Figure 4 - uploaded by Alauddin Bhuiyan
Content may be subject to copyright.
Three patterns with different sizes and shapes. Pattern 1) 20×21 pixels (left), 2) 17×16 pixels (center) and 3) 12×12 pixels
Source publication
The Automated Anti-personnel Mine (APM) detection and classification is currently a broad issue. The detection success depends on the feature selection that we obtain from the sensors. Ground Penetrating Radar (GPR) is one of the established sensors for detecting buried APM. In this paper, we introduce a method which improves the accuracy of detect...
Context in source publication
Similar publications
Type 2 fuzzy systems have been under investigation for a while and the projection of type 2 understanding for uncertainty management onto the connectionist models -i.e. neural networks- seems an interesting field of research. This paper considers neurons having multiple bias values defining a new structure that resembles the uncertainty handling ca...
We developed a multi-channel electromyogram acquisition system using PSOC microcontroller to acquire multi-channel EMG signals. An array of 4 times 4 surface electrodes was used to record the EMG signal. The obtained signals were classified by a back-propagation-type artificial neural network. B-spline interpolation technique has been utilized to m...
This paper presents two pattern recognition approaches using Partial Discharges fingerprints as input features to classify PD patterns. A multi-layer perceptron (MLP) backpropagation neural network and a support vector machine (SVM) were trained to recognize three types of PD patterns. Experimental results showed that the algorithms can achieve hig...
The growing interests in multiway data analysis and deep learning have drawn tensor factorization (TF) and neural network (NN) as the crucial topics. Conventionally, the NN model is estimated from a set of one-way observations. Such a vectorized NN is not generalized for learning the representation from multiway observations. The classification per...
In this article a prototype system for the detection of tumors in gastroscopic images is presented, in which fuzzy pyramid linking technique was used to help to identify tumors in an image, and neural networks were applied to give the capacity to learn by examples and to generalize this knowledge to make inferences from incomplete information. Test...
Citations
... Electromagnetic wave detection is a non-invasive object detection technique widely used in agriculture [1]. The classification or analysis of radar signals in terms of differences in electromagnetic properties has become a key research issue [2][3][4], because the modeling of agricultural soil requires efficiency and precision in order to provide the elusive physical parameters needed to meet the requirements of artificial intelligence algorithms that require multi-sample supervised learning. Artificial intelligence methods can quickly locate objects through GPR image processing, but it is very difficult to collect unknown sample characteristics, such as the electrical conductivity and water content of an object in actual farmland soil [5]. ...
... Obviously, the electromagnetic classification in the two-dimensional case can be divided into two independent groups, such as E x , E y , and H z as a group that can be denoted as TE electromagnetic field group (formula (1) and (2)). Similarly, E z , H x , and H y can be denoted as TM electromagnetic field group (formula (3) and (4)). This FDTD equation can be simplified into two-dimensional problems. ...
... The sand fraction, clay, and silt fraction are set according to the parameters described in Section 2.1. The bulk density of soils is 2 g/cm 3 . The volumetric water fraction range was 0.001-0.25. ...
Unknown objects in agricultural soil can be important because they may impact the health and productivity of the soil and the crops that grow in it. Challenges in collecting soil samples present opportunities to utilize Ground Penetrating Radar (GPR) image processing and artificial intelligence techniques to identify and locate unidentified objects in agricultural soil, which are important for agriculture. In this study, we used finite-difference time-domain (FDTD) simulated models to gather training data and predict actual soil conditions. Additionally, we propose a multi-class support vector machine (MSVM) that employs a semi-supervised algorithm to classify buried object materials and locate their position in soil. Then, we extract echo signals from the electromagnetic features of the FDTD simulation model, including soil type, parabolic shape, location, and energy magnitude changes. Lastly, we compare the performance of various MSVM models with different kernel functions (linear, polynomial, and radial basis function). The results indicate that the FDTD-Yee method enhances the accuracy of simulating real agricultural soils. The average recognition rate of the hyperbola position formed by the GPR echo signal is 91.13%, which can be utilized to detect the position and material of unknown and underground objects. For material identification, the directed acyclic graph support vector machine (DAG-SVM) model attains the highest classification accuracy among all soil layers when using an RBF kernel. Overall, our study demonstrates that an artificial intelligence model trained with the FDTD forward simulation model can effectively detect objects in farmland soil.
... Using template matching on B-scans for identification is proposed in [25,26]. In [27], another pattern recognition method is adopted for buried object classification. Input B-scan is first binarised, and then segmented using seeded region growing algorithm. ...
Buried object identification is an integral part of the humanitarian demining process which enables discarding clutter objects, thereby reducing the overall cost of landmine localisation drastically. In this study, we present a novel buried object identification method for a ground penetrating radar (GPR) and electromagnetic induction sensor (EMI)-based dual sensor hand-held system developed at TUBITAK BILGEM. The proposed approach relies on grouping the buried objects into three groups according to their metal content using their EMI responses. Features are extracted from GPR responses of the objects and landmine/clutter discrimination is achieved within each object group using k-nearest neighbour algorithm. The identification results are presented on an extensive real data set of mine simulants and clutter objects, which is collected from three different terrains with different types of soil and different burial depths. We show that the proposed method outperforms a popular method based on edge histogram descriptors.
... Using template matching on B-scans for identification is proposed in [21], [22]. In [23], another pattern recognition method is adopted for buried object classification. Input B-scan is first binarized, and then segmented using seeded region growing algorithm. ...
... Therefore, GPR has become a very important geophysical method for archaeological exploration where buried cultural resources are rare, contested, or simply off-limits (Goodman and Nishimura, 1993;Gracia et al., 2000;Edwards et al., 2000;Conyers and Goodman, 1997;Leucci and Negri, 2006;Goodman et al., 2007;Francese et al., 2009;Conyers, 2009;Shaaban et al., 2009). Also, the application of GPR in engineering, mining, and environmental geology studies is well documented and highly effective (Ulricksen, 1982;Pilon, 1992;Grandjean and Gourry, 1996;Annan, 1999;Jack and Jackson, 1999;Bhuiyan and Nath, 2006). ...
Ground Penetrating Radar (GPR) survey was conducted along 31 profiles in the ancient city of Ur. The results revealed a considerable number of anomalies interpreted as walls, both intact and demolished. These walls have a range of dimensions, varying from 0.75–7 m, and begin approximately 0.2 to 0.5 m below surface and continue down to various depths. The most important anomalies that have been distinguished are the arched gate and a grave at depths of about 0.3 m and 3.6 m respectively. The final processing of the results help to establish 3D models, which reveal the extensions and dimensions of subsurface archaeological features. Most of the walls have regular structure, effecting shapes such as rectangles and squares. These features can be related to a system of architecture that might include housing, civic buildings, and temples. In the future, these results may help to guide archaeologists and geophysicists to excavate and understand new portions of Ur, as well as other unexplored locations in Iraq and surrounding areas.
... For target detection, many feature extraction and classification techniques are proposed in literature. Artificial Neural Networks (ANNs), Support Vector Machine (SVM) are efficient machine learning approaches for detection and classification of landmines [7,8]. Target-detection and localization for pipes, voids and tanks is achieved in [9] using image processing techniques along with unsupervised neural network classifier. ...
Images from ground penetrating radar (GPR) may be obscured by high clutter noise over the target signal, making target detection difficult. In this contribution, a decluttering technique is applied to GPR images. This clutter removal/reduction is achieved through wavelet decomposition and application of two methods using the third and fourth order statistics: skewness and kurtosis. These higher order statistics remove clutter but retain target signatures. Different scenarios are considered for real GPR images collected in our controlled lab environment set up and peak signal to noise ratio are compared for the two methods. Further features of targets and non-targets are extracted from de-noised images. These features are used in training a neural network classifier. This classifier is applied to various real GPR images with promising results for detection of targets.
... Once the spatial location is obtained, the temporal location can be estimated. By far, the most common approach for temporal estimation relies on extracting keypoints at locations of high energy (e.g., local maxima) in the GPR A-scans [6], [8], [20], [29], [35]- [39]. These energy-based methods often yield multiple keypoints at each spatial location. ...
... Figure 2 illustrates a previously proposed method [37], which we term the maxsmoothed-energy keypoint (MSEK) approach. MSEK is used in this work, and is representative of most existing temporal identification approaches, though some others do exist [8], [10], [39]- [41]. ...
... Existing strategies typically choose which keypoints to utilize based on one of two ordering criteria. The first ordering criterion is to utilize keypoints at maximum energy locations, in the same way it is done during training [8], [10], [35], [39], [45]. This is denoted as "En" in Table 1. ...
Ground penetrating radar (GPR) is one of the most popular and successful sensing modalities that has been investigated for landmine and subsurface threat detection. Many of the detection algorithms applied to this task are supervised and therefore require labeled examples of target and non-target data for training. Training data most often consists of 2-dimensional images (or patches) of GPR data, from which features are extracted, and provided to the classifier during training and testing. Identifying desirable training and testing locations to extract patches, which we term "keypoints", is well established in the literature. In contrast however, a large variety of strategies have been proposed regarding keypoint utilization (e.g., how many of the identified keypoints should be used at targets, or non-target, locations). Given the variety keypoint utilization strategies that are available, it is very unclear (i) which strategies are best, or (ii) whether the choice of strategy has a large impact on classifier performance. We address these questions by presenting a taxonomy of existing utilization strategies, and then evaluating their effectiveness on a large dataset using many different classifiers and features. We analyze the results and propose a new strategy, called PatchSelect, which outperforms other strategies across all experiments.
... The application of Ground Penetrating Radar in engineering, mining, and environmental geology studies is well documented and highly effective [6][7][8][9][10][11]. ...
The researchers have tried to focus on how to determine the number of pipes that are present in one obtained hyperbola in radargram profile. Ground Penetration Radar (GPR) survey was performed to distinguish between two zero-spaced iron pipes in radargram. The field work was carried out by constructing artificial rectangular models with dimensions of length, width, and depth equal to 10.0, 1.0, 0.65 meter respectively that filled with dry clastic mixture deposit, three twin sets of air filled iron pipes of 15.24 cm (6 inch) diameter were buried horizontally and vertically inside the mixture at different distances together. Visual and Numerical interpretation were chosen to get the best results. In the visual interpretation, the amplitude variations show that the height of the positive peaks increases with the increase of the space distance between the buried pipes. Numerical interpretation appeared that the decrease in the width of the bands means an increase of the space between the pipes. The second part of the numerical analysis comprises measuring the amplitude value variation, among the signal forms; relying on the value of amplitude in each hyperbola the distinction process becomes quite easy. Depending on the variations in amplitude, the identification and discrimination of two closely spaced underground pipes will be feasible. The big values refer to highly spaced pipes while the low values denote the slightly spaced pipes. It is worth mentioning that the lowest value indicates the amplitude of only one buried iron pipe.
... Ground penetrating Radar, Ground probing or surface penetrating radar has been found to be an especially attractive option [4, 5]. The application of Ground Penetrating Radar in engineering, mining, and environmental geology studies is well documented and highly effective [6][7][8][9][10][11]. The new development of GPR techniques, acquisition, and processing used multi frequency antenna, ,complex 3D datasets, and computer programs to increase data collection and resolution over larger areas [12][13][14][15][16][17]. ...
The researchers have tried to focus on how to determine the number of pipes that are present in one obtained hyperbola in radargram profile. Ground Penetration Radar (GPR) survey was performed to distinguish between two zero-spaced iron pipes in radargram. The field work was carried out by constructing artificial rectangular models with dimensions of length, width, and depth equal to 10.0, 1.0, 0.65 meter respectively that filled with dry clastic mixture deposit, three twin sets of air filled iron pipes of 15.24 cm (6 inch) diameter were buried horizontally and vertically inside the mixture at different distances together. Visual and Numerical interpretation were chosen to get the best results. In the visual interpretation, the amplitude variations show that the height of the positive peaks increases with the increase of the space distance between the buried pipes. Numerical interpretation appeared that the decrease in the width of the bands means an increase of the space between the pipes. The second part of the numerical analysis comprises measuring the amplitude value variation, among the signal forms; relying on the value of amplitude in each hyperbola the distinction process becomes quite easy. Depending on the variations in amplitude, the identification and discrimination of two closely spaced underground pipes will be feasible. The big values refer to highly spaced pipes while the low values denote the slightly spaced pipes. It is worth mentioning that the lowest value indicates the amplitude of only one buried iron pipe.
... In contrast with [10] and [11] a single GPR trace was used in order to retain the applicability of the ANN to handheld devices that often do not offer accurate positioning information. B-Scans, after simple post-processing were used as inputs in [13], but again the training set was measured at a single minelane. In [14], a hybrid complexvalued ANN was employed in order to use both magnitude and phase information from eleven distinct frequencies (0.8-1.2 GHz) as inputs. ...
A numerical modelling case study is presented aim- ing to investigate aspects of the applicability of artificial neural networks (ANN) to the problem of landmine detection using ground penetrating radar (GPR). An essential requirement of ANN and machine learning in general, is a coherent training set. A good training set should include data from as many as possible scenarios. Therefore, a training set consisting of simulated data from a diverse range of models with varying topography, soil’s inhomogeneities, landmines, false alarm targets, heights of the antenna, depths of the landmines, has been produced and used. Previous approaches, have employed limited training sets and as a result they often have underestimated the capabilities of ANN. In this preliminary study, a 2D Finite-Difference Time-Domain (FDTD) model has been used as the training platform for ANN. Although a 2D approach is clearly a simplification that cannot directly translate to a practical application, it is a computationally efficient approach to examine the performance of ANN subject to a complete training set. The results are promising and provide a good basis to further expand this approach in the future by employing even more realistic, but computationally expensive, 3D models and well characterised real data sets.
... They consist in summing up intensities along A-scans within a C-scan, and thus, a surface 2-D map of potentially suspicious places is generated [1]. In [5], Bhuiyan and Nath worked with B-scans and use a seeded-growing approach for image segmentation and discovering places of interest. Marble [2] also considered B-scans and proposed a binary large objects detection technique. ...
... Overlapping regions of 3 × 3 cells constitute blocks for the normalization (33). 5 Since the extraction is repeated for B-scans (crossing the middle of the scanning window) both across and along track, the total number of features is twice the grid size times the number of bins: n = 2 · 4 · 3 · n θ . In other words, the full vector of HOG features is a concatenation of H(c, l) values for all cells, all bins, and two B-scan orientations. ...
... Training of the detectors (in each CV fold) was carried out by the RF algorithm, with 100 trees, as proposed in [21]. The trees in RF are not restricted in terms of splits or depth (contrarily to 5 In [21], the authors downsampled the image along t by the factor of 3. The downsampling helped to make the gradient magnitude distribution over angle bins closer to uniform, because their successive A-scans were taken 5 cm apart-horizontal image changes were more rapid than vertical ones. The downsampling step was redundant in our case. ...
This paper aimed to devise an efficient algorithm applicable to ground penetrating radar (GPR) and to enable an automatic landmine detection. Proposed is a machine learning approach in which we put the main emphasis on fast performance of the scanning procedure analyzing the C-scans, i.e., 3-D images defined over the coordinate system, i.e., along track by across track by time, where the time axis can be associated with depth. The approach is based on our proposition of 3-D Haar-like features. Learning of the detector is carried out by boosted decision trees. Practical experiments on metal and plastic antitank mines in a garden soil are carried out. A prototype mobile platform is designed to scan the subsurface of the ground, equipped with a GPR based on a standard vector network analyzer and our original antenna system. We report the results, particularly the following: detection sensitivity, false alarm rates, receiver operating characteristic curves, and times of learning and detection.
