基于全卷积神经网络的SAR图像目标分类

近年来，卷积神经网络(Convolutional Neural Network，CNN)在合成孔径雷达(Synthetic Aperture Radar，SAR)图像目标分类中取得了较好的分类结果。CNN结构中，前面若干层由交替的卷积层、池化层堆叠而成，后面若干层为全连接层。全卷积神经网络（All Convolutional Neural Network, A-CNN）是对CNN结构的一种改进，其中池化层和全连接层都用卷积层代替，该结构已在计算机视觉领域被应用。针对公布的MSTAR数据集，提出了基于A-CNN的SAR图像目标分类方法，并与基于CNN的SAR图像分类方法进行对比。实验结果表明，基于A-CNN的SAR图像目标分类正确率要高于基于CNN的分类正确率。

SAR Image Target Classification Based on All Convolutional Neural Network

Convolutional neural network (CNN) has achieved good performance in the target classification of synthetic aperture radar (SAR) images. In the CNN structure, the preceding layers are stacked by alternating convolutional and pooling layers, while the latter layers are the fully-connected layers. The all convolutional neural network (A-CNN) is an improvement on the CNN structure, where the pooling and fully-connected layers are replaced by convolutional layers. It has been applied in the field of computer vision. In this paper, a method of SAR image target classification based on A-CNN is proposed. A comparison of the proposed method with CNN-based method by use of the published MSTAR data set. The experimental results show that the accuracy of SAR image classification based on A-CNN is higher than that of CNN.