Clutter Objects Research Articles

Keypoint based autonomous registration of terrestrial laser point-clouds

Registration of terrestrial laser scanning point-clouds involves handling huge datasets, irregular point distribution, multiple views, and relatively low textured surfaces. Common approaches for the registration are based in large on the iterative closest point (ICP) model, which is non-linear and requires good initial values to secure convergence to the actual solution. Computation of the initial approximations to launch the ICP algorithm requires the extraction of distinct features, devising methods for matching them among the datasets, and then computing the transformation parameters. In this paper we present a computational approach for the registration problem. In essence we exploit 3D rigid-body transformation invariant features to reduce significantly the computational load involved in the matching between key features. Generally, with the partial overlap among datasets and among the extracted features the identification of corresponding key features can be viewed as a sub-graph matching problem. This problem is hard to solve, but as the actual matched entities are subjected to a six parameters transformation it becomes manageable. We show that distances, which are invariant to rigid body transformation, can be applied for solving this problem. We then show that by using selected keypoints, the matching process can be optimized. We also show how the information embedded within the range data is utilized to improve the quality of the selected points. Following the presentation of our algorithm, we demonstrate its application on a sequence of scans taken in areas featuring a clutter of objects. Results and the analysis show its efficiency and robustness.

An experiment to investigate aural classification of coherent-source active sonar data

Active sonar performance is degraded by the presence of environmental clutter objects, which lead to false alarm echoes. Earlier research suggests that perceptual signal features similar to those employed in the human auditory system can be used to automatically discriminate between impulsive‐source target and clutter echoes, thereby improving sonar performance by reducing the number of false alarms [V.W. Young et al., Can. Acoust. 34(3), 50–51 (2006)]. The present work attempts to extend these findings to the realm of coherent‐source active sonar echoes. To this end, an experiment was conducted on the Malta Plateau using a cardioid towed‐array and broadband FM sweeps running from 500 Hz up to 3.5 kHz. The data set consists of hundreds of pulse‐compressed echoes from several targets, shipwrecks, and clutter objects. It includes both monostatic and bistatic echoes, as well as echoes recorded at different aspect angles and at different propagation ranges between roughly 5 km and 30 km. In this paper the experiment is reviewed and preliminary results are presented for an automatic classifier that employs features based on a simple auditory model.

Using Triplet Arrays for Broadband Reverberation Analysis and Inversions

Some directional characteristics of observed clutter and reverberation are presented using line arrays with cardioid elements. It is shown that the cardioid arrays break the left/right ambiguity for reverberation sources above ~600-700 Hz. Data analysis uses a cardioid beamforming algorithm developed by researchers at the NATO Undersea Research Centre (NURC, La Spezia, Italy) but the normalization for this algorithm is derived by the author. It is shown that this normalized algorithm has an upper frequency limitation not previously noted. Reverberation data are taken from two recent experiments. One was the 2004 Boundary Characterization Experiment near the Malta Plateau. The area is rich in clutter objects like wrecks and mud-volcanoes and has some subbottom features that may be important. Sources were monostatic coherent pulses and impulsive sources. The receiver was the NURC cardioid array. The other experiment was the U.S. Office of Naval Research's 2003 Geoclutter effort to study shallow-water bottom reverberation and clutter in the STRATAFORM area off New Jersey. Sources were bistatic coherent pulses. The receiver was the five-octave research array (FORA). The STRATAFORM has benign surface morphology but strong clutter is observed. Some highlights of the reverberant returns from that area are discussed that include returns from probable fish schools. From these two data sets, it is shown that up to 15-20-dB left/right discrimination is attainable on broadside cardioid beams in the 1200-1750-Hz region for both linear frequency modulation (LFM) pulses and sound-underwater-source (SUS) broadband data. Also, a (previously developed) constrained simulated annealing (SA) inversion technique is applied to the directional reverberation data from the NURC array. A two-layer fluid model of the bottom is assumed. The algorithm is best at estimating compressional speeds, layer thickness, and attenuations in that order. Extracted bottom parameters at the site are compared with independent inversion results previously reported by Holland. Inverting reverberation from directional arrays is shown to give somewhat better quality results since, unlike standard line arrays, the reverberation data apply to only one bearing along the seafloor.

Independent component analysis for UXO detection in highly cluttered environments

Statistical signal processing techniques have shown progress in discriminating UXO from clutter when the objects occur in isolation. Under this condition, only a single object contributes to the measured sensor data. For multiple closely spaced subsurface objects, however, the unprocessed sensor measurement is a mixture of the responses from several objects. Consequently, the unprocessed measurements cannot be used directly to discriminate UXO from clutter. In this paper, we implement independent component analysis (ICA), a well-established blind source separation (BSS) technique, to recover the unobserved object signatures from the mixed measurement data obtained by simulating electromagnetic induction (EMI) sensor data, and then use the recovered signatures for UXO/clutter discrimination. Discrimination performance depends on multiple factors, including the number of clutter objects in proximity to the UXO, the separation distance between the UXO and clutter, and the number of mixed measurements available. Simulation results are presented illustrating the impact of these factors on discrimination performance.

LIBS for landmine detection and discrimination

The concept of utilizing laser-induced breakdown spectroscopy (LIBS) technology for landmine detection and discrimination has been evaluated using both laboratory LIBS and a prototype man-portable LIBS systems. LIBS spectra were collected for a suite of landmine casings, non-mine plastic materials, and "clutter-type" objects likely to be present in the soil of a conflict area or a former conflict area. Landmine casings examined included a broad selection of anti-personnel and anti-tank mines from different countries of manufacture. Other materials analyzed included rocks and soil, metal objects, cellulose materials, and different types of plastics. Two "blind" laboratory tests were conducted in which 100 broadband LIBS spectra were obtained for a mixed suite of landmine casings and clutter objects and compared with a previously-assembled spectral reference library. Using a linear correlation approach, "mine/no mine" determinations were correctly made for more than 90% of the samples in both tests. A similar test using a prototype man-portable LIBS system yielded an analogous result, validating the concept of using LIBS for landmine detection and discrimination.

Feature Extraction from GPR Data for Identification of Landmine-Like Objects Under Rough Ground Surface

Feature extraction from ground penetrating radar (GPR) data for identification of landmine-like objects under rough ground surface is studied. The feature proposed here is a time interval between two pulses reflected from top and bottom sides of landmine-like objects. Since the time interval is closely related to the thickness and permittivity of the objects, it is expected that this feature is suitable for target identification. Monte Carlo simulations using data set generated by an FDTD method show that good identification performance is obtained and that this feature is suitable for discrimination of landmines from confusing clutter objects.

Spatial distribution model for tracking extended objects

A Bayesian filter has been developed for tracking an extended object in clutter based on two simple axioms: (i) the numbers of received target and clutter measurements in a frame are Poisson distributed (so several measurements may originate from the target) and (ii) target extent is modelled by a spatial probability distribution and each target-related measurement is an independent ‘random draw’ from this spatial distribution (convolved with a sensor model). Diffuse spatial models of target extent are of particular interest. This model is especially suitable for a particle filter implementation, and examples are presented for a Gaussian mixture model and for a uniform stick target convolved with a Gaussian error. A rather restrictive special case that admits a solution in the form of a multiple hypothesis Kalman filter is also discussed and demonstrated.

Processing of reverberation data from 400–3500 Hz using line arrays with left/right discrimination

The author recently participated in two experiments using directional towed arrays. One was the 2004 boundary characterization experiment near the Malta Plateau. That experiment was led by the NATO Undersea Research Centre (NURC). The area is rich in clutter objects like wrecks and mudvolcanoes and has some sub-bottom features that may be important. Sources were monostatic coherent pulses and SUS. The receiver was the NURC cardioid array. The other experiment was ONR’s 2003 Geoclutter effort to study shallow-water bottom reverberation and clutter in the STRATAFORM off New Jersey. That experiment was led by M.I.T. Sources were bistatic coherent pulses. The receiver was the five-octave research array (FORA). The STRATAFORM is known to have benign surface morphology but strong clutter is observed. Some highlights of the reverberant returns are discussed that include the correlation of returns with manmade targets and probable fish schools. The purpose of this work is to assess the directional characterization of the observed clutter and reverberation. The cardioid arrays should yield good directional estimates of reverberation sources above ∼1600−1800 Hz. Examples from the data analysis are presented using a cardioid beamforming algorithm developed by NURC. [Work supported by ONR Code 32, Grant N00014-03-1-0113.]

Investigation of a combined seismic, radar, and induction sensor for landmine detection

An experimental system to collect co-located ground penetrating radar (GPR), electromagnetic induction (EMI), and seismic data was developed to investigate the benefits of the fusion of these sensors. In the experiments, a range of mines and clutter objects were buried at various depths in the sandbox at Georgia Tech. Multiple burial scenarios were investigated with a variety of antipersonnel and antitank mines and typical clutter objects. The seismic system used in these experiments is an extension of our existing seismic mine detection system. The system uses electrodynamic shakers to generate a seismic wave which propagates across the simulated minefield, and a specially designed radar is used to measure the displacement of the surface caused by the seismic wave. The GPR makes use of modified resistive-vee antennas and operates over the frequency range of 500 MHz to 8 GHz. These antennas are very clean in that they have very little self-clutter and very low radar cross section to lessen the reflections between the ground and the antennas. The EMI sensor collects broadband data (300 Hz to 100 kHz) so that the relaxation frequencies of the buried targets can be used to aid discrimination. [Work supported by ARO.]

Combining magnitude and phase features in acoustic landmine detection

The utility of acoustic-to-seismic coupling systems for landmine detection has been clearly established. They have been shown to be able to detect very low metal content landmines that are difficult to detect for ground-penetrating radars. This technique measures the difference in ground vibration velocity of regions with and without buried landmines when subject to acoustic excitation. For most applications, only the magnitude of the surface velocity is used to construct recognition algorithms. We recently introduced phase-based features in the classification scheme, significantly reducing false alarms at given detection probabilities. Here the focus is on the analysis of ground velocity data collected in the time domain with a moving array of laser Doppler vibrometers. The processing techniques used to extract the magnitude and phase information are described, as well as examples to demonstrate how combined magnitude and phase features help improve the detection of buried mines with weak signatures and reduce false alarms. Simple mass-and-spring models are shown to be useful in understanding the observed features. Also analyzed are the phase signatures of a number of ground regions with buried man-made clutter objects and how they can be useful in separating these clutters from actual landmine regions.

Discrimination Mode Processing for EMI and GPR Sensors for Hand-Held Land Mine Detection

Signal processing algorithms for hand-held mine detection sensors are described. The goals of the algorithms are to provide alarms to a human operator indicating the likelihood of the presence of a buried mine. Two modes of operations are considered: search mode and discrimination mode. Search mode generates an initial detection at a suspected location and discrimination mode confirms that the suspected location contains a land mine. Search mode requires that the signal processing algorithm generate a detection confidence value immediately at the current sample location and no delay in producing an alarm confidence is tolerable. Search mode detection has a high false-alarm rate. Discrimination mode allows the operator to interrogate the entire suspected location to eliminate false alarms. It does not require that the signal processing algorithm produce an alarm confidence immediately for the current sample location, but rather allows the system to process all the data acquired over the region before producing an alarm. This paper proposes discrimination mode processing algorithms for metal detectors (MDs), or electromagnetic induction sensors (EMIs), ground-penetrating radars (GPRs), and their fusion. The MD discrimination mode algorithm employs a model-based approach and uses the target model parameters to discriminate between mines and clutter objects. The GPR discrimination mode algorithm uses the consistency of detection as well as the shape of the detection peaks over several sweeps to improve the discrimination accuracy. The performances of the proposed algorithms were examined on a dataset collected at a government test site, and performance was compared with baseline techniques. Experimental results showed that the proposed method can reduce the probability of false alarm by as much as 70% at a 100% correct detection rate and performed comparable to the best human operator on a blind test with data collected at approximately 1000 locations.

Image segmentation and similarity of color-texture objects

We aim for content-based image retrieval of textured objects in natural scenes under varying illumination and viewing conditions. To achieve this, image retrieval is based on matching feature distributions derived from color invariant gradients. To cope with object cluttering, region-based texture segmentation is applied on the target images prior to the actual image retrieval process. The retrieval scheme is empirically verified on color images taken from textured objects under different lighting conditions.

(Spectral) pattern recognition as a versatile tool towards automatic landmine detection: A new European approach

A mobile acousto-optical sensor (Laser Vibrometer) is being used for the detection and discrimination of buried landmines. Analysis of measurement data, obtained in a number of field tests, reveals that buried mines (anti-tank mines as well as anti personnel mines) can reliably be discriminated from nonlethal clutter objects, such as stones, wood, cans, etc. due to their individual shapes and spectral properties. Increasing operational demands on these acoustic soundings (speed, reliability, and spatial/spectral resolution) implies the introduction of new procedures and strategies to identify and suppress misleading features as well as improved correlation of useful data. These procedures will include improved techniques for real time: (a) processing of the spectrally resolved image of the soil surface as well as (b) spectral analysis of the on-target soil vibrations. Image processing is comprising background linearization and suppression, shape recognition and gradient detection. Spectral analysis is including the use of look-up tables for cross-correlation and identification of characteristic features. This paper is presenting an introduction to the detection technique, sounding strategies as well as subsequent spectral data evaluation toward automated target recognition.

Audible and visual representations of the signals from a seismic landmine detection system

A system using high frequency seismic waves to detect buried landmines is being investigated using both experimental and numerical models. The simplest detection cue for the localization of buried landmines has been found to be resonances of the trigger mechanism and overlying soil as these resonances create substantial localized surface displacements over the buried landmine in comparison to other buried objects such as rocks, sticks, and manmade clutter. The prototype system utilizes a noncontact radar sensor to detect surface–normal displacements and a remotely-located surface-contacting source to generate seismic waves in the ground. Processing of the measured data shows the location of mines both graphically and audibly, either of which could be used as inputs for automatic target recognition algorithms. Visual images created by a wavenumber-domain signal processing algorithm and audible representations will be compared for landmines and typical clutter objects. [Work supported by ONR and ARO.]

Anti-Personnel Land Mine Detection and Discrimination Using Acoustic Data

An acoustic-based land mine detection system was field-tested by the University of Mississippi with promising results. This system uses a Laser Doppler Vibrometer (LDV) to measure the velocity of the vibration at the surface of the soil induced by acoustic energy in various frequency bands. In this paper, automated methods for detecting and discriminating Anti-Personnel (AP) mines from clutter objects are presented. Pre-processing methods rely on nonlinear filters realized as Choquet integrals. These filters are robust to the non-Gaussian, impulse type noise in this type of data. Detection follows pre-processing and relies on adaptive thresholding over each frequency band and three-dimensional (3D) connected component analysis. Features are extracted from the 2D frequency slices of the 3D connected components. The features are adaptively aggregated over frequency and used for discriminant analysis. Experiments are performed using anti-personnel mines, clutter objects, and blank areas (background samples with no mines or clutter objects). The algorithm detects 92% of the mines for a wide range of parameters. For some threshold values, 100% of the mines are detected and 92% of the mines are classified as mines with no false alarms.

Ordnance/Clutter Discrimination Based on Target Eigenvalue Analysis

A procedure is described for deriving a quantitative measure of the “ordnance-likeness” of an unknown target from electromagnetic induction data, which thus can be used as a clutter discriminant. This parameter provides a measure of the axial symmetry and aspect ratio of the target and is formed by taking the ratio of the eigenvalues of the target's magnetic polarizability tensor. The eigenvalues are derived by measuring the matrix elements of the polarizability tensor from above-ground spatial data and then performing an eigenvalue decomposition of this tensor. The method is linear, and does not require a nonlinear parameter search. The eigenvalue ratio is smaller than one for objects that are symmetric and long (ordnance-like) and greater than one for objects that are flat and/or irregular in shape (clutter-like). This ratio is independent of the orientation of the target. The procedure is demonstrated using data derived from samples of metallic clutter and ordnance-like objects (i.e., cylinders, sections of pipe and actual ordnance) at different orientations.

Sensor Fusion for the Detection of Landmines

The detection of buried anti-personnel mines (APMs) is widely considered as a problem which may only be solved with a combination of two or more complementary sensors. We present processing and fusion results obtained from a multisensor data set, acquired with a pulse induction metal detector (MD), a pulsed ultra wide band ground penetrating radar (GPR) and a 3–5μ thermal infra-red (IR) camera. Various types of soils, clutter objects and burial conditions were recorded. Anti-personnel mines included minimum metal mines as well as mines with a significant metal content. We use a special projection to map a 3D GPR data cube, with time or depth as vertical co-ordinate, into a horizontal plane view 2D image. Object contours are then derived, based on an edge extraction method, followed by an automatic detection of circular shapes with a Hough-transform. In the association step, the stand-off IR image, the metal detector and GPR images and related detections are mapped onto a common cartesian grid on the ground surface. Detection results are fused on a decision level, using a Bayesian approach. Our results indicate that the GPR performance approximately matches that of the metal detector. With both sensors all metallic mines and around 60% of the minimum metal mines were detected. In the case of two false alarms per square meter combined detection probability clearly exceeds single sensor performance.

PicToSeek: combining color and shape invariant features for image retrieval

We aim at combining color and shape invariants for indexing and retrieving images. To this end, color models are proposed independent of the object geometry, object pose, and illumination. From these color models, color invariant edges are derived from which shape invariant features are computed. Computational methods are described to combine the color and shape invariants into a unified high-dimensional invariant feature set for discriminatory object retrieval. Experiments have been conducted on a database consisting of 500 images taken from multicolored man-made objects in real world scenes. From the theoretical and experimental results it is concluded that object retrieval based on composite color and shape invariant features provides excellent retrieval accuracy. Object retrieval based on color invariants provides very high retrieval accuracy whereas object retrieval based entirely on shape invariants yields poor discriminative power. Furthermore, the image retrieval scheme is highly robust to partial occlusion, object clutter and a change in the object's pose. Finally, the image retrieval scheme is integrated into the PicToSeek system on-line at http://www.wins.uva.nl/research/isis/PicToSeek/ for searching images on the World Wide Web.

Assessing Object Mastery in the Home

In this study, the causal relationship between introduced noise and object clutter on infant performance was tested empirically. The infants were studied in their own homes and they actively engaged with the test materials, demonstrating object mastery. Surprisingly, however, the noise and object clutter did not interfere with the quality of object mastery for these twelve month olds. The differences between home and laboratory testing, as well as the implications of such testing, are discussed in-depth.

Content-based image retrieval by viewpoint-invariant color indexing

We aim at content-based image retrieval by color information indexing robust against varying imaging conditions. To this end, we propose a new set of color features robust to a large change in viewpoint, object geometry and illumination. From the proposed set, various color features are selected to construct color pattern-cards for each image. Matching measurers are defined, expressing similarity between color pattern-cards, robust to a substantial amount of object occlusion and cluttering. Based on the color pattern-cards and matching measures, a hashing scheme is presented offering constant run-time image retrieval independent of the number of images in the image database. To evaluate accuracy of the image retrieval scheme, experiments have been conducted on a database consisting of 500 images taken from multicolored man-made objects in real world scenes. The results show that high image retrieval accuracy is achieved. Also, robustness is demonstrated against a change in viewing position, partial occlusion, and a substantial amount of object cluttering. Finally, the image retrieval scheme is integrated into the PicToSeek system on-line at http://www.wins.uva.nl/research/isis/PicToSeek/ for searching images on the World Wide Web.