Problem Of Target Recognition Research Articles

Point-wise discriminative auto-encoder with application on robust radar automatic target recognition

Deep learning methods have a wide range of applications on various recognition problems for their ability to extract hierarchical features. However, most of the deep algorithms learn features from both the target area and the background area fulfilled with noise or clutter. In radar target recognition applications, the performance of applying deep learning methods directly on data with noise or clutter background is restricted because high-level features can be polluted by a significant amount of irrelevant patterns in the background. Therefore, a deep learning based model, point-wise discriminative auto-encoder, is proposed in this paper to extract noise and clutter robust features from the target area. We bind the original auto-encoder (AE) with a target area extraction net, which can learn the target area automatically from the raw data, to reduce the influence of the noise or clutter background. Moreover, we take advantage of the label information by adding a supervised constraint in our algorithm to help the target area extraction net learn the target area precisely and to extract discriminative high-level features. And then, our proposed method is applied to both high-resolution range profile (HRRP) data and synthetic aperture radar (SAR) images in radar automatic target recognition (RATR) problems. Experimental results on the measured HRRP data and SAR images show the advantages of our proposed method in real applications.

Statistical Modeling With Label Constraint for Radar Target Recognition

Motivated by the problem of radar target recognition, we develop a label-aided factor analysis (LA-FA) model for statistical modeling of high-resolution range profile (HRRP) under the prerequisite that the HRRP data are Gaussian distributed. The LA-FA model is the extension of the multitask learning-based factor analysis (MTL-FA) model, which is mainly applied to the recognition problem with small training data size. Compared to the MTL-FA model, our LA-FA model introduces the discrete class labels via Sigmoid-Bernoulli hierarchy to restrict the learning of model parameters, which offers the potential to enhance the separability of statistical models from different classes, thus beneficial to the improvement of recognition capability. In addition, since the noise level of a test sample is usually different from those of the training samples in the real application, we introduce a noise-robust modification method for Gaussian-based models. The proposed modification method is implemented by updating the noise level parameter of the statistical model according to the estimated signal-to-noise ratio (SNR) of test HRRP. Experiments on measured HRRP data demonstrate the better recognition performance of the LA-FA model with limited training data and our noise-robust model modification method under low test SNR. Especially, when there are 20 training HRRP samples per frame, the recognition rate of our LA-FA model is 7% higher than that of the MTL-FA model, and moreover, the recognition accuracy of the noise-robust LA-FA model is 3% higher than that of the LA-FA model without modification under the condition of SNR = 15 dB.

Conversion of Polarization Characteristics of Sounding Signals by a Radar Target

Исследовано влияние движения объекта рассеяния на характеристики собственного базиса. Определено влияние модуляции поляризационных характеристик зондирующего сигнала на вероятность обнаружения цели. Предложена методика расчета потенциальных возможностей повышения технических характеристик РЛС путем целенаправленного изменения во времени структуры зондирующего сигнала. Предложено использовать многопозиционную поляризационную модуляцию для решения задачи распознавания цели. Определены пути повышения арсенала многопозиционной поляризации. Для снятия априорной неопределенности состояния радиолокационного канала, связанной с динамически изменяющимися эффективными площадями рассеяния, использована метрика Бхаттачария. Получены на основе распределения Уишарта распределения вероятностей выборочных значений основных показателей состояния радиолокационного канала. Синтезированы алгоритмы и определены объемы выборки, необходимые для получения практически не смещенных и эффективных оценок основных показателей состояния эффективной площади рассеяния радиолокационной цели. Предложено использовать инварианты поляризационной матрицы рассеяния мощности цели для решения задачи классификации целей. На основе распределения Уишарта определены распределения выборочных значений следа и определителя матрицы рассеяния мощности. Получены алгоритмы вычисления инвариантов и определены интервалы наблюдений, при которых оценки выборочных значений становятся практически не смещенными и эффективными.

Aircraft Type Identification Using Reinforcement Active Learning

As a subclass of Automatic Target Recognition problem, Automatic Aircraft Recognition plays an important role in air traffic management and modern battlefield for automatic monitoring and detection. The research on Automatic Aircraft Recognition is still in the exploratory stage. Since aircrafts move at high speeds in complex background, it is still a challenging task of fast data processing and accurate aircraft type recognition. Besides, active learning has recently attracted many researcher’s interesting. Based on this, we employ a learning-based approach which combines active learning with reinforcement learning to learn how and when to request labels for the aircraft type recognition problem. The experimental results show that the model can achieve a good prediction accuracy with few label requests.

SCATTERING TARGET IDENTIFICATION BASED ON RADIAL BASIS FUNCTION ARTIFICIAL NEURAL NETWORKS IN THE PRESENCE OF NON-STATIONARY NOISE

The paper deals with the radar target discrimination problem performed on complex radar images. The approach based on radial basis function (RBF) artificial neural network (ANN) is proposed for the identification of point scatterers placed within a radar image. The renewed concept of simple adaptive units as the foundation for network assembling allows one to design an ANN-based feature extraction scheme for the two-dimensional signal processing. It was shown that ANN implementing RBF neural processing units could be applied for the identification of radar targets described by the set of separated scatterers, even in cases where the relative distance between the scatterers is comparable to or less than the effective width of each scatterer. The obtained results indicate a high accuracy estimation of separate scatterer centers in the presence of noise which is not limited to the stationary case but supposed to be cyclostationary. It was also shown that the parameters describing the coordinates of scattering centers could be successfully extracted from the trained ANN after about one hundred epochs spent on ANN training process, which is carried out by means of modified gradient descent method. The main result is to demonstrate the possibility of using neural networks to automatically analyze radar images, which is an integral part of a set of tasks that form the target recognition problem. The proposed algorithm implements an approach of identification systems made using a neural network training procedures.

Intelligent fault detection of high voltage line based on the Faster R-CNN

To realize intelligent fault detection of high voltage line, a deep convolution neural network method based on Faster R-CNN method is proposed to locate the broken insulators and bird nests. With the region proposal network, the Faster R-CNN chooses a random region in the features of the image as the proposal region, and trains them to get the corresponding category and location for a certain component in the image. Since the internal and regional features of the image can be learned, the Faster R-CNN method transforms the problem of target classification into the problem of target detection and recognition. Based on the ResNet-101 network model, the damage of insulators and bird nests in the electric power line can be located effectively.

Synthetic Aperture Radar Target Recognition Using Weighted Multi-Task Kernel Sparse Representation

As an extension of traditional sparse representation (SR), kernel SR has received great interest recently in the areas of computer vision and pattern recognition. It shows a considerable capacity to map linearly inseparable data into high-dimensional feature space via nonlinear mapping technique, and has been widely used in target recognition problems. In this paper, we propose a new weighted multi-task kernel sparse representation method to solve the synthetic aperture radar (SAR) target recognition problem. To capture the spatial and spectral information of a SAR target simultaneously, the proposed method explores the monogenic signal transformation to generate multi-scale monogenic features at first. Then, the proposed method provides a unified framework, named multi-task kernel sparse representation, for SAR target classification. The framework implicitly maps monogenic features into a high-dimensional kernel feature space by using the nonlinear mapping associated with a kernel function. In the kernelized subspace, SAR target recognition is formulated as a joint covariate selection problem across a group of related tasks. Furthermore, a multi-task weight optimization scheme is developed to compensate for the heterogeneity of the multi-scale features and enhance the recognition performance. Extensive experimental results tested on the public moving and stationary target acquisition and recognition (MSTAR) dataset demonstrate that our proposed method achieves better recognition performance than other existing competitive algorithms.

Target identity recognition method based on RCS distribution and spatial location

In order to solve the problem of difficult target recognition caused by target tracking discontinuity in complex battlefield environment, We propose a target identity recognition method based on the combination of RCS distribution and spatial position. First select a certain spatial position rule of the target to be identified, predict of the target trajectory using the Runge-Kutta equation, get ballistic data covering the target tracking all time period, match the predicted full trajectory with the ballistic data of the remaining time period of the target to be identified, Obtain identity recognition result. At the same time, the probability density distribution is obtained for the smoothed target RCS sequence, calculate the K-L distance between the probability density distributions to obtain the identity recognition result. Finally, the above two identity recognition results are fused to obtain the final identity recognition result. The method is simple and fast, and the similarity recognition effect can be achieved when the multi-target flight trajectory is close. The correctness of the method is verified by simulation data.

Return of function after CNS axon regeneration: Lessons from injury-responsive intrinsically photosensitive and alpha retinal ganglion cells.

This review addresses issues relating to the survival and axon regeneration of both intrinsically photosensitive retinal ganglion cells (ipRGC) and αRGC, and possible ensuing patterns of functional recovery after optic nerve crush, all of which are broadly relevant to recovery from injury in the central nervous system (CNS) as whole. Although much needs to be clarified about the connectivity, function and patterns of myelination of regenerated CNS axons, the results of recent research on activity-induced αRGC axon regeneration associated with functional restitution have highlighted key focal obstacles to recovery including neurotrophic support, axon misguidance, target recognition failure and dysmyelination. Pan RGC survival/axon regeneration requires receptor binding and downstream signalling by a cocktail of growth factors, more generally defined in the CNS by the individual trophic requirements of neuronal subsets within a given disconnected centre. Resolution of the problem of failed axon guidance and target recognition is complicated by a confounding paradox that axon growth inhibitory ligand disinhibition required for axon regeneration may mask axon guidance cues that are essential for accurate re-innervation. The study of the temporal parameters of remyelination of regenerated αRGC axons may become feasible if they establish permanent homologous connections, allowing time for new myelin sheaths to fully form. Unless near complete re-innervation of denervated targets is re-instated in the CNS, debilitating dysfunctional neurological sequelae may ensue from the resulting imbalance in connectivity.

A Sparse-Driven Anti-Velocity Deception Jamming Strategy Based on Pulse-Doppler Radar with Random Pulse Initial Phases.

This paper focuses on developing an anti-velocity jamming strategy that enhances the ability of a pulse-Doppler (PD) radar to detect moving targets in the presence of translational and/or micro motion velocity jamming generated by the digital radio frequency memory (DRFM) repeat jammers. The strategy adopts random pulse initial phase (RPIP) pulses as its transmitted signal and thus gets DRFM jammers not adaptable to the randomness of initial phase of the transmitted pulses in the pulse repetition interval (PRI) domain. The difference between the true target echo and the false target jamming signal at each PRI is then utilized to recognize the true and false target signals. In particular, an entropy based multi-channel processing scheme is designed to extract the information of the received signal without the assumption that true and false targets must be both included within one coherent processing interval (CPI). Information such as the component of the received signal (target echo only, jamming only or both) or the operating manner of DRFM repeat jammer can be gained (if jamming exists). Meanwhile, we solve the false target recognition problem under sparse theory frame and our previous work named the short-time sparse recovery (STSR) algorithm is introduced to recover the motion parameters of the true and/or false targets in the time-frequency domain. It should be pointed out that both the translational false target jamming and micro motion target jamming can be recognized in our strategy. The performance of the proposed strategy is compared with the correlated processing (CP) method used by most extant strategies. It is shown that the proposed strategy can successfully recognize the existence of true and/or false targets and keep its power in recovering corresponding motion parameters even when the jamming environment is strong.

Learning using privileged information for HRRP‐based radar target recognition

A novel machine learning method named extended support vector data description with negative examples (ESVDD-neg) is developed to classify the fast Fourier transform-magnitude feature of complex high-resolution range profile (HRRP), motivated by the problem of radar automatic target recognition. The proposed method not only inherits the close non-linear boundary advantage of support vector data description with negative examples model but also incorporates a new learning paradigm named learning using privileged information into the model. It leads to the appealing application with no assumptions regarding the distribution of data and needs less training samples and prior information. Besides, the second order central moment is selected as privileged information for better recognition performance, weakening the effect of translation sensitivity, and the normalisation contributes to eliminating the amplitude sensitivity. Hence, there will be a remarkable improvement of recognition accuracy not only with small training dataset but also under the condition of low signal-to-noise ratio. Numerical experiments based on two publicly UCI datasets and HRRPs of four aircrafts demonstrate the feasibility and superiority of the proposed method. The noise robust ESVDD-neg is ideal for HRRP-based radar target recognition.

Genetic algorithm-tuned adaptive pruning SVDD method for HRRP-based radar target recognition

ABSTRACTA novel machine learning method named adaptive pruning support vector data description (APSVDD) is developed to classify the FFT-magnitude feature of complex high-resolution range profile (HRRP), motivated by the problem of radar automatic target recognition (RATR). The APSVDD method not only inherits the advantage of least square support vector machine (LSSVM) model, which owns low computational complexity with linear equality constraints so that it is convenient to prune the boundary of SVDD dynamically and rapidly, but also overcomes the shortcoming of ability to deal with outliers in SVDD so that it can enclose targets and exclude outliers simultaneously. Genetic algorithm (GA) tunes the pruning direction of ‘shear’ dynamically, reducing the empirical risk. And fuzzy membership contributes to decision of classes for multiclass fuzzy areas. Besides, similar to the LSSVM, the distribution information within classes is found by least square method and applied for adjusting the pruning depth of ‘shear’ in APSVDD. Hence, there will be a remarkable improvement in recognition accuracy and generalization performance. Numerical experiments based on two publicly UCI datasets and remotely sensed data of four aircrafts can demonstrate the feasibility, repeatability and superiority of the proposed method. The APSVDD is ideal for HRRP-based radar target recognition.

Self-cleaning windows for downhole and process chemistry environment

This review addresses issues relating to the survival and axon regeneration of both intrinsically photosensitive retinal ganglion cells (ipRGC) and αRGC, and possible ensuing patterns of functional recovery after optic nerve crush, all of which are broadly relevant to recovery from injury in the central nervous system (CNS) as whole. Although much needs to be clarified about the connectivity, function and patterns of myelination of regenerated CNS axons, the results of recent research on activity-induced αRGC axon regeneration associated with functional restitution have highlighted key focal obstacles to recovery including neurotrophic support, axon misguidance, target recognition failure and dysmyelination. Pan RGC survival/axon regeneration requires receptor binding and downstream signalling by a cocktail of growth factors, more generally defined in the CNS by the individual trophic requirements of neuronal subsets within a given disconnected centre. Resolution of the problem of failed axon guidance and target recognition is complicated by a confounding paradox that axon growth inhibitory ligand disinhibition required for axon regeneration may mask axon guidance cues that are essential for accurate re-innervation. The study of the temporal parameters of remyelination of regenerated αRGC axons may become feasible if they establish permanent homologous connections, allowing time for new myelin sheaths to fully form. Unless near complete re-innervation of denervated targets is re-instated in the CNS, debilitating dysfunctional neurological sequelae may ensue from the resulting imbalance in connectivity.

Honeywell confirms contract awards for Al-Zour refining & PC complex

This review addresses issues relating to the survival and axon regeneration of both intrinsically photosensitive retinal ganglion cells (ipRGC) and αRGC, and possible ensuing patterns of functional recovery after optic nerve crush, all of which are broadly relevant to recovery from injury in the central nervous system (CNS) as whole. Although much needs to be clarified about the connectivity, function and patterns of myelination of regenerated CNS axons, the results of recent research on activity-induced αRGC axon regeneration associated with functional restitution have highlighted key focal obstacles to recovery including neurotrophic support, axon misguidance, target recognition failure and dysmyelination. Pan RGC survival/axon regeneration requires receptor binding and downstream signalling by a cocktail of growth factors, more generally defined in the CNS by the individual trophic requirements of neuronal subsets within a given disconnected centre. Resolution of the problem of failed axon guidance and target recognition is complicated by a confounding paradox that axon growth inhibitory ligand disinhibition required for axon regeneration may mask axon guidance cues that are essential for accurate re-innervation. The study of the temporal parameters of remyelination of regenerated αRGC axons may become feasible if they establish permanent homologous connections, allowing time for new myelin sheaths to fully form. Unless near complete re-innervation of denervated targets is re-instated in the CNS, debilitating dysfunctional neurological sequelae may ensue from the resulting imbalance in connectivity.

Research on Camouflage Target Detection via Computational Analysis with Spectral Imaging

Camouflage target detection is one of most important techniques in a great number of applications such as remote sensing, security monitoring and industrial production. In this paper, high-resolution spectral imaging together with computational analysis on spectral data cube is adopted to solve camouflage target recognition problem. Spectral data cubes of certain experimental scene are collected by a self-developed visual-band imaging spectrometer, and spectral information divergence (SID) is adopted as difference descriptor to discriminate abnormal target. SIDs of the whole spectral region and several specific ranges are all measured to quantitatively evaluate the effort of band selection on camouflage detection. This is proven to be a low-cost and effective tool for camouflage target detection and can be further developed beyond visual band spectral imaging technique.

An efficient algorithm to decompose a compound rectilinear shape into simple rectilinear shapes

Detection of a compound object is a critical problem in target recognition. For example, buildings form an important class of shapes whose recognition is important in many remote sensing based applications. Due to the coarse resolution of imaging sensors, adjacent buildings in the scenes appear as a single compound shape object. These compound objects can be represented as the union of a set of disjoint rectilinear shaped objects. Separating the individual buildings from the resulting compound objects in a segmented image is often difficult but important nevertheless. In this paper we propose a new and efficient technique to decompose a compound shape into a set of simple rectilinear shapes. First, the true interior and exterior corner points of the compound object are extracted. A modified corner detector based on polygonal approximation is proposed to accurately determine the boundaries of compound shapes. The compound shape is then split at the interior corner points to minimize the difference between the perimeter of the compound object and the sum of the perimeters of the decomposed objects. We have systematically compared the results our algorithm with those of existing approaches and the results show that the proposed algorithm is more accurate than the algorithms in the literature in terms of accuracy of perimeter estimation and computational cost.

Development of Mobile Robot Target Recognition and Following Behaviour Using Deep Convolutional Neural Network and 2D Range Data

Abstract In this work, a novel approach allowing a mobile robot to recognize and follow dynamic targets is proposed. The designed behavior can simplify subsequent development of autonomous navigation and obstacle avoidance tasks as well as the capability of mobile platforms to operate within multi-agent formations. In the proposed approach, the target recognition problem is considered as an image classification task and it tackled by using a state-of-the-art deep convolutional neural network working with a sequence of images obtained from the environmental scene. The subsequent target following problem is implemented by using 2D range data. The proposed approach has been implemented on a KUKA youBot omnidirectional mobile platform using robot operating system middleware and equipped with an onboard USB camera and a 2D LiDAR. The conducted experiments have shown that the mobile robot is capable to detect and follow accurately another mobile platform (iRobot Create) in a shared environment. The target following robotized platform is also capable to maintain a pre-specified relative position and orientation with respect to the followed dynamic object by controlling its own linear and angular velocity.

SAR Automatic Target Recognition Using Joint Low-Rank and Sparse Multiview Denoising

In recent years, many researchers have focused on the automatic target recognition problem for high-resolution synthetic aperture radar (SAR) systems. Most have directly employed the training data as the dictionary, which introduces error from speckle noise. In this letter, a joint low-rank and sparse multiview denoising (JLSMD) dictionary is generated, which combines multiview training samples for denoising. To extract the dictionary, we fully consider the low-rank property of multiview target images and the sparsity of speckle noise for SAR systems. The designed dictionary is more accurate than the training data in representing the target. With the help of the proposed JLSMD dictionary, we develop three algorithms based on the sparse representation classification and the support vector machine approach. We carry out experiments on the moving and stationary target acquisition and recognition public data set to evaluate the excellent performance of the proposed methods against several state-of-the-art methods, including deep learning methods.

Deep kernel learning method for SAR image target recognition.

With the development of deep learning, research on image target recognition has made great progress in recent years. Remote sensing detection urgently requires target recognition for military, geographic, and other scientific research. This paper aims to solve the synthetic aperture radar image target recognition problem by combining deep and kernel learning. The model, which has a multilayer multiple kernel structure, is optimized layer by layer with the parameters of Support Vector Machine and a gradient descent algorithm. This new deep kernel learning method improves accuracy and achieves competitive recognition results compared with other learning methods.

About the problems and methods for finding of small size objects in the image

This work is devoted to the development of a new system of search and identification unmanned vehicle using a machine vision system, namely of guidance system. The problem of automatic aerial target recognition based on of the digital filtering streaming video and questions of informative signature formation. Machine vision systems are used for the treatment and recognition of different images obtained from a camera. A vision system is a means of tracking, monitoring and automatic decision-making in certain situations. This article discusses of design features of guidance system for unmanned aerial vehicle. The guidance system operates in real time. Is proposed detection algorithm of small-sized unmanned aerial vehicle. From the results obtained, generated control command for the guidance the unmanned aircraft to target. The main objective of the search algorithm and identification of air targets, there is a search for the reference point of the desired geometric object. To find the sought point of the contour, the used to group of digital filters for three channels of streaming video. Represented and implemented algorithm the block guidance when using frame difference. The proposed algorithm for finding of small objects can detect moving objects according to the optical sensor to the changing position of the optical sensor. The work concludes by demonstrating the proposed algorithm.