Search

High Resolution Range Profiles (HRRPs) have become a key area of focus in the domain of Radar Automatic Target Recognition (RATR). Despite the success of deep learning based HRRP recognition, these methods needs a large amount of training samples to generate good performance, which could be a severe challenge under non‐cooperative circumstances. Currently, deep learning based models treat HRRPs as sequences, which may lead to ignorance of the internal relationship of range cells. This letter proposes HRRPGraphNet, a novel graph‐theoretic approach, whose primary innovation is the use of the graph‐theory of HRRP which models the spatial relationships among range cells through a range cell amplitude‐based node vector and a range‐relative adjacency matrix, enabling efficient extraction of both local and global features in noneuclidean space. Experiments on the aircraft electromagnetic simulation dataset confirmed HRRPGraphNet's superior accuracy and robustness compared with existing methods, particularly in limited training sample condition. This underscores the potential of graph‐driven innovations in enhancing HRRP‐based RATR, offering a significant advancement over sequence‐based methods. [ABSTRACT FROM AUTHOR]

Sparse representation-based classification (SRC) possesses remarkable characteristics for application in synthetic aperture radar (SAR) automatic target recognition (ATR), for instance, inherent feature extraction and robustness to articulation, and so on. However, the performance of SRC is highly sensitive to parameters such as sufficient training samples, SAR images quality, and targets' changing conditions in depression, pose, configuration, and so on. Unfortunately, the training sample resources for SAR ATR are often expensive and scarce. Further, the targets in SAR images, even with slightest changes in conditions, display mutable characteristics attributable to unique SAR image formation, and speckle noise corruption. To overcome these obstacles, this Letter proposes to establish several compact and complementary dictionaries using monogenic signal's components of SAR images and Fisher discriminative dictionary learning. Then, an optimal decision fusion (ODF) strategy is proposed, which utilises SRC and the latter dictionaries for robust SAR ATR. Compared with single classifiers or multiple parallel classifiers, the proposed ODF increases the accuracy of recognition, while at the same time, decreases the complexity of the system. The proposed methods have considerable features making them applicable in practical situations. Based on the experimental results, the proposed methods outperform state-of-the-art approaches. [ABSTRACT FROM AUTHOR]

A new simulation and processing methodology based on open source tools to produce high fidelity synthetic aperture radar (SAR) simulations of ground vehicles of varying types, as well as analysis of an applied automatic target recognition (ATR) technique is presented in this Letter. This work is based around the RaySAR open-source model and the outputs have been configured for both monostatic and bistatic geometries. Input CAD models of various military and civilian vehicles are used to produce the SAR imagery. This output imagery was then used to train a tiny you only look once convolutional neural network (CNN) classifier. The classification success of the CNN applied was showed to produce significantly accurate results and the whole pipeline of processing enabled rapid evaluation of potential ATR methods against targets of choice. [ABSTRACT FROM AUTHOR]

Automatic target recognition of ground surveillance radars is a beneficial practice that reduces cost, error, duration of detection and response time. The identification of echoes from eight classes using an RF pulsed Doppler radar is studied. Instead of traditional methods, the wavelet packet analysis (WPA) is used. WPA, by providing arbitrary time-frequency resolution, enables analysing signals of stationary and non-stationary nature. It has better time representation than Fourier analysis and better high-frequency resolution than wavelet analysis. WPA subimages are further analysed to obtain feature vectors of log energy entropy in all cases except one case where norm entropy is used. These features are fed into a multilayer perceptron for classification which is done in a hierarchical scheme and composed of binary classification steps. Hundred per cent accuracy was possible in all steps except two which were realised at 99.5 and 98.2%. An overall classification rate of 98.2% was achieved. The results were compared with other published results that report accuracy in the same application field. The proposed method achieved the highest accuracy although it had the highest number of classes. The method provides a promising tool for the automatic recognition of ground surveillance radar targets. [ABSTRACT FROM AUTHOR]

To improve the accuracy and robustness of high-resolution range profile (HRRP) target recognition, in this paper, the multi-scale fusion sparsity preserving projections (MSFSPP) approach is proposed for feature extraction. Compared with traditional multi-scale feature extraction method, the proposed MSFSPP approach utilises features in every scale and their sparse reconstructive relationship to construct multi-scale fusion features which contain more discriminating information. Support vector machine is employed to verify the classification performance of features extracted by MSFSPP and related feature extraction methods. Simulation results based on the measured aircraft datasets show that the proposed MSFSPP approach has outperformance with a small amount of data. [ABSTRACT FROM AUTHOR]

An interview with Michael Woollard from University College London, is presented. Topics discussed include the background to the work that is presented in Woollard Electronics Letters submission; information on the fastest convolutional neural network (CNN) approaches; and challenges he'd overcome during the research for the letter.

A feature processing technique that is specifically designed for synthetic aperture radar images, where features are extracted with merits of compactness, high discrimination ability, and shift invariance is proposed. To this end, a rich feature set is first constructed with a wealth of discrimination information. Then, the redundancy and dimensionality of the rich feature set are reduced such that a much more compact and efficient feature set can be achieved. Finally, the advantages of the compact feature set are further explored by learning the relationships among features statistically in discriminative fashion. Experimental results show the efficiency of the proposed method. [ABSTRACT FROM AUTHOR]

High-resolution range profile is the significant characteristic of radar targets in automatic target recognition. Traditional feature extractions of range profiles in target classification are constrained to the original scale. This Letter proposes a multi-scale target classification method based on the scale-space theory. Target range profile feature is extended from single scale to multiple scales. The minimum Kullback-Leibler mean divergence (MKMD) algorithm is developed to achieve the automatic optimal scale factor selection. Classification evaluations on aircraft models using support vector machine and 3-nearest neighbour classifiers demonstrate that the application of scale-space theory in multi-scale feature extraction could effectively enhance the classification performance. The feasibility of the proposed MKMD algorithm is also validated by an enumeration method. [ABSTRACT FROM AUTHOR]

A method combing simple linear and nonlinear filters is proposed for real-time 'non-uniformity correction' of infrared video sequences, which suppresses 'ghosting' artefacts due to both lack of motion and strong edges. In this 'least mean square' (LMS)-based method, a mean filter is used first, when the 'fixed pattern noise' (FPN) level is high, taking advantage of its noise smoothing capability. When the FPN level drops to a low level, a sigma filter is used instead to reduce edge smearing. The sigma filter is also used to detect abnormal pixels like dead pixels and pixels contaminated by impulse noise, in addition to adaptive adjustment of the learning rate, with no extra cost. Experiments with simulated data and real infrared sequences show that the proposed method outperforms several other LMS methods. It is of the same computational complexity as Scribner's method, which makes it a good candidate for real-time hardware implementation. [ABSTRACT FROM AUTHOR]