Target Detection Techniques Research Articles

Research on semi-supervising learning algorithm for target model updating in target tracking

Target detection and tracking technique is one of the hot subjects in image processing and computer vision fields, which has significant research value not only in military areas such as imaging guidance and military target tracking, but also for civil use such as security and monitoring and the intelligent man-machine interaction. In this paper, for target deformation, scale changing, rotation, and other issues in the long-term stable target tracking, a bootstrapping feedback learning algorithm is proposed, which may improve the target model and the classifier discriminating capacity as well as the fault tolerance ability; and it also makes fewer errors during the updating, and then the proof of convergence of the algorithm is given. Experimental results show that among the same tracking algorithms, utilization of the learning method to update the target model and classifier is more stable and more adaptable than unusing it in the processes of target scale changing, deformation, rotation, perspective changing and fuzzy. And compared with the existing conventional method, this method has a better robustness, and a high value in practical application and research.

Trail Coverage : A Coverage Model for Efficient Intruder Detection near Geographical Obstacles in WSNs

In wireless sensor network (WSNs), the coverage strategy for achieving higher performance in intruder detection probability near irregular geographical obstacles inside a military terrain, is studied in this paper. The real life problem of the of intruder detection near irregular geographical obstacles has not been focused so far, in which coverage requirement is very high, hard and expensive, if we require the entire boundary of an obstacle in the region has to be covered. Therefore, a novel coverage problem based on data fusion detection system, called Trail Coverage is proposed in this paper. A sensor network with Trail Coverage provides higher performance rate in detection probability of any intruder that moves along the shortest passes between the geographical obstacles inside a military terrain. We consider various preliminary steps to provide a strong basement for this Trail Coverage model. We state a probabilistic model to locate the obstacles inside the region, with randomly deployed sensor nodes, during initial network configuration. Based on computational geometry techniques we design a O1􀀀 Continuous Obstacle Hull to find the trail region between the obstacles, for the node deployment, to achieve Trail Coverage. We adopt fusion based target detection technique to obtain effective performance in intruder detection probability at the trail region. Through analytical results, we show, the significance of the Trail Coverage strategy to meet the desired performance requirements of an intruder detection system, while maintaining greater network coverage with a given fusion range

Spatial and temporal features selection for low-altitude target detection

Target detection in plane position indicator (PPI) radar images aims at separating moving targets from complicated background image. Background subtraction is a powerful mechanism for such applications. Since there is still much clutter left in the foreground image after background subtraction, an optimal classification plane (OCP) should be constructed to distinguish targets from clutters. Due to the complexity and variability of the foreground and background images, the threshold value of each position of the OCP should be selected adaptively corresponding to each pixel of the foreground image. In this paper, a novel method is proposed to improve the classification results of the OCP with the spatial and temporal features from the PPI radar image sequence. Firstly, to select the thresholds in the OCP adaptively, a new formula is developed with the spatial features from two statistical models. The statistics from the foreground model reflect the aggregation degree of the concerned pixels, while those from the background model reflect their relative positions. Secondly, to further reduce the false alarm rate, a novel strategy based on the temporal features is incorporated to modify the OCP. Our method with the optimal parameter values is compared with the other successful techniques for target detection. Quantitative evaluations show that the proposed method provides better results.

Uncooperative pose estimation with a LIDAR-based system

This paper aims at investigating the performance of a LIDAR-based system for pose determination of uncooperative targets. This problem is relevant to both debris removal and on-orbit servicing missions, and requires the adoption of suitable electro-optical sensors on board of a chaser platform, as well as model-based techniques for target detection and pose estimation. In this paper, a three dimensional approach is pursued in which the point cloud generated by a LIDAR is exploited for pose estimation. Specifically, the condition of close proximity flight to a large debris is considered, in which the relative motion determines a large variation of debris appearance and coverage in the sensor field of view, thus producing challenging conditions for pose estimation. A customized three dimensional Template Matching approach is proposed for fast and reliable pose initial acquisition, while pose tracking is carried out with an Iterative Closest Point algorithm exploiting different measurement-model matching techniques. Specific solutions are envisaged to speed algorithm convergence and limit the size of the point clouds used for pose initial acquisition and tracking to allow autonomous on-board operation. To investigate proposed approach effectiveness and achievable pose accuracy, a numerical simulation environment is developed implementing realistic debris geometry, debris-chaser close-proximity flight, and sensor operation. Results demonstrate algorithm capability of operating with sparse point clouds and large pose variations, while achieving sub-degree and sub-centimeter accuracy in relative attitude and position, respectively.

A Non-destructive Bulk Currency Detection System (BCDS) for Screening Smuggled Currency

Bulk cash smuggling is a serious issue in recent years. It is estimated that $6–36B flow illegally between the United States and Mexico each year during otherwise legitimate border crossings. A nondestructive bulk currency detection system (BCDS) for screening pedestrians, vehicles and shipping containers will significantly improve boarder security, protect domestic economics and deter criminal and terrorist activity. However, such a practical device is not yet available because neither the target detection technique nor the instrumentation has been defined. In this paper, we evaluated the currency characteristics that provide a basis for detection, relevant physical/chemical instrumental approaches, and a rationale for selection of an approach to build a practical BCDS. US currency releases volatile compounds. These emissions can be used as a chemical signature in currency detection. The literature regarding characterization of volatile compounds associated with US currency is limited. To identify the unique vapor signature of US currency, we analyzed VOC emission from currency bills using thermal desorption GC/MS. This enabled calculation of emission rates of volatile compounds. The results were consistent with prior studies of currency emissions. From our data and the literature, a tentative fingerprint that represents currency can be selected from among the complex background present in real currency. The analysis shows that a most-likely US currency vapor signature contains a specific set of volatiles at low concentrations. A successful practical BCDS based on volatiles must be able to accurately identify and quantify a speciated set of trace chemicals from a complex and highly variable matrix. Variations of GC/MS with SPME preconcentration can result in near-term methods. Future developments are needed to address optimization of methodology and meet the need for rapid sampling, detection with speciation. Sensor arrays can also provide a platform for currency detection and would improve logistics over larger instruments.

Efficient target detection using an adaptive compressive imager

The goal of a target detection system is to determine the location of potential targets in the field of view of the sensor. Traditionally, this is done using high-quality images from a conventional imager. For wide-field-of-view scenarios, this can pose a challenge for both data acquisition and system bandwidth. In this paper, we discuss a compressive sensing technique for target detection that dramatically reduces the number of measurements that is required to perform the task, as compared with the number of pixels in conventional images. This, in turn, can reduce the data rate from the sensor electronics, and along with it, the cost, complexity, and the bandwidth requirements of the system. Specifically, we discuss a two-stage approach that, first, adaptively searches a large area using shift-invariant masks to determine the locations of potential targets (i.e., the regions of interest) and then revisits each location to discriminate between target and clutter using a different set of specialized masks.We show that the overall process is not only highly efficient (i.e., dramatically reduces the number of measurements as compared with the number of pixels) but does so without appreciable loss in target detection performance.

Trends in correlation-based pattern recognition and tracking in forward-looking infrared imagery.

In this paper, we review the recent trends and advancements on correlation-based pattern recognition and tracking in forward-looking infrared (FLIR) imagery. In particular, we discuss matched filter-based correlation techniques for target detection and tracking which are widely used for various real time applications. We analyze and present test results involving recently reported matched filters such as the maximum average correlation height (MACH) filter and its variants, and distance classifier correlation filter (DCCF) and its variants. Test results are presented for both single/multiple target detection and tracking using various real-life FLIR image sequences.

Target Area Detection Based on Piecewise Membership FSVM

Aiming at the target detection technique is time-consuming, noise and other issues, improved the application of FSVM in target detection, a piecewise fuzzy membership function is proposed based on LDA algorithm. The algorithm projected the original high-dimensional samples onto a one-dimensional space by LDA ,the original sample’s fuzzy weights is segment calculated based on the distribution of the projection point, reduce the impact of noise and outliers on classification results. In the simulation experiments, this method can effectively reduce the impact of unbalanced data to FSVM, improve the classification accuracy.

Pulse Compression Techniques for Target Detection

LPI is the property of radar that, because of its low power, wide bandwidth, frequency variability, or other design attributes makes it difficult for it to be detected by means of a passive intercept receiver. Desirable properties of LPI based on periodic autocorrelation, ambiguity function, peak and integrated side lobes. LPI modulation techniques include frequency modulation, phase modulations. In phase modulation Barker, Frank, P4simulations are studied. The best code is opted for target detection based on desirable properties, Doppler tolerance and reduction in side lobes. When the transmitted and received signals are correlated a peak is generated as the indication of target detection. These correlated peaks are added non coherently to achieve target detection.

Fast Two-Dimensional CFAR Procedure

An FMCW radar system, which transmits a sequence of rapid chirp signals, is considered to measure target range and radial velocity simultaneously even in multitarget situations. The main outcome of the coherent processing procedure applied to the received echo signal is the two-dimensional range-Doppler-matrix (RDM), which is the basis for an adaptive constant false alarm rate (CFAR) target detection technique. The well-known one-dimensional cell averaging (CA) CFAR procedure suffers from masking effects in all multitarget situations and is additionally very limited in the proper choice of the reference window length. In contrast the ordered statistic (OS) CFAR is very robust in multitarget situations but requires a high computation power. Therefore two-dimensional CFAR procedure based on a combination of OS and CA CFAR is proposed. In this case any masking effects in multitarget situations can be avoided without increasing the computation complexity. In addition the size of the reference window can be adjusted in order to optimize the detection performance.

Mobile image sensors for object detection using color segmentation

The intelligent mobile robot with sensors and image processing embedded system combines the suction and aerodynamic attraction to achieve good balance between strong adhesion force and high mobility. Experimental results showed that the robot can move upward on the wall at the speed of 2.9 m/min and carry 5 kg payload in addition to 2.5 kg self-weight, which record the highest payload capacity among climbing robots of similar size. It also implements object detection ability using effective color transform and segmentation technique for the exact target detection on the wall using a embedded camera system, communication module and several active sensors.

Fusion of Radiation and Electromagnetic Induction Data for Buried Radioactive Target Detection and Characterization

In general, buried penetrators made of Depleted Uranium (DU) become hazardous waste. In addition to the detection of DU waste, it is also of interest to know their state of oxidation. However, radioactive target detection techniques usually do not differentiate between metal and oxide. In this study, data fusion techniques are applied to combine results from both the radiation detection and the electromagnetic induction (EMI) detection, so that further differentiation among DU metal, DU oxide, and non-DU metal debris may be achieved. A two-step approach is developed to accomplish decision level fusion. The approach is based on techniques such as majority voting (MV) and weighted majority voting (WMV), in combination with a set of decision rules. The fusion approach has been tested successfully with survey data collected on simulation targets.

Building detection from urban SAR image using building characteristics and contextual information

With the urgent demand on urban synthetic aperture radar (SAR) image interpretation, this article deals with detecting buildings from a single high-resolution SAR image. Based on our previous work in building detection from SAR images, aiming at extracting buildings with their whole and accurate boundaries from the built-up area, a general framework using the marker-controlled watershed transform is introduced to combine both building characteristics and contextual information. First, the characteristics of the buildings and their surroundings are extracted as markers by the target detection techniques. Second, the edge strength image of the SAR image is computed using the ratio of exponentially weighted averages detector. The marker-controlled watershed transform is implemented with the markers and the edge strength image to segment buildings from the background. Finally, to remove false alarms, building features are considered. Especially, a shape analysis method, called direction correlation analysis, is designed to keep linear or L-shaped objects. We apply the proposed method to high-resolution SAR images of different scenes and the results validate that the new method is effective with high detection rate, low false-alarm rate, and good localization performance. Furthermore, comparison between the new method and our previous method reveals that introducing contextual information plays an important role in improve building detection performance.

Comparison of machine learning techniques for target detection

This paper focuses on machine learning techniques for real-time detection. Although many supervised learning techniques have been described in the literature, no technique always performs best. Several comparative studies are available, but have not always been performed carefully, leading to invalid conclusions. Since benchmarking all techniques is a tremendous task, literature has been used to limit the available options, selecting the two most promising techniques (AdaBoost and SVM), out of 11 different Machine Learning techniques. Based on a thorough comparison using 2 datasets and simulating noise in the feature set as well as in the labeling, AdaBoost is concluded to be the best machine learning technique for real-time target detection as its performance is comparable to SVM, its detection time is one or multiple orders of magnitude faster, its inherent feature selection eliminates this as a separate task, while it is more straightforward to use (only three coupled parameters to tune) and has a lower training time.

Prohibited Zones Coverage Protection Using Aerial Mines

Mining readiness is continually emphasized. Training is conducted to ensure that selected aerial forces remain in a state of mine readiness. Naval air forces and mine assembly personnel are capable of rapid deployment to operate from overseas bases and carriers. Tactics for mine laying in a hostile environment are under constant development, and prepositioned mine stocks are held in a high state of readiness to support approved mining plans. Research and development continues for new mines and components, including such features as improved capabilities against submarine and surface targets, better resistance against minesweeping and hunting, more flexibility, easier and less expensive maintenance, simpler and faster preparation for laying, and improvement of mine detection and control systems with increased sophistication. A sophisticated technique for aerial mine presented in this paper including the technologies of the wireless communication to construct a mesh of aerial mine, and GPS technologies for the orientation positioning, also the ultrasonic techniques for target detection.

Dual-Module Data Fusion of Infrared and Radar for Track Before Detect

A track before detect method based on data fusion of infrared and radar is proposed to increase the probability of correct track initiation and shorten initiation time. Track before detect is a new technique for dim target detection and tracking which is useful when the signal-to-noise ratio of target is low. Particle filter and dynamic programming for track before detect are currently proposed for detecting and tracking dim targets in low signal-to-noise ratio background. In this paper, we apply them in dual-module data fusion of infrared and radar. Particle Filter is applied to process the acquired data from radar. Dynamic programming is applied to process the acquired data from Infrared. Sensor receives data and generates the first stage decision. The decisions are subsequently transmitted to the fusion center where they are combined into a final decision on distributed fusion architectures. The proposed method is applied to simulate track before detect in dual module system of infrared and radar. The simulation results show that the proposed method increase the probability of correct track initiation and shorten initiation time.

Multiresolution Non-Local Mean Filter for Infrared Dim and Small Target Background Suppression

Structured clouds and ground building background suppression are difficult problems for dim and small target detection technique. In this paper, the dim and small target background suppression method based on combined curvelet transform with modified non-local means filter was presented to solve the problem. And local gradient statistics is introduced to boost ability of method which suppresses false by background structure. The innovation was that the curvelet transform was adopted to decompose the input infrared image, which extracts multi-scale and directional detail features of the image. Moreover non-local means filter improved by local gradient characteristic was introduced to suppress background details and enhance target information for suppression background. Compared with two-dimensional least mean square (TDLMS) and modified partial differential equation (MPDE) methods, through visual quality and value index, several groups of experimental results demonstrate that the presented method can suppress complicated background in dim and small target image effectively.

Anomaly Detection in Hyperspectral Images Based on an Adaptive Support Vector Method

Recently, anomaly detection (AD) has attracted considerable interest in a wide variety of hyperspectral remote sensing applications. The goal of this unsupervised technique of target detection is to identify the pixels with significantly different spectral signatures from the neighboring background. Kernel methods, such as kernel-based support vector data description (SVDD) (K-SVDD), have been presented as the successful approach to AD problems. The most commonly used kernel is the Gaussian kernel function. The main problem using the Gaussian kernel-based AD methods is the optimal setting of sigma. In an attempt to address this problem, this paper proposes a direct and adaptive measure for Gaussian K-SVDD (GK-SVDD). The proposed measure is based on a geometric interpretation of the GK-SVDD. Experimental results are presented on real and synthetically implanted targets of the target detection blind-test data sets. Compared to previous measures, the results demonstrate better performance, particularly for subpixel anomalies.

Bistatic detection and imaging of proud targets in shallow water using coherent space‐time‐frequency processing.

An ongoing challenge for underwater sonar systems is to discriminate a man made target (typically having an elastic shell) from surrounding clutter returns especially in the presence of multipath. Network of autonomous systems of unmanned vehicles provides a practical means for bistatic measurements and thus potentially bistatic enhancement for target detection. One popular technique for automatic target detection is synthetic aperture imaging. An aspect that complicates this technique is the evolutional, time dependent aspect of the bistatic echo spectrum from elastic shells especially for the acoustic guided waves echo components at low to midfrequency. To this end, time‐frequency analysis, and, in particular, Wigner‐Ville analysis have been shown to provide a robust processing tool for interpreting the evolutional time dependent aspect of the scattered acoustic wave field especially around the midfrequency enhancement. Time‐frequency methods will be used to enhance the sonar images created from monostatic or bistatic low‐frequency measurements. The proposed approach will be tested using numerical simulations and experimental at‐sea data. [Work supported by the ONR.]

Effects of linear projections on the performance of target detection and classification in hyperspectral imagery

We explore the use of several linear dimensionality reduction techniques that can be easily integrated into the hyperspectral imaging sensor. We investigate their effect on the performance of classical target detection and classification techniques for hyperspectral images. Specifically, each N -dimensional spectral pixel is embedded to an M -dimensional measurement space with M ⪡ N by a linear transformation (e.g., random measurement matrices, uniform downsampling, principal component analysis). The detectors/classifiers are then applied to the M -dimensional measurement vectors and their performances are compared to those obtained from the entire N -dimensional spectrum. Through extensive experiments on several hyperspectral imagery data sets, we demonstrate that only a small amount of measurements are necessary to achieve comparable performance to that obtained by exploiting the full N -dimensional pixels.