现代雷达2018，Vol.40Issue(3)：38-42,5.DOI:10.16592/j.cnki.1004-7859.2018.03.009
基于卷积神经网络迁移学习的SAR图像目标分类
SAR Image Target Classification Based on Convolutional Neural Network Transfer Learning
刘晨 1曲长文 1周强 2李智 1李健伟1
作者信息
- 1. 海军航空工程学院电子信息工程系,山东烟台264001
- 2. 海军航空工程学院科技部,山东烟台264001
- 折叠
摘要
Abstract
In order to improve the convolutional neural network classification results for small sample target set,the method based on convolutional neural network transfer learning was proposed.Firstly,pre-training the parameter of convolutional neural network by large sample data set.Secondly,replace fully connected layer with the extreme learning machine in convolutional neural,complete the transfer learning of convolutional neural network.Finally,training the network structure parameters of the extreme learning machine by the small sample data set,and obtained classification model.The experimental results show that the method has improved the accuracy of image classification and reduced training time,outperforming other methods on MSTAR and SAR ship target data sets.关键词
迁移学习/卷积神经网络/超限学习机/合成孔径雷达图像Key words
transfer learning/convolutional neural network/extreme learning machine/SAR image data分类
信息技术与安全科学引用本文复制引用
刘晨,曲长文,周强,李智,李健伟..基于卷积神经网络迁移学习的SAR图像目标分类[J].现代雷达,2018,40(3):38-42,5.
