SAR target identification method combining few-sample learning and target attribute features

Abstract

The invention provides an SAR target identification method combining few-sample learning and target attribute characteristics. The SAR target identification method comprises the following steps: acquiring a source domain data set R, a target domain data set E, a target domain support set ES, a target domain query set EQ and a target attribute characteristic set A of R and ES; constructing a visualclassification network F1; performing iterative training on the visual classification network F1 by using R; constructing a visual attribute classification network F; performing iterative training onthe visual attribute classification network F by using the source domain data set R, the target domain support set ES and the target attribute feature set A thereof; and obtaining a target recognition result of the SAR image by using the target domain support set ES, the target domain query set EQ and the target attribute feature set A'o thereof. According to the SAR target identification method,the SAR target recognition performance under the condition that training samples of known categories are limited is improved by combining few-sample learning and target attribute characteristics.

Description

technical field [0001] The invention belongs to the technical field of radar image processing, and relates to a SAR image target recognition method, in particular to a SAR target recognition method combining few-sample learning and target attribute features, which can be used for target recognition when known category label samples are limited. Background technique [0002] Synthetic aperture radar (SAR) has the characteristics of all-day, all-weather and strong penetrating power, and is widely used in the fields of reconnaissance, detection guidance and remote sensing. In recent years, SAR image automatic target recognition technology SAR ATR has developed rapidly. The basic SAR image automatic target recognition system generally includes three stages: target detection, target identification and target recognition. Target recognition is used to distinguish the categories of targets, as the last link of automatic target recognition system, it has important research significa...

AI Technical Summary

Problems solved by technology

This method improves the accuracy of SAR target recognition in the case of limited training data in the target task, but its disadvantage is that when training the angle rotation generation network ARGN, only the source task sample pair is used to extract the angle of the feature The rotation generation network ARGN has been trained, and the training set data of the target task is not used for training, so the trained angle rotation generation network ARGN is used to extract the features of the training set and test set of the target task, and the training set and the test set of the target task are obtained. The separability of the features in the test set is not high enough, and only the features extracted by the network are used as the basis for classification, so the accuracy of SAR image target recognition is still not high enough

For example, Tang, Zhang et al. disclosed a SAR target recognition method based on few-shot learning and improved Siamese network at the 2019 International Symposium on Earth Science and Remote Sensing, using a large number of known categories of source domain data to train a classifier Model, using the trained classification model to perform target recognition on the target domain data with only a small number of known category labels, this method improves the accuracy of SAR target recognition under the condition of limited known category labels in the target domain data, but its existence The disadvantage is that only the features extracted by the network are used as the basis for classification, so the accuracy of SAR image target recognition is still not high enough

Images