Polarized SAR Classification Method Based on Deep Bidirectional LSTM Siamese Network

Abstract

The invention discloses a polarization SAR classification method based on a deep convolutional bidirectional LSTM twin network, which mainly solves the problem of low classification accuracy caused by fewer polarization SAR data label samples in the existing method. The implementation steps are: 1) input the polarimetric SAR image to be classified and its real object mark, and perform Lee filtering; 2) extract the time series feature vector from the filtered data and divide the training set and test set; 3) the training set 4) Build a deep convolutional bidirectional LSTM twin network and use the training set and sample pair training set to train it; 5) Use the trained network to classify the test set samples to obtain the ground truth object category. The present invention expands the training set and extracts differentiated features under the twin architecture, and makes more reasonable and full use of spatial neighborhood information for bidirectional time-series modeling under the condition of small-sample labeling, so that the classification accuracy of the model is greatly improved.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a polarization SAR data object classification method, specifically a polarization SAR classification method based on a deep convolution bidirectional LSTM twin network, which can be used for image classification, object classification and target recognition. Background technique [0002] Polarization SAR is a new type of radar technology that can perform full polarization measurement on the target. In the full polarization mode, HH, HV, VH can be obtained by alternately transmitting and simultaneously receiving two kinds of electromagnetic waves of horizontal polarization and vertical polarization. , VV and four different polarization scattering echo information form the polarization scattering matrix of the target. The polarization scattering matrix contains the complete electromagnetic scattering characteristics of the measured ground object, can fully express an...

AI Technical Summary

Problems solved by technology

Although this method makes full use of the target decomposition characteristics of polarimetric SAR data, the shortcomings of this method are that the manual design and extraction of features are complex, and the professional knowledge of polarimetric SAR is relatively high.

Although this method makes full use of the supervision information of labeled samples of polarimetric SAR data and improves the classification accuracy, this method only makes a simple linear connection for the spatial neighborhood information, and its utilization method is unreasonable, and it is easy to introduce data redundancy. residual and noise interference, which interferes with data classification

Images