Abstract: Inverse synthetic aperture radar (ISAR) images are often used for classifying and recognizing targets. Moreover, the use of fully polarimetric ISAR (Pol-ISAR) images enhances classification capabilities. In this paper, the authors propose a novel automatic target recognition (ATR) technique based on the use of fully Pol-ISAR images and neural networks (NNs). In order to reduce the amount of data processed by the classifier, the brightest scattering centers are first extracted by means of the Pol-CLEAN technique, and then, their scattering matrices are decomposed using Cameron's decomposition. A classifier based on the use of multilayer perceptron NN that makes use of the features extracted from the Pol-ISAR images is then implemented. A proof-of-concept test is performed on real data acquired during a controlled experiment in an anechoic chamber.

Abstract: The problem of feature selection is a difficult combinatorial task in machine learning and of high practical relevance. In this paper, we consider feature selection method for multimodally distributed data, and present a large margin feature weighting method for k-nearest neighbor (kNN) classifiers. The method learns the feature weighting factors by minimizing a cost function, which aims at separating different classes by large local margins and pulling closer together points from the same class, based on using as few features as possible. The consequent optimization problem can be efficiently solved by linear programming. Finally, the proposed approach is assessed through a series of experiments with UCI and microarray data sets, as well as a more specific and challenging task, namely, radar high-resolution range profiles (HRRP) automatic target recognition (ATR). The experimental results demonstrate the effectiveness of the proposed algorithms.

Abstract: A quantitative model analysis is presented to justify the extraction of high range resolution (HRR) profiles from synthetic aperture radar (SAR) images as motion-invariant features for identifying moving ground targets. A comparative study is conducted to assess the effectiveness in the identification process between using HRR profiles and SAR images as target signatures. The results indicate that HRR profiles are just as viable as SAR image for identification. Furthermore, a score-level multi-look fusion identification method has been investigated. It is found that a correct accurate identification rate of greater than 99 percent, a low false alarm rate, and a high level of identification confidence can be achieved, providing very robust performance.

Abstract: Simple yet robust techniques for detecting targets in infrared (IR) images are an important component of automatic target recognition (ATR) systems. In our previous works, we have developed IR target detection and tracking algorithms based on image correlation and intensity. In this paper, we discuss these algorithms, their performances and problems associated with them and then propose novel algorithms to alleviate these problems. Our proposed target detection and tracking algorithms are based on frequency domain correlation and Bayesian probabilistic techniques, respectively. The proposed algorithms are found to be suitable for real-time detection and tracking of static or moving targets, while accommodating for detrimental affects posed by the clutter and background noise. Finally, limitations of all these algorithms are discussed.

Abstract: The work presented here introduces a procedure for the automatic recognition of ground-based targets from high range resolution (HRR) profile sequences that may be obtained from a synthetic aperture radar (SAR) platform. The procedure incorporates an adaptive target mask and uses a superresolution algorithm to identify the cross-range positions of target scattering centers. These are used to generate a pseudoimage of the target whose low-order discrete cosine transform coefficients form the recognizer feature vector. Within the recognizer, the states of a hidden Markov model (HMM) are used to represent the target orientation and a Gaussian mixture model is used for the feature vector distribution. In a closed-set identification experiment, the misclassification rate for ten MSTAR targets was 2.8%. Also presented are results from open-set experiments and investigates the effect on recognizer performance of variations in feature vector dimension, azimuth aperture, and target variants.

Abstract: Automatic license plate recognition is an important form in the automatic target recognition. In recent years, there has a lot of research in license plate recognition, and many license plate recognition algorithms have been proposed and used. In this paper, a new license plate recognition approach is put forward based on the Radial Basis Function Neural Networks (RBFNN). Also discussed are the problem of feature of vehicle license plate feature, the input data pattern of the RBFNN, the architecture of the automatic recognition system, the problem of normalization of the image-size, and the problem of training algorithm of hidden layer’s neural nodes. Experiments have been conducted for video monitored by vehicle monitor. The results show that compared with BP neural network, the RBF neural network can decrease the error recognition rate, the complexity of the system architecture, the training time, and the recognition time efficiently.

Abstract: This paper presents a new supervised classification approach for automated target recognition (ATR) in SAS images. The recognition procedure starts with a novel segmentation stage based on the Hilbert transform. A number of geometrical features are then extracted and used to classify observed objects against a previously compiled database of target and non-target features. The proposed approach has been tested on a set of 1528 simulated images created by the NURC SIGMAS sonar model, achieving up to 95% classification accuracy.

Abstract: Hyperspectral-based automatic target recognition (ATR) and classification systems often project the high-dimensional hyperspectral reflectance signatures onto a lower dimensional subspace using techniques such as principal component analysis, Fisher's linear discriminant analysis (LDA), and stepwise LDA. In a general classification framework, these projections are suboptimal and, in the absence of sufficient training data, are likely to be ill conditioned. In recent work, the authors proposed a divide-and-conquer approach that partitions the hyperspectral space into contiguous subspaces followed by a multiclassifier and decision-fusion (MCDF) framework. Although this technique alleviated the small-sample-size problem and provided a good recognition performance in light and moderate pixel mixing, the performance significantly decreased under severe mixing conditions, as it does with conventional ATR techniques. In this letter, the authors propose a kernel discriminant analysis-based projection in each subspace of the partition, followed by the MCDF framework to ensure robust recognition even in severe pixel-mixing conditions. The performance of the proposed system (as measured by overall recognition accuracies) is greatly superior to conventional dimensionality-reduction techniques as well as the more recently proposed LDA-based MCDF technique.

Abstract: Aiming at the common support vector machine’s biased disadvantage and computational complexity, an unbiased least squares support vector machine (LSSVM) model is proposed in this paper. The model eliminates the bias item of LSSVM by improving the form of structure risk, then the unbiased least squares support vector classifier and the unbiased least squares support vector regression are deduced. Based on this model, we design a new learning algorithm using Cholesky factorization according to the characteristic of kernel function matrix, in this way the calculation of Lagrangian multipliers is greatly simplified. Several experiments on diffenert datasets are carried out, including the common datasets classification, synthetic aperture radar image automatic target recognition and chaotic time series prediction. The experimental results of correct recognition rate and the fitting precision testify that the unbiased LSSVM model has good universal ability and fitting accuracy, better generalization capability and stability, and have a great improvement in learning speed.

Abstract: In this paper, we present a system for aircraft automatic target recognition (ATR) using Inverse Synthetic Aperture Radar (ISAR) and based on Knowledge discovery from data process adapted to radar domain. We propose a method for target shape extraction from ISAR images based on the combination of two methods, SUSAN modified and active deformable contours via level set. In the second part of this work, we propose to fuse two commonly used shape descriptors algorithms based on moments Invariant and Fourier descriptors. We have investigated the impact of the information fusion on the recognition rate. The classification scheme is ensured using support vector machine (SVM) classifier. Several combination strategies are compared at score/decision/feature level. Experimental results of the proposed method are provided and discussed.

Abstract: This paper presents a new approach of improving automatic target recognition (ATR) performance by tuning adaptively the Gabor filter. The Gabor filter adopts the network structure of two layers, and its input layer constitutes the adaptive nonlinear feature extraction part, whereas the weights between output layer and input layer constitute the linear classifier. From the statistic property of high-resolution range profile (HRRP), its extracted nonstationarity degree of features is tracked to extract the discriminative features of Gabor atoms. Two experimental examples show that the Gabor filter approach with simple structure has higher recognition rate in radar target recognition from HRRP as compared with several existing methods.

Abstract: Light detection and ranging (Lidar) instruments collect high density and accurate three dimensional (3D) point clouds of scanned surfaces of objects 3D building modelling from terrestrial Lidar requires the raw point cloud data to be processed Through processing, noise and outliers are eliminated from the point cloud, and a 3D photo-realistic model is generated using image data This effectively reduces redundant data and enhances the visual representation This paper deals with point cloud processing and proposes methods to automate several of the processing procedures Specifically, we implemented automatic 3D point cloud registration, automatic target recognition used for geo-referencing, automatic plane detection algorithm used for surface modelling, and texture mapping The proposed approach leads to the generation of accurately geo-referenced three dimensional (3D) photo-realistic models from point clouds and digital imagery

Abstract: This study provides a comparative analysis of different infrared image enhancement techniques for sea surface targets. Image enhancement is the most important preprocessing step of automatic target recognition and tracking algorithms. For efficient target detection and tracking, it is important to develop infrared image enhancement techniques to increase the contrast between the target and background, emphasize edges, and suppress medium and sensor noises and the background clutter. In this paper, performances of different histogram modification, filtering, and morphological processing techniques are compared for sea surface targets' detection using both real and synthetic images. It is observed that balanced contrast limited histogram equalization and contrast enhancement technique provides best result for different scenarios in quantitative and subjective tests.

Abstract: In this paper, a novel descriptive feature parameter extraction method from synthetic aperture radar (SAR) images is proposed. The new approach is based on region covariance (RC) method which involves the computation of a covariance matrix whose entries are used in target detection and classification. In addition the region co-difference matrix is also introduced. Experimental results of object detection in MSTAR (moving and stationary target recognition) database are presented. The RC and region co-difference method delivers high detection accuracy and low false alarm rates. It is also experimentally observed that these methods produce better results than the commonly used principal component analysis (PCA) method when they are used with different distance metrics introduced.

Abstract: The paper addresses the problem of target recognition using High-resolution Radar Range Profiles (HRRP). A novel approach of feature extraction and dimension reduction based on extended high order central moments is proposed in order to reduce the dimension of range profiles. Features extracted from radar HRRPs are normalized and smoothed, and then comparative analysis of the similar approaches is done. The range profiles are obtained by step frequency technique using the two-dimensional backscatters distribution data of four different aircraft models. The template matching method by nearest neighbor rules, which is based on the theory of kernel methods for pattern analysis, is used to classify and identify the range profiles from four different aircrafts. Numerical simulation results show that the proposed approach can achieve good performance of stability, shift independence and higher recognition rate. It is helpful for real-time identification and the engineering implements of automatic target recognition using HRRP. The number of required templates could be reduced considerably while maintaining an equivalent recognition rate.

Abstract: Automatic target recognition(ATR) is an important task in image application. This paper concentrates on two key subroutines of ATR system: Dimensionality reduction and Classifier. After pretreatment on original features a self-organizing neural network trained with the Hebbian rule is used to extract the principal component features. Then a classifier based on Directed Acyclic Graph Support Vector Machines(DAGSVM) is adopted to recognize more than two types of aircraft targets. The experiment results show the proposed method achieves better subset features and higher recognition rate.

Abstract: The Target Separation Algorithms (TSAs) are used to improve the results of Automated Target Recognition (ATR). The task of the TSAs is to separate two or more closely spaced targets in Regions of Interest (ROIs), to separate targets from objects like trees, buildings, etc., in a ROI, or to separate targets from clutter and shadows. The outputs of the TSA separations are inputs to ATR, which identify the type of target based on a template database. TSA may include eight algorithms. These algorithms may use average signal magnitude, support vector machines, rotating lines, and topological grids for target separation in ROI. TSA algorithms can be applied together or separately in different combinations depending on case complexity, required accuracy, and time of computation.

Abstract: This paper deals with the processing adopted for shape extraction from the 2D-presentation (image) in radar field for target recognition The goal is to provide helpful in- formation to human operator for target recognition and data interpretation However, extracting the target characteristics from a radar echoes is the rather difficult task Hence, sev- eral kinds of radar signatures can be employed to acquire in- formation about target (15, 18) based on radar image This ap- plication aims to design a target recognition system based on measures of radar signatures more exactly based on two dimen- sional images (ISAR image - Inverse Synthetic Aperture radar) This paper presents one approach for retrieval system for tar- get recognition based on ISAR-images in radar experimenta- tion field Then, we propose efficient features that deal with target shape which are extracted using Watersheds transforma- tion Of course, the target shape gives a better human interpre-

Abstract: Visual tracking is one of the most important field of computer vision. It has immense number of applications ranging from surveillance to hi-fi military applications. This paper is based on the application developed for automatic visual tracking and fire control system for anti-aircraft machine gun (AAMG). Our system mainly consists of camera, as visual sensor; mounted on a 2D-moving platform attached with 2GHz embedded system through RS-232 and AAMG mounted on the same moving platform. Camera and AAMG are both bore-sighted. Correlation based template matching algorithm has been used for automatic visual tracking. This is the algorithm used in civilian and military automatic target recognition, surveillance and tracking systems. The algorithm does not give robust performance in different environments, especially in clutter and obscured background, during tracking. So, motion and prediction algorithms have been integrated with it to achieve robustness and better performance for real-time tracking. Visual tracking is also used to calculate lead angle, which is a vital component of such fire control systems. Lead is angular correction needed to compensate for the target motion during the time of flight of the projectile, to accurately hit the target. Although at present lead computation is not robust due to some limitation as lead calculation mostly relies on gunner intuition. Even then by the integrated implementation of lead angle with visual tracking and control algorithm for moving platform, we have been able to develop a system which detects tracks and destroys the target of interest.

Abstract: Validation of automatic target recognition (ATR) algorithm needs huge amount of real data, which is mostly infeasible. Hence we need statistical separability indices (SI) to evaluate the performance of ATR algorithms using limited amount of data. In this paper we explain five such different SIs. For parametric classifiers, we use the classic Bhattacharya distance as the SI and propose a simpler modified Bhattacharya distance. For non-parametric schemes we use the classic geometrical SI and propose two new geometrical SIs, viz. modified geometrical SI and nearest neighbor based separability index. The utilities and implications of these SIs are demonstrated by using them in real ATR exercises.

Abstract: Generalized principal component analysis integrating class information (ICGPCA) is proposed for feature extraction in this paper. Firstly we compute wavelet coefficients of images using DB2 wavelet and extract the approximate sub-image of wavelet transformation, and then extract the feature of the sub-image using ICGPCA which maximizes the between-class scatter and minimizes the within-class scatter. Experimental results adopting nearest neighbour classifier (NNC) and support vector machine (SVM) classifier show that the proposed method can extract effective features with lower dimensions, consequently enhance the correct probability of recognition and decrease the recognition computation effectively. The recognition rate without target azimuth information arrives at nearly 97%. (4 pages)

Abstract: Taking into account sparsity of the reflectivity function of several radar targets of interest, efficient low-complexity Automatic Target Recognition (ATR) systems can be designed. A low-dimensional 2D spatial model, where information on the radar target signature is compressed, can be estimated using High Range Resolution (HRR) data from a sparse system of view angles. Incoherent tomographic processing of HRR data from a distributed surveillance system, made up of several radar nodes, is studied in this paper. A sparse angular sampling scheme is proposed, which exploits diversity due to both the distributed radar system and the target motion. The novelty is in the exploitation of this locally dense, but otherwise sparse set of viewing angles of the targets, obtained using a sparse network of radars. The de-ghosting efficiency of such a sampling scheme is demonstrated geometrically. This results in identification of minimal information resources for unambiguous estimation of a 2D target model, useful for radar target classification.

Abstract: This paper deals with the topic of Automatic Target Recognition (ATR) of Non Cooperative Targets. Specifically the focus is on the ATR of ships from multiple ISAR images. An investigation on different techniques for the fusion of the information provided by multiple image frames is presented. The techniques exploit the principles of multi-feature based ATR and apply them to the case of availability of several images. The recognition process makes use of a wire-frame models library which undergoes a step of candidate models selection before feeding the target model declaration step. Both cases of centralized and decentralized data fusion techniques are considered. The performance of the proposed techniques is investigated in depth by means of simulated data. Moreover the paper provides a comparison of the results of the proposed techniques when applied to live ISAR images of ship targets.

Abstract: Airborne radar tracking in moving ground vehicle scenarios is impacted by sensor, target, and environmental dynamics. Moving targets can be assessed with 1-D High Range Resolution (HRR) Radar profiles with sufficient signal-to-noise (SNR) present which contain enough feature information to discern one target from another to help maintain track or to identify the vehicle. Typical radar clutter suppression algorithms developed for processing moving ground target data not only remove the surrounding clutter but also a portion of the target signature. Enhanced clutter suppression can be achieved using a multi-channel signal subspace (MSS) algorithm which preserves target features. In this paper, we (1) exploit extra information from enhanced clutter suppression for automatic target recognition (ATR), (2) present a gain comparison using displaced phase center antenna (DPCA) and MSS clutter suppressed HRR data; and (3) develop a confusion-matrix identity result for simultaneous tracking and identification (STID). The results show that more channels for MSS increase ID over DCPA, result in a slightly noisier clutter suppressed image, and preserve more target features after clutter cancellation

Abstract: A passive automatic target recognition (ATR) system includes a range map processor configured to generate range-to-pixel map data based on digital elevation map data and parameters of a passive image sensor. The passive image sensor is configured to passively acquire image data. The passive ATR system also includes a detection processor configured to identify a region of interest (ROI) in the passively acquired sensor image data based on the range-to-pixel map data, and an ATR processor configured to generate an ATR decision for the ROI.

Abstract: In this paper we extend the investigation on the use of Doppler signatures for human motion detection. As it is well known, human movements generate additional Doppler frequencies on top of the main Doppler carrier. In recent times research has been trying to exploit this effect for dynamic feature extraction and, consequently, automatic target recognition. Here a simulation tool is developed and measurements are presented as a term of comparison. Several scenarios that model arbitrary trajectories and multiple targets are analyzed. Finally we provide an insight into the potential using more than one sensor. This may improve dramatically the performance of the system as it does in many other radar applications.

Abstract: This paper presents aircraft target recognition (ATR) system using Inverse Synthetic Aperture Radar (ISAR). The methodology used to design the complete processing chain from the acquisition step to the recognition (classification) step is based on the artificial intelligence approach. This process is known as Knowledge Discovery from Data (KDD) which we have adapted to radar target recognition system. We propose a new method for target shape extraction from ISAR images based on the combination of a modified SUSAN Algorithm and Variational of Level Set. To guarantee the invariance in translation and rotation of the extracted shape, the momentinvariants and Fourier descriptors are used. In the second part of this work, We have investigated the impactof the information fusion on our recognition system. Therefore, three combination strategies: probability theory, majority vote and belief theory are applied at score and decision level. The classification results are obtained using Support Vector Machine (SVM) classifier. In the last section, experimental results are provided and discussed.