摘要
针对合成孔径雷达图像的分类优化方法,提出一种基于多特征与卷积神经网络的SAR图像分类方法Canny-WTD-CNN.将Canny算子提取的边缘特征,与小波阈值去噪法提取的小波特征进行自适应融合,作为卷积神经网络的输入;以softmax为分类器,对SAR图像进行分类识别检测.最后利用MSTAR公开数据集的三类目标数据进行试验,并给出该方法与其他方法结果的对比,表明该方法的有效性,识别率达到99.14%.
A new method of optimization Synthetic Aperture Radar(SAR)image classification based on multi-feature and Convolutional Neural Networks,Canny-WTD-CNN,was proposed.The edge features extracted by Canny operator was adaptively fused with the wavelet features extracted by wavelet threshold denoising method,as the input of Convolutional Neural Networks,and softmax was used as the classifier to classify and identify SAR images.Finally,this paper used the three types of target data of MSTAR database to test,and the comparison between the method and other methods was given.The effectiveness of the method was shown,and the recognition rate reached 99.14%.
作者
王永乐
续婷
杜敦伟
白艳萍
WANG Yong-le;XU Ting;DU Dun-wei;BAI Yan-ping(School of Science,North University of China,Taiyuan 030051,China;Beijing Institute of Mechanical and Electrical Engineering,Beijing 100074,China)
出处
《数学的实践与认识》
北大核心
2020年第6期140-147,共8页
Mathematics in Practice and Theory
基金
国家自然科学基金(61774137)
山西省自然科学基金(201801D121026,201701D121012,201701D221121)
山西省回国留学人员科研项目(2016-088)
关键词
SAR图像
卷积神经网络
小波变换
边缘提取
图像分类
SAR images
convolutional neural networks
wavelet transform
edge extraction
image classification
