Bayesian Mitigation of Sensor Position Errors to Improve Unexploded Ordnance Detection

: The research described in this report was conducted in fulfillment of Project MM-1572, A Complex Approach to UXO Discrimination: Combining Advanced EMI Forward and Statistical Signal Processing, submitted to the Strategic Environmental Research and Development Program (SERDP) in response to the Munitions Management Statement of Need MMSON-07-04, Advanced Technologies for Detection, Discrimination, and Remediation of Munitions and Explosives of Concern (MEC): UXO Technology. The well-known and prohibitive cost of carefully excavating all geophysical anomalies detected at lands contaminated with unexploded ordnance (UXO) is one of the greatest impediments to performing an efficient and thorough cleanup of former battlefields and of Department of Defense (DoD) and Department of Energy (DOE) sites. Innovative discrimination techniques are required that can quickly and reliably distinguish between hazardous UXO and non-hazardous metallic items. The key to success lies in the development of advanced processing techniques that can analyze and process sophisticated magnetic or electromagnetic induction data, with its novel waveforms, ever improving quality, and vector or tensor character, so as to maximize the probability of correct classification and minimize the falsealarm rate.

Many of the ongoing efforts to develop strategies for detecting and locating subsurface unexploded ordnance (UXO) use features based on phenomenological models to discriminate between UXO and harmless clutter. The process of generating features requires model inversion to fit the phenomenological model to the measured sensor data. In commonly-used model inversion processes, the standard measures of model fit error do not incorporate the spatial distribution of the data used in the model inversion. This study incorporates the Fisher information in a joint metric optimization to assess the spatial distribution of data and how well the model parameters are supported by the data used in the model inversion. The outcomes of this study indicate that some outliers in the feature space can be mitigated by considering the Fisher information in the model inversion process, resulting in improved unexploded ordnance detection rates in a test using data collected at Camp Sibert, Alabama.

Data processing in high energy physics experiments is a multi-tiered process in which raw detector signals are first processed locally into physics objects, and then collated into event records which can be scrutinized by a fast online trigger system. The resulting selection of events are reconstructed and pass through a number of software filters before arriving at a final offline analysis where hard physical constants are extracted. Although sophisticated statistical data analysis techniques are routinely employed high energy physics, the use of statistical signal processing in the field is has until now been rare. Our paper will begin with an overview of a typical high energy physics data acquisition system, outlining the technologies and tradeoffs involved at each stage. We will then proceed to argue that the dominant roles of model dependence and systematic errors in final physics analyses render statistical signal processing techniques largely inapplicable at this level. We observe, however, that at the low-level pattern recognition and event reconstruction levels, statistical signal processing techniques have been making inroads in high energy physics for a number of years, and examples from the literature will be cited. The viability of the technique for second level triggers will be assessed. Parallels to other other approaches, such as neural networks, will also be drawn. It will be argued that the falling cost of computing hardware favors the growth of statistical signal processing methods in high energy physics

Mobile, heterogeneous, and smart sensor networks produce pervasive structural health monitoring (SHM) information. With various embedded sensors, smartphones have emerged to innovate SHM by empowering citizens to serve as sensors. By default, smartphones meet the fundamental smart sensor criteria, thanks to the built-in processor, memory, wireless communication units and mobile operating system. SHM using smartphones, however, faces technical challenges due to citizen-induced uncertainties, undesired sensor-structure integration, and lack of control over the sensing platform. Previously, the authors presented successful applications of smartphone accelerometers for structural vibration measurement and proposed a monitoring framework under citizen-induced spatiotemporal uncertainties. This study aims at extending the capabilities of smartphone-based SHM with a special focus on the lack of control over the sensor (i.e., the phone) positioning by citizens resulting in unknown sensor orientations. Using smartphone gyroscope, accelerometer, and magnetometer; instantaneous sensor orientation can be obtained with respect to gravitational and magnetic north directions. Using these sensor data, mobile operating system frameworks return processed features such as attitude and heading that can be used to correct misaligned sensor signals. For this purpose, a coordinate transformation procedure is proposed and illustrated on a two-story laboratory structural model and real-scale bridges with various sensor positioning examples. The proposed method corrects the sensor signals by tracking their orientations and improves measurement accuracy. Moreover, knowing structure’s coordinate system a priori, even the data from arbitrarily positioned sensors can automatically be transformed to the structural coordinates. In addition, this paper also touches some secondary mobile and heterogeneous data issues including imperfect sampling and geolocation services. The coordinate system transformation methods proposed in this study can be implemented in other non-smartphone-based SHM systems as long as similar instrumentation is available.

This paper reports vehicle based electromagnetic induction (EMI) array sensor data inversion and discrimination results. Recent field studies show that EMI arrays, such as the Minelab Single Transmitter Multiple Receiver (STMR), and the Geophex GEM-5 EMI array, provide a fast and safe way to detect subsurface metallic targets such as landmines, unexploded ordnance (UXO) and buried explosives. The array sensors are flexible and easily adaptable for a variety of ground vehicles and mobile platforms, which makes them very attractive for safe and cost effective detection operations in many applications, including but not limited to explosive ordnance disposal and humanitarian UXO and demining missions. Most state-of-the-art EMI arrays measure the vertical or full vector field, or gradient tensor fields and utilize them for real-time threat detection based on threshold analysis. Real field practice shows that the threshold-level detection has high false alarms. One way to reduce these false alarms is to use EMI numerical techniques that are capable of inverting EMI array data in real time. In this work a physically complete model, known as the normalized volume/surface magnetic sources (NV/SMS) model is adapted to the vehicle-based EMI array, such as STMR and GEM-5, data. The NV/SMS model can be considered as a generalized volume or surface dipole model, which in a special limited case coincides with an infinitesimal dipole model approach. According to the NV/SMS model, an object's response to a sensor's primary field is modeled mathematically by a set of equivalent magnetic dipoles, distributed inside the object (i.e. NVMS) or over a surface surrounding the object (i.e. NSMS). The scattered magnetic field of the NSMS is identical to that produced by a set of interacting magnetic dipoles. The amplitudes of the magnetic dipoles are normalized to the primary magnetic field, relating induced magnetic dipole polarizability and the primary magnetic field. The magnitudes of the NSMS are determined directly by minimizing the difference between measured and modeled data for any known object and any type of EMI sensor data. The EMI array data are inverted via the combined NV/SMS and differential evolution inversion method that uses a search scheme to estimate the location of the target. First, the applicability of the NV/SMS and DE algorithms to STMR and GEM-5 data sets is demonstrated by comparing the modeled data against the actual data, and finally the discrimination studies are conducted using as discrimination parameters the total NV/SMS and the principal axis of the induced magnetic polarizability tensor for each target.

In this paper, an image fusion technique based on statistical signal processing is studied. Without the multiscale transform (MST) for the original images, the fusion is directly made via using Expectation-Maximization (EM) algorithm. Unlike the standard EM fusion algorithm, we only use the neighboring pixels in the window to estimate the centered pixel. The improvement of the iterative procedure is helpful to obtain the final fusion result effectively. At last, some performance evaluations are suggested and applied to compare the performance of these fusion methods. Consequently, the obtained results and comparisons indicate that the method here has significant advantage. Keywords-image fusion; statistical signal processing; EM algorithm

Single photon laser altimeter simulator and statistical signal processing

The authors propose three phenomenological models for wideband electromagnetic induction (EMI) response of buried conductors, such as unexploded ordnance (UXO) or metal parts in landmines. The models are based on analytic solutions for spheres, cylinders, and wire loops, and produce physically reasonable predictions for a variety of targets at all frequencies of interest including matches to theory in the low- and high-frequency limits. All three produce excellent fits to test data and run quickly enough to be of practical use in data inversion schemes. The authors present a three-parameter model capable of exactly matching permeable spheres and cylinders, a four-parameter version which adds the capability to match wire loops, and a five-parameter version which adds the capability to match signals due to driving bands, a feature found only on UXO. Driving bands, also called rotating bands, are soft metal rings near the tail of a projectile designed to engage rifles in the gun bore when the projectile Is fired. The author observe that driving bands produce a distinctive loop-like signal in EMI spectra, possibly because they are in the shape of a loop and typically have much greater conductivity than the body of the UXO. They demonstrate that the five-parameter model is capable of accurately fitting this signal and expressing its presence or absence through model fit parameters.

Monitoring the Comprehensive Test Ban Treaty (CTBT) to magnitude levels below 4.0 will require use of regional seismic data recorded at distances of less than 2000 km. To improve regional discriminant performance we tested three different methods of correcting for path effects, and the third method includes a correction for source-scaling. We used regional recordings of broadband from stations in and near China. Our first method removes trends between phase ratios and physical parameters associated with each event-station path. This approach requires knowledge of the physical parameters along an event-station path, such as topography, basin thickness, and crustal thickness. Our second approach is somewhat more empirical. We examine spatial distributions of phase amplitudes after subtracting event magnitude and correcting for path distance. For a given station, phase, and frequency band, we grid and then smooth the magnitude-corrected and distance-corrected amplitudes to create a map representing a correction surface. We reference these maps to correct phase amplitudes prior to forming discrimination ratios. Our third approach is the most complicated, but also the most rigorous. For a given station and phase, we invert the spectra of a number of well-recorded earthquakes for source and path parameters. We then use the values obtained from the inversion to correct phase amplitudes for the effects of source size, distance, and attenuation. Finally,the amplitude residuals are gridded and smoothed to create a correction surface representing secondary path effects. We find that simple ratio- parameter corrections can improve discrimination performance along some paths (such as Kazakh Test Site (KTS) to WMQ), but for other paths (such as Lop Nor to AAK) the corrections are not beneficial. Our second method, the empirical path correction surfaces, improves discrimination performance for Lop Nor to AAK paths. Our third method, combined source and path corrections, has only been tested at WMQ (primarily using KTS explosions and northwest China earthquakes), but the method successfully corrects for source-scaling and propagation effects. We favor this combined source and path correction technique over the other correction methods, and we plan to test this method at other Asian stations. 11 refs., 7 figs.

The effect on differential eyelid conditioning of 18 h of food deprivation was assessed on New Zealand White and American Dutch rabbits. Food deprivation did not increase the rate of acquisition of the conditioned response, nor did it increase general responsivity. Food deprivation did improve discrimination performance. The Dutch rabbits acquired the CR more rapidly than the albino rabbits and showed a greater responsivity in the conditioning situation. The albino rabbits showed superior discrimination performance.

Many mysteries related to the behavior of cancer stem cells (CSCs), their role in the formation of tumors, and the evolution of tumors with time remain unresolved. Biologists conduct experiments and collect data from them; these are then used for modeling, inference, and prediction of various unknowns that define the CSC system and are vital for its understanding. The aim of this chapter is to provide a summary of our progress in statistical signal processing models and methods for the advancement of the theory and understanding of cancer and the CSC paradigm. The chapter comprises three parts: model building, methods for the forward problem, and methods for the inverse problem.

2005 IEEE/SP 13th Workshop on Statistical Signal Processing - Book of Abstracts

A clinical decision support system (CDSS) is used as an aid in decision-making processes of health care providers in their day-to-day activities. This research attempts diagnosis of multiple diseases based on symptoms provided by patients. The work also recommends laboratory tests related to the predicted diseases and medications based on their results. The methodology adopted for implementation of CDSS is the Bayesian network approach. The modeling of the Bayesian network structure was undertaken in consultation with experts from medical domain. Clinical data has been used for estimation of network parameters such as conditional probability tables thereby bringing in machine learning into Bayesian methodology. The model developed is a learning model wherein the system input is saved for future training of the model. The results indicate that Bayesian approach is suitable for implementing a CDSS for multiple disease diagnosis. The proposed work will be useful towards increasing physicians throughput.

Gait analysis using foot-worn inertial measurement units has proven to be a reliable tool to diagnose and monitor many neurological and musculoskeletal indications. However, only few studies have investigated the robustness of such systems to changes in the sensor attachment and no consensus for suitable sensor positions exists in the research community. Specifically for unsupervised real-world measurements, understanding how the reliability of the monitoring system changes when the sensor is attached differently is from high importance. In these scenarios, placement variations are expected because of user error or personal preferences. In this manuscript, we present the largest study to date comparing different sensor positions and attachments. We recorded 9000 strides with motion-capture reference from 14 healthy participants with six synchronized sensors attached at each foot. Spatial gait parameters were calculated using a double-integration method and compared to the reference system. The results indicate that relevant differences in the accuracy of the stride length exists between the sensor positions. While the average error over multiple strides is comparable, single stride errors and variability parameters differ greatly. We further present a physics model and an analysis of the raw sensor data to understand the origin of the observed differences. This analysis indicates that a variety of attachment parameters can influence the systems’ performance. While this is only the starting point to understand and mitigate these types of errors, we conclude that sensor systems and algorithms must be reevaluated when the sensor position or attachment changes.

Recent Developments in Statistical Signal Processing