Figure 5 - uploaded by Keith Copsey
Content may be subject to copyright.
6: The training data classification rate versus the model order (number of mixture components for each class).
Source publication
Similar publications
Bayesian hierarchical models are increasing popular in economics. When using hierarchical models, it is useful not only to calculate posterior expectations, but also to measure the robustness of these expectations to reasonable alternative prior choices. We use variational Bayes and linear response methods to provide fast, accurate posterior means...
Citations
... Thus, training sets trained with these data can be used for target detection in real SAR images. In this study, for the first time, ATR were tried to be carried out from real SAR images by using groundbased ISAR images [4], [7], [8][3] and a database trained with popular artificial intelligence algorithms such as Artificial Neural Networks (ANN) [9], k-nearest neighbor (KNN) [10] and Support Vector Machine (SVM) [4]. The features of 2D SAR and ISAR images were extracted with Cooccurrence Matrix (GLCM), and Gray Level Size Zone Matrix (GLSZM) methods, and the performances of artificial intelligence algorithms and feature extraction techniques were tested. ...
... The first target consists of ISAR images of the T72 tank presented by MSTAR. While the second target is the backhoe loader given in MSTAR data, the third target is the ISAR images of the main battle tank obtained with a depression angle of 10 degrees presented by [7]. In the training data, the ISAR images of the 2nd and 3rd targets were reproduced automatically by rotation, while the images of the 1st target were taken directly from the MSTAR data. ...
Automatic target recognition (ATR) from SAR images is critical, especially for military applications. Although many machine learning-based ATR methods are effectively used for this purpose, training and test data in all studies so far have been made with the same SAR images. However, obtaining real SAR images of different targets is a costly and time consuming process. In this study, for the first time, ATR in SAR images was performed using training data consisting of images obtained with ISAR imaging. As training data, ISAR images of 3 different targets were obtained and tested with 3 different SAR image sets in the MSTAR dataset. The data were trained with KNN and SVM by applying two different feature extraction algorithms, GLSZM and GLCM. From the results obtained, it has been revealed that the GLCM+SVM algorithm is the best model with 81,27% accuracy. Although these results are not at the desired level, the performance of the most realistic test and training method for military applications has been demonstrated.
... It has been assumed in the above discussion that the variance of the thermal noise is known. If desired, it is possible to include the noise variance as an additional unknown parameter to be estimated as part of the Bayesian framework [40]. On the other hand, a heuristic method for estimating the noise variance, such as that suggested in [29], can be used. ...
... It has been assumed in the above discussion that the variance of the thermal noise is known. If desired, it is possible to include the noise variance as an additional unknown parameter to be estimated as part of the Bayesian framework [40]. On the other hand, a heuristic method for estimating the noise variance, such as that suggested in [29], can be used. ...
... It has been assumed in the above discussion that the variance of the thermal noise is known. If desired, it is possible to include the noise variance as an additional unknown parameter to be estimated as part of the Bayesian framework [39]. Alternatively a heuristic method for estimating the noise variance, such as that suggested in [28], could be used. ...
Super-resolution of signals and images can improve the automatic detection and recognition of objects of interest. However, the uncertainty associated with this process is not often taken into consideration. This is important because the processing of noisy signals can result in spurious estimates of the scene content. This study reviews a variety of super-resolution techniques and presents two non-parametric Bayesian super-resolution algorithms that not only take uncertainty into account, but also retain knowledge about the output uncertainty in the form of a full probability distribution. One of the two Bayesian techniques is based on an analytical calculation re-interpreted as super-resolution, and the other is a novel numerical algorithm. Although the algorithms are presented as stand-alone techniques for image analysis, such Bayesian super-resolution algorithms can increase automatic target recognition performance over standard super-resolution.
... If computers could reliably perform this task automatically then potentially thousands of images per hour could be processed. The subject of automatic target recognition (ATR) has been developing significantly since the 1960s [25]. Research programmes have been accelerated with the advent of high-speed low-cost computers and an increasing military need for reliable ATR. ...
... Here we outline only the Bayesian classifier and k-nearest-neighbour algorithm as those will be used later in this thesis when comparing super-resolution algorithms in a Bayesian framework. For a recent review of other target recognition algorithms see [25]. ...
... Generalization could potentially be improved by increasing the value of k although it is suggested in [83] that this is unnecessary. The issue of generalization is covered in detail in [25] and [83]. ...
Processing of synthetic aperture radar (SAR) images for automatic target recognition (ATR) is a critical application especially in military surveillance. In particular, numerous machine learning-based SAR ATR methods have been proposed for this task. However, data training and testing stages of all these methods are based on the exploitation of SAR signatures of the target under investigation. Considering the high variability of radar targets, obtaining such signature data is obviously a costly and time consuming process. In this study, therefore, a feasibility analysis of the use of inverse-SAR (ISAR) training data in SAR ATR has been made for the first time. The turntable ISAR and circular SAR images of three different vehicles are used in training and testing is performed by means of SAR images of three similar targets within the publicly available MSTAR dataset. Also, three most prominent machine learning methods, namely KNN, SVM and ANN are used in conjunction with three different feature extraction algorithms namely, GLRLM, GLSZM and GLCM. The obtained results reveal that the GLCM+SVM algorithm pair is the most effective model with 85% accuracy.
