Vincent Baltazart

Thin-Pavement Thickness Estimation Using GPR With High-Resolution and Superresolution Methods

In the field of civil engineering, sounding the top layer of carriageways, i.e., the pavement layer, is classically performed using standard ground-penetrating radar (GPR), whose resolution is bandwidth dependent. The layer thickness is deduced from both the time delays of backscattered echoes and the known dielectric constant of the medium. This paper focuses on superresolution and high-resolution techniques, which serve to improve the time resolution of GPR signals, and presents a parametric technique and five subspace methods, namely, estimation of signal parameters via rotational invariance techniques (ESPRIT), multiple-signal classification (MUSIC) algorithm, Min-Norm, and their polynomial versions root-MUSIC and root-Min-Norm. The performance of these algorithms will be compared in terms of resolution power as well as root-mean-square error on the estimated thickness. The paper also presents the results of computer tests and radar measurements in the far field.

Automatic Crack Detection on Two-Dimensional Pavement Images: An Algorithm Based on Minimal Path Selection

This paper proposes a new algorithm for automatic crack detection from 2D pavement images. It strongly relies on the localization of minimal paths within each image, a path being a series of neighboring pixels and its score being the sum of their intensities. The originality of the approach stems from the proposed way to select a set of minimal paths and the two postprocessing steps introduced to improve the quality of the detection. Such an approach is a natural way to take account of both the photometric and geometric characteristics of pavement images. An intensive validation is performed on both synthetic and real images (from five different acquisition systems), with comparisons to five existing methods. The proposed algorithm provides very robust and precise results in a wide range of situations, in a fully unsupervised manner, which is beyond the current state of the art.

Support Vector Regression method applied to thin pavement thickness estimation by GPR

In the field of civil engineering, sounding the layers is classically performed using standard ground-penetrating radar (GPR), whose vertical resolution is bandwidth dependent. The layer thicknesses are deduced from both the time delays of backscattered echoes and the dielectric constants of the layers. In contrast with the conventional spectral analysis approaches, we propose in this paper to use one of the most powerful machine learning algorithm, namely the Support Vector Machine(SVM), to perform Time Delay Estimation (TDE) of backscattered radar signals. In particular, this paper demonstrates the super time resolution capability of such technique in the context of overlapping and totally correlated echoes when thin pavement layers survey is under scope.

Generalization of PILE Method to the EM Scattering From Stratified Subsurface With Rough Interlayers: Application to the Detection of Debondings Within Pavement Structure

This paper presents the numerical method, generalized propagation-inside-layer expansion (GPILE), to calculate the scattered electromagnetic field by any stratified 1-D medium composed of three random rough interfaces separating homogeneous media. GPILE is a generalization of the propagation-inside-layer expansion method, which considers only two interfaces. Both methods rely on the rigorous implementation of the Maxwell equations, with a simple matrix formulation and which have a straightforward physical interpretation. In particular, this method allows us to distinguish the primary echo of the upper surface and also the multiple echoes arising from the intermediate and lower interfaces. This method is applied in this paper to simulate the ground-penetrating radar (GPR) signal at nadir. The simulated signals are analyzed to study the sensitivity of the GPR signal to any material debonding within the pavement layered structure.

Blind deconvolution via sparsity maximization applied to GPR data

ABSTRACT In measuring thin layer thickness of asphalt pavements using GPR, the sparse reflectivity series representing the layered structure of the pavement is convolved with the radar wavelet. This convolution results in masking closely spaced reflections. An ideal deconvolution retrieves the reflectivity series, and thus improves the time resolution and facilitates quantitative data interpretation. In this paper, we cast the convolutional model as a multidimensional data model which renders blind deconvolution via independent component analysis possible. We use a nonlinearity related to the double exponential density whose heavy-tailed nature provides further insight into the sparse nature of the reflectivity series. The method is tested on synthetic and real GPR data from a thin PVC slab. The results show the accuracy of the time delay estimates and verify the high resolution capability of the proposed approach.

Assessment of statistical-based clutter reduction techniques on ground-coupled GPR data for the detection of buried objects in soils

A bi-static Ground Penetrating Radar (GPR) has been developed for the detection of cracks and buried pipes in urban grounds. It is made of two shielded Ultra Wide Band (UWB) bowtie-slot antennas operating in the frequency band [0.3;4] GHz. GPR signals contain not only responses of targets, but also unwanted effects from antenna coupling in air and in the soil, system ringing, and soil reflections that can mask the proper detection of useful information. Thus, it appears necessary to propose and assess several clutter reduction techniques as pre-processing techniques to improve the signal-to-noise ratio, discriminate overlapping responses issued from the targets and the clutter, and ease the use of data processing algorithms for target detection, identification or reconstruction. In this work, we have evaluated on Bscan profiles three different statistical data analysis such as mean subtraction, Principal Component Analysis (PCA), and Independent Component Analysis (ICA) considering a shallow and a medium depth target. The receiver operating characteristics (ROC) graph has allowed to evaluate the performance of each data processing in simulations and measurements to further draw a comparison in order to select the technique most adapted to a given soil structure with its radar probing system.

A new A-star algorithm adapted to the semi-automatic detection of cracks within grey level pavement images

The detection of cracking on the road surface is an important issue in many countries to insure the maintenance and the monitoring of the roadways. This paper proposes a method which adapts the single pair shortest path A* algorithm to the detection of cracks within pavement images. The proposed A* algorithm computes the crack skeleton by calculating the minimal path between a pair of pixels which belong to the crack structure. Compared with the widespread and ubiquitous Dijkstra’s algorithm and to its bidirectional version, the proposed A* reduces the amount of the visited pixels; it is thus about 4 times faster than Dijkstra while keeping a large similarity coefficient with the ground truth.

Progress in monitoring the debonding within pavement structures during accelerated pavement testing on the fatigue carousel

The paper gives an overview of the ongoing experiment to survey debonding areas within pavement structure during accelerated pavement testing on the Ifsttar’s fatigue carrousel. Several defects have been embedded during the construction phase and have been probed by different NDT techniques. Among them, the paper focuses on radar NDT&E techniques which have been used to detect and locate the artificial defects at the early stage of the experiment and then to perform the survey at different loading cycles. Besides, the test-site has been used to test different GPR materials, including the 3D GPR technology. The experiment has also motivated some related studies in the field of GPR data modelling and data processing, which are summarized in the paper. The experiment is expected to be pursued beyond 300 loading kcycles to reach larger pavement degradations.

Subspace Leakage Suppression for Joint Parameter Estimation of Quality Factors and Time Delays in Dispersive Media

Linear prediction methods, based on a Hankel data matrix, suffer from subspace leakage and degraded resolution when applied to data models that do not result in a mode matrix with Vandermonde structure, such as the constant-Q model. In the absence of noise, the Vandermonde structure ensures the equivalence between the number of backscattered signals and the rank of the data matrix. This paper first identifies the origin of subspace leakage residing in subspace-based and linear prediction methods when applied to data of the constant-Q model. Second, it proposes a frequency-distortion technique, based on the extension theorems, for suppressing the leakage and preserving the time resolution performance of these methods. The effectiveness of the distortion technique is then demonstrated on GPR simulated data by extending the damped MUSIC algorithm to the joint parameter estimation of the constant-Q model.

Evaluation of an UWB ground-coupled radar in the detection of discontinuities using polarization diversity: FDTD modeling and experiments

An UWB ground-coupled radar has been designed to operate from 460 MHz to beyond 4 GHz and essentially for civil engineering applications. Full-wave modeling using the FDTD approach has allowed to study in details the antenna radiation characteristics in air, in the presence of a soil and as a constituent in a bistatic GPR system. The polarization diversity in the E and H-planes is an important aspect which has been studied in order to further detect the orientation of damages (cracks or delaminations) in civil engineering structures. The analysis of the hyperbola signatures has allowed to evaluate the ability of the radar to detect small canonical buried objects.