autoencoder; ( 6), training softmax classifier; (7), adjusting network parameters; (8), image classification; (9), coloring; (10), output classification result map. The invention reduces the time complexity, reflects the essential characteristics of data, can better learn higher-dimensional features from low-dimensional features, has good denoising effect, and improves the classification accuracy of images.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a polarization synthetic aperture radar SAR (Synthetic Aperture Radar, SAR) image classification method based on a wavelet sparse autoencoder in the technical field of polarization synthetic aperture radar image classification. The invention classifies polarization synthetic aperture radar SAR images by combining a generation wavelet function and a sparse self-encoder, and can be used for polarization synthetic aperture radar SAR image target detection and target recognition. Background technique [0002] Polarization SAR has become one of the important development directions of SAR at home and abroad, and polarization SAR image classification is an important research technology of SAR image interpretation. Polarimetric SAR can describe the target more comprehensively, and its measurement data contains rich target information. Therefore, polarimetric SAR has very o...

AI Technical Summary

Problems solved by technology

The calculation complexity of this method is relatively small, and the accuracy has been improved compared with the classical method. However, the shortcomings of this method are: this method belongs to unsupervised classification, without preprocessing, and can only rely on scattering information Clustering ground objects does not learn the deep features and detailed features of the data, which makes the classification accuracy of polarimetric SAR low

The shortcomings of this method are: the LLSVM classifier cannot guarantee that the obtained solution is the global optimal solution, and it lacks sparsity, which easily leads to overfitting, and cannot overcome the influence of outliers and noise, making the classification accuracy of polarimetric SAR low

Images