Radar HRRP target category labeling method based on convolutional self-encoding

Abstract

The invention provides a radar HRRP target category labeling method based on convolutional self-coding, and aims to solve the problems of low manual interpretation efficiency and low accuracy of massive HRRP samples. The method is mainly divided into three stages: stage 1, constructing convolution self-coding, and training the convolution self-coding to convergence by using all HRRP samples; 2, constructing a convolutional neural network by taking a convolutional self-encoding encoder as a feature extractor, and training the convolutional neural network by utilizing a labeled HRRP sample to obtain an initial labeling model; 3, labeling the label-free HRRP sample of which the one-hot coding meets the labeling condition, performing parameter updating on the labeling model by using the label-free HRRP sample, and repeating the step 3 until the HRRP sample meeting the condition is not increased any more to obtain a final labeling model. Compared with a traditional labeling method, the method has the advantages that label-free HRRP sample information is fully utilized, the labeling efficiency and accuracy are greatly improved, and the labeling problem of massive HRRP samples is solved.

Description

technical field [0001] The invention belongs to radar target automatic recognition technology, and aims at the problems of low efficiency and poor accuracy of manual interpretation of massive sample data, and proposes a radar HRRP target category labeling method based on convolutional self-encoding. Background technique [0002] HRRP (High Resolution Range Profiles) contains a large amount of target structure, target scattering point intensity and other information to build a target HRRP sample library. In actual situations, most of the targets to be identified are non-cooperative targets, and their HRRP labels require manual interpretation by professionals. In the face of massive data, the efficiency of manual labeling is low and the accuracy rate is poor. In view of the above situation, the present invention proposes a radar HRRP target category labeling method based on convolutional self-encoding. Compared with traditional labeling methods, this method makes full use of ...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide a radar HRRP target category labeling method based on convolutional self-encoding, which greatly improves the accuracy of sample labeling, in view of the problems of low manual labeling efficiency and poor accuracy during the establishment of the radar target sample database. efficiency and accuracy

Images