Kenneth D. Morton

Histograms of Oriented Gradients for Landmine Detection in Ground-Penetrating Radar Data

Ground-penetrating radar (GPR) is a powerful and rapidly maturing technology for subsurface threat identification. However, sophisticated processing of GPR data is necessary to reduce false alarms due to naturally occurring subsurface clutter and soil distortions. Most currently fielded GPR-based landmine detection algorithms utilize feature extraction and statistical learning to develop robust classifiers capable of discriminating buried threats from inert subsurface structures. Analysis of these techniques indicates strong underlying similarities between efficient landmine detection algorithms and modern techniques for feature extraction in the computer vision literature. This paper explores the relationship between and application of one modern computer vision feature extraction technique, namely histogram of oriented gradients (HOG), to landmine detection in GPR data. The results presented indicate that HOG features provide a robust tool for target identification for both classification and prescreening and suggest that other techniques from computer vision might also be successfully applied to target detection in GPR data.

Laser-induced breakdown spectroscopy-based geochemical fingerprinting for the rapid analysis and discrimination of minerals: the example of garnet

Laser-induced breakdown spectroscopy (LIBS) is an analytical technique real-time geochemical analysis that is being developed for portable use outside of the laboratory. In this study, statistical signal processing and classification techniques were applied to single-shot, broadband LIBS spectra, comprising measured plasma light intensities between 200 and 960 nm, for a suite of 157 garnets of different composition from 92 locations worldwide. Partial least squares discriminant analysis was applied to sets of 25 LIBS spectra for each garnet sample and used to classify the garnet samples based on composition and geographic origin. Careful consideration was given to the cross-validation procedure to ensure that the classification algorithm is robust to unseen data. The results indicate that broadband LIBS analysis can be used to discriminate garnets of different composition and has the potential to discern geographic origin.

Multiple-Instance Hidden Markov Model for GPR-Based Landmine Detection

Hidden Markov models (HMMs) have previously been successfully applied to subsurface threat detection using ground penetrating radar (GPR) data. However, parameter estimation in most HMM-based landmine detection approaches is difficult since object locations are typically well known for the 2-D coordinates on the Earths surface but are not well known for object depths underneath the ground/time of arrival in a GPR A-scan. As a result, in a standard expectation maximization HMM (EM-HMM), all depths corresponding to a particular alarm location may be labeled as target sequences although the characteristics of data from different depths are substantially different. In this paper, an alternate HMM approach is developed using a multiple-instance learning (MIL) framework that considers an unordered set of HMM sequences at a particular alarm location, where the set of sequences is defined as positive if at least one of the sequences is a target sequence; otherwise, the set is defined as negative. Using the MIL framework, a collection of these sets (bags), along with their labels is used to train the target and nontarget HMMs simultaneously. The model parameters are inferred using variational Bayes, making the model tractable and computationally efficient. Experimental results on two synthetic and two landmine data sets show that the proposed approach performs better than a standard EM-HMM.

Target Classification and Identification Using Sparse Model Representations of Frequency-Domain Electromagnetic Induction Sensor Data

Frequency-domain electromagnetic induction (EMI) sensors can measure object-specific signatures that can be used to discriminate landmines from harmless clutter. In a model-based signal processing paradigm, the object signatures can often be decomposed into a weighted sum of parameterized basis functions, such as the discrete spectrum of relaxation frequencies (DSRF), where the basis functions are intrinsic to the object under consideration and the associated weights are a function of the target-sensor orientation. The basis function parameters can then be used as features for classifying the target. One of the challenges associated with effectively utilizing a model-based signal processing paradigm such as this is determining the correct model order for the measured data, as the number of basis functions containing fundamental information regarding the target under consideration is not known a priori. In this paper, sparse Bayesian relevance vector machine (RVM) regression is applied to simultaneously determine both the number of parameterized basis functions and their relative contributions to the measured signal assuming a DSRF signal model. The target is then classified utilizing the basis function parameters as features within a statistical classifier. Results for data measured with a prototype frequency-domain EMI sensor at a standardized test site are presented, and indicate that RVM regression followed by distance-based statistical classifiers utilizing the resulting model-based features provides an effective approach for classifying and identifying landmine targets.

Context-dependent landmine detection with ground-penetrating radar using a Hidden Markov Context Model

Context-dependent approaches to landmine detection have been developed in recent years to exploit the sensitivity of ground-penetrating radar (GPR) to changes in environmental conditions. Previous approaches to context-dependent fusion have only considered the special case of statistically independent observations. This work proposes the use of Hidden Markov Models, trained on the GPR background, for modeling the context of observation sequences. The performances of context-dependent fusion using two statistical context models were compared in an experiment with field data. One approach utilized a Hidden Markov Context Model (HMCM), and the other utilized a Gaussian mixture. Experimental results illustrated that the HMCM improved performance of context-dependent fusion. These results suggest that spatial dependencies are an important source of contextual information for landmine detection that warrants further investigation.

Bayesian Context-Dependent Learning for Anomaly Classification in Hyperspectral Imagery

Many remote sensing applications involve the classification of anomalous responses as either objects of interest or clutter. This paper addresses the problem of anomaly classification in hyperspectral imagery (HSI) and focuses on robustly detecting disturbed earth in the long-wave infrared (LWIR) spectrum. Although disturbed earth yields a distinct LWIR signature that distinguishes it from the background, its distribution relative to clutter may vary over different environmental contexts. In this paper, a generic Bayesian framework is proposed for training context-dependent classification rules from wide-area airborne LWIR imagery. The proposed framework combines sparse classification models with either supervised or discriminative context identification to pool information across contexts and improve classification overall. Experiments are performed with data from a LWIR landmine detection system. Contexts are learned from endmember abundances extracted from the background near each detected anomaly. Classification performance is compared with single-classifier approaches using the same information as well as other baseline anomaly detectors from the literature. Results indicate that utilizing context for classifying anomalies in HSI could lead to more robust performance over varying terrain.

Mandarin Chinese tone identification in cochlear implants: Predictions from acoustic models

It has been established that current cochlear implants do not supply adequate spectral information for perception of tonal languages. Comprehension of a tonal language, such as Mandarin Chinese, requires recognition of lexical tones. New strategies of cochlear stimulation such as variable stimulation rate and current steering may provide the means of delivering more spectral information and thus may provide the auditory fine-structure required for tone recognition. Several cochlear implant signal processing strategies are examined in this study, the continuous interleaved sampling (CIS) algorithm, the frequency amplitude modulation encoding (FAME) algorithm, and the multiple carrier frequency algorithm (MCFA). These strategies provide different types and amounts of spectral information. Pattern recognition techniques can be applied to data from Mandarin Chinese tone recognition tasks using acoustic models as a means of testing the abilities of these algorithms to transmit the changes in fundamental frequency indicative of the four lexical tones. The ability of processed Mandarin Chinese tones to be correctly classified may predict trends in the effectiveness of different signal processing algorithms in cochlear implants. The proposed techniques can predict trends in performance of the signal processing techniques in quiet conditions but fail to do so in noise.

A hidden Markov context model for GPR-based landmine detection incorporating stick-breaking priors

In recent years, context-dependent algorithm fusion has been proposed for improving landmine detection with ground-penetrating radar (GPR) across changing environmental and operating conditions. While context-dependent fusion techniques generally assume independent observations, previous work showed that spatial information may be exploited by modeling context with a hidden Markov model (HMM). However, the degree of performance improvement was found to depend the number of states included in the HMM. In this work, stick-breaking priors were employed to automate learning of the number of HMM states, and therefore the number of contexts to consider. The improved spatially-dependent fusion technique was evaluated on GPR data collected over various targets at multiple test sites, and performance was compared to another context-dependent technique which assumed independent observations. Results illustrate the potential for nonparametric, spatially-dependent context modeling to exploit contextual information in sequentially-collected GPR data and improve overall classification performance.

Variational Bayesian Learning for Mixture Autoregressive Models With Uncertain-Order

Autoregressive (AR) models are fundamental tools for modeling a variety of signals in many fields of study. Selecting the appropriate order for AR models is typically accomplished using an information criterion to compare the models learned for all orders under consideration. The use of an information criterion for model order selection becomes increasingly computationally demanding when AR models are used as part of larger statistical models, such as mixture models, that have their own model order selection issues. Statistical models utilizing Dirichlet process (DP) priors provide a mechanism for automatically selecting the number of components within a mixture model, and have previously been utilized with AR components. These previous investigations utilize different priors for the AR parameters to enable automatic selection of the AR order and each makes us of computationally expensive Markov chain Monte Carlo (MCMC) sampling. This paper develops and evaluates a variational Bayesian (VB) inference procedure for the parameters of DP mixtures of AR components with uncertain order, to enable rapid parameter inference for AR based statistical models that provide automatic model order selection and is suitable for large scale problems. The previously utilized priors for AR models with uncertain order are evaluated to determine which is more appropriate for VB inference and the ability of the VB inference procedure for the developed model to correctly determine the number of mixture components and the AR order of each of the mixture components is shown to be comparable to computationally intensive MCMC inference. The VB inference procedure is then applied to an acoustic signal classification problem to illustrate the efficacy of AR based statistical models utilizing automated model order selection for real-world signal processing tasks.

Comparison of a distance-based likelihood ratio test and k-nearest neighbor classification methods

Several studies of the k-nearest neighbor (KNN) classifier have proposed the use of non-uniform weighting on the k neighbors. It has been suggested that the distance to each neighbor can be used to calculate the individual weights in a weighted KNN approach; however, a consensus has not yet been reached on the best method or framework for calculating weights using the distances. In this paper, a distance likelihood ratio test was discussed and evaluated using simulated data. The distance likelihood ratio test (DLRT) shares several characteristics with the distance-weighted k-nearest neighbor methods but approaches the use of distance from a different perspective. Results illustrate the ability of the distance likelihood ratio test to approximate the likelihood ratio and compare the DLRT to two other k-neighborhood classification rules that utilize distance-weighting. The DLRT performs favorably in comparisons of the classification performance using the simulated data and provides an alternative non-parametric classification method for consideration when designing a distance-weighted KNN classification rule.