Combining deep convolutional neural network and SVM to SAR image target recognition

Citation

Gao F, Huang T, Wang J, Sun J, Yang E & Hussain A (2018) Combining deep convolutional neural network and SVM to SAR image target recognition. In: 2017 IEEE International Conference on Internet of Things (iThings) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data (SmartData), volume 2018-January. The 3rd International Conference on Smart Data (SmartData-2017), 21.06.2017-23.06.2017. Exeter, UK: Institute of Electrical and Electronic Engineers, pp. 1082-1085. https://doi.org/10.1109/iThings-GreenCom-CPSCom-SmartData.2017.165

Abstract
To address the challenging problem on target recognition from synthetic aperture radar (SAR) images, a novel method is proposed by combining Deep Convolutional Neural Network (DCNN) and Support Vector Machine (SVM). First, an improved DCNN is employed to learn the features of SAR images. Then, a SVM is utilized to map the leant features into the output labels. To enhance the feature extraction capability of DCNN, a class of separation information is also added to the cross-entropy cost function as a regularization term. As a result, this explicitly facilitates the intra-class compactness and separability in the process of feature learning. Numerical experiments are performed on the Moving and Stationary Target Acquisition and Recognition (MSTAR) database. The results demonstrate that the proposed method can achieve an average accuracy of 99.15% on ten types of targets.

Keywords
Support vector machines; Cost function; Target recognition; Synthetic aperture radar; Training; Databases; Feature extraction