A Method of SAR Image Automatic Target Recognition Based on Convolution Auto-Encode and Support Vector Machine

Abstract

In this paper, a method of Synthetic Aperture Radar (SAR) image Automatic Target Recognition (ATR) based on Convolution Auto-encode (CAE) and Support Vector Machine (SVM) is proposed. Using SVM replaces the traditional softmax as the classifier of the CAE model to classify the feature vectors extracted by the CAE model, which solves the problem that the softmax classifier is less effective in the nonlinear case. Since the SVM can only solve the binary classification problem, and in order to realize the classification of the class objectives, the SVM were designed to achieve the classification of the input samples. After unsupervised training for CAE, the coding layer is connected with SVM to form a classification network. CAE can extract the features of the data by an unsupervised method, and the nonlinear classification advantage of SVM can classify the features extracted by CAE and improve the accuracy of the object recognition. At the same time, the high-accuracy identification of key targets is required in some special cases. A new initialization method is proposed, which initializes the network parameters by pretraining the key targets and changes the weights of different targets in the loss function to obtain better feature extraction, so it can ensure good multitarget recognition ability while realizing the high recognition accuracy of the key targets.

CASVM: An Efficient Deep Learning Image Classification Method Combined with SVM

“…However, using such techniques can result in drawbacks like the inability to do multi-classification or issues like low multi-classification accuracy [4]. Researchers have discovered one-to-all or one-to-one [44][45][46][47] SVM classification approaches to address the issue of limiting multiclassification and have concentrated their research on feature extraction models paired with SVM to increase multi-classification accuracy [48].…”

Synthetic aperture radar target recognition based on adaptive decision fusion of multiple views

“…Most of the classifiers used in SAR ATR were mature ones, such as K-nearest neighbor, 11 SVM, 9,25,26 and sparse representation-based classification (SRC). [26][27][28] With emerging deep learning techniques, intelligent models [29][30][31][32][33][34] have been applied to SAR ATR including convolutional neural networks (CNN), [30][31][32] ResNet, 33 and convolution autoencode. 34 Those methods conducted target recognition through end-to-end frameworks, in which features and classifiers were considered together.…”

“…Most of the classifiers used in SAR ATR were mature ones, such as K-nearest neighbor, 11 SVM, 9 , 25 , 26 and sparse representation-based classification (SRC) 26 – 28 With emerging deep learning techniques, intelligent models 29 – 34 have been applied to SAR ATR including convolutional neural networks (CNN), 30 – 32 ResNet, 33 and convolution autoencode 34 . Those methods conducted target recognition through end-to-end frameworks, in which features and classifiers were considered together.…”

A Moving Target Tracking Framework Based on a Set and Its Topological Space

“…In addition, some scholars introduced feature vectors to achieve high-precision target recognition. For example, a synthetic aperture radar (SAR) image automatic target recognition method based on convolutional self-coding (CAE) and support vector machine (SVM) [14] realizes high recognition accuracy for key targets by pretraining and initializing network parameters for key targets. Moreover, deep learning [15][16][17][18][19][20] and machine learning have obvious advantages in improving the accuracy of target tracking.…”