基于多层自编码器的SAR目标和阴影共享表示

Q2 Physics and Astronomy

雷达学报 Pub Date : 2013-04-01 DOI:10.3724/SP.J.1300.2013.20085

Zhi-jun Sun, Lei Xue, Yang-ming Xu, Zhijun Sun

{"title":"基于多层自编码器的SAR目标和阴影共享表示","authors":"Zhi-jun Sun, Lei Xue, Yang-ming Xu, Zhijun Sun","doi":"10.3724/SP.J.1300.2013.20085","DOIUrl":null,"url":null,"abstract":"Automatic Target Recognition (ATR) of Synthetic Aperture Radar (SAR) images is investigated. A SAR feature extraction algorithm based on a multilayer auto-encoder is proposed. The method makes use of a probabilistic neural network and Restricted Boltzmann Machine (RBM) modeling probability distribution of the environment. Through the formation of a more expressive multilayer neural network, the deep learning model learns the shared representation of the target and its shadow outline reflecting the target shape characteristics. Targets are classified automatically through two recognition models. The experiment results based on the MSTAR verify the effectiveness of the proposed algorithm.","PeriodicalId":37701,"journal":{"name":"雷达学报","volume":"1 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2013-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":"{\"title\":\"Shared Representation of SAR Target and Shadow Based on Multilayer Auto-encoder\",\"authors\":\"Zhi-jun Sun, Lei Xue, Yang-ming Xu, Zhijun Sun\",\"doi\":\"10.3724/SP.J.1300.2013.20085\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Automatic Target Recognition (ATR) of Synthetic Aperture Radar (SAR) images is investigated. A SAR feature extraction algorithm based on a multilayer auto-encoder is proposed. The method makes use of a probabilistic neural network and Restricted Boltzmann Machine (RBM) modeling probability distribution of the environment. Through the formation of a more expressive multilayer neural network, the deep learning model learns the shared representation of the target and its shadow outline reflecting the target shape characteristics. Targets are classified automatically through two recognition models. The experiment results based on the MSTAR verify the effectiveness of the proposed algorithm.\",\"PeriodicalId\":37701,\"journal\":{\"name\":\"雷达学报\",\"volume\":\"1 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2013-04-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"1\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"雷达学报\",\"FirstCategoryId\":\"1093\",\"ListUrlMain\":\"https://doi.org/10.3724/SP.J.1300.2013.20085\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"Physics and Astronomy\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"雷达学报","FirstCategoryId":"1093","ListUrlMain":"https://doi.org/10.3724/SP.J.1300.2013.20085","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"Physics and Astronomy","Score":null,"Total":0}

引用次数: 1

摘要

研究了合成孔径雷达(SAR)图像的自动目标识别(ATR)问题。提出了一种基于多层自编码器的SAR特征提取算法。该方法利用概率神经网络和受限玻尔兹曼机(RBM)对环境的概率分布进行建模。深度学习模型通过形成更具表现力的多层神经网络，学习目标的共享表示及其反映目标形状特征的阴影轮廓。通过两种识别模型对目标进行自动分类。基于MSTAR的实验结果验证了该算法的有效性。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文本刊更多论文

Shared Representation of SAR Target and Shadow Based on Multilayer Auto-encoder

Automatic Target Recognition (ATR) of Synthetic Aperture Radar (SAR) images is investigated. A SAR feature extraction algorithm based on a multilayer auto-encoder is proposed. The method makes use of a probabilistic neural network and Restricted Boltzmann Machine (RBM) modeling probability distribution of the environment. Through the formation of a more expressive multilayer neural network, the deep learning model learns the shared representation of the target and its shadow outline reflecting the target shape characteristics. Targets are classified automatically through two recognition models. The experiment results based on the MSTAR verify the effectiveness of the proposed algorithm.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

雷达学报 Physics and Astronomy-Instrumentation

CiteScore

4.10

自引率

0.00%

发文量

882

期刊介绍： Journal of Radars was founded in 2012 by the Institute of Space and Astronautical Information Innovation of the Chinese Academy of Sciences (formerly the Institute of Electronics) and the China Radar Industry Association (CRIA), which is located in the high-end academic journal and academic exchange platform in the field of radar, and is committed to promoting and leading the scientific and technological development in the field of radar. The journal can publish Chinese papers and English papers, and is now a bimonthly journal. Journal of Radars focuses on theory, originality and foresight, and its scope of coverage mainly includes: radar theory and system, radar signal and data processing technology, radar imaging technology, radar identification and application technology. Journal of Radars has been included in domestic core journals and foreign Scopus, Ei and other databases, and was selected as ‘China's high-quality science and technology journals’, and ranked the first in the category of electronic technology and communication technology in the ‘Chinese Core Journals List (2023 Edition)’.