A Novel Convolutional Neural Network Architecture for SAR Target Recognition

Abstract

Among many improved convolutional neural network (CNN) architectures in the optical image classification, only a few were applied in synthetic aperture radar (SAR) automatic target recognition (ATR). One main reason is that direct transfer of these advanced architectures for the optical images to the SAR images easily yields overfitting due to its limited data set and less features relative to the optical images. Thus, based on the characteristics of the SAR image, we proposed a novel deep convolutional neural network architecture named umbrella. Its framework consists of two alternate CNN-layer blocks. One block is a fusion of six 3-layer paths, which is used to extract diverse level features from different convolution layers. The other block is composed of convolution layers and pooling layers are mainly utilized to reduce dimensions and extract hierarchical feature information. The combination of the two blocks could extract rich features from different spatial scale and simultaneously alleviate overfitting. The performance of the umbrella model was validated by the Moving and Stationary Target Acquisition and Recognition (MSTAR) benchmark data set. This architecture could achieve higher than 99% accuracy for the classification of 10-class targets and higher than 96% accuracy for the classification of 8 variants of the T72 tank, even in the case of diverse positions located by targets. The accuracy of our umbrella is superior to the current networks applied in the classification of MSTAR. The result shows that the umbrella architecture possesses a very robust generalization capability and will be potential for SAR-ART.

Near-Infrared Road-Marking Detection Based on a Modified Faster Regional Convolutional Neural Network

“…After an image passes through each stage, its feature map shrinks by half. After several convolutions and pooling operations, coarse feature maps are obtained at the last convolution stage [13,48].…”

“…At each sliding window position, there are nine (9) anchors, which translate through a feature map from which varying sizes of proposals are obtained [9]. The RoIs are sent into two siblings 1 × 1 convolutional layers for the box-regression and box-classification task [48]. The maximum possible proposal for each location is denoted by k. The classification (cls) layer is assigned 2k scores (there is an object, or there is no object) for each proposal.…”

End‐to‐End Ship Detection in SAR Images for Complex Scenes Based on Deep CNNs

“…SAR images are different from optical images, and the microwave imaging mechanism is more complicated. The artificial recognition of SAR image objects is difficult and requires strong domain knowledge, hence, the automatic target recognition (ATR) system of SAR images is necessary [4]. SAR ocean images are heterogeneous and contain the ships, upwelling, breaking waves, and a lot of artifacts such as radio frequency interferences (RFIs) and azimuth ambiguities [5].…”

Target Recognition of Synthetic Aperture Radar Images Based on Two-Phase Sparse Representation

“…Based on the extracted features, the classifiers are designed to make decisions on the target labels. The nearest neighbor (NN) [13,27], support vector machine (SVM) [28,29], sparse representation-based classification (SRC) [29][30][31], adaptive boosting [32], and the recent deep learning classifiers (e.g., convolutional neural network (CNN) [33][34][35][36][37]) are popular classification schemes in SAR ATR.…”