A SYNTHETIC-APERTURE ALGORITHM FOR GROUND-PENETRATING RADAR IMAGING

ABSTRACT: A formulation for ground-penetrating radar (GPR) imag-ing using the synthetic-aperture concept is introduced. We show that itis possible to form a 3D image by inverse Fourier transforming the mul-tifrequency, multispatial scattered field. The proposed algorithm forGPR imaging is tested with measured and simulated data. The resultantimages demonstrate good agreement between the measured and simu-lated cases. © 2004 Wiley Periodicals, Inc. Microwave Opt TechnolLett 42: 412–414, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.20320 Key words: ground penetrating radar (GPR); synthetic aperture radar(SAR); radar imaging 1. INTRODUCTION The imaging of buried objects or inhomogeneities undergroundusing ground-penetrating radar (GPR) has been a topic of interestfor a wide variety of applications ranging from mine detection toarcheology. Many GPR imaging algorithms have been proposed inthe literature [1–5]. Although good depth resolution can usually berealized in GPR images using frequency diversity, good resolutionin the cross-range dimensions is much harder to achieve. Capineriet al. [3] proposed a method for obtaining good resolution in GPRimages out of B-scan data by applying the Hough transformationtechnique. Morrow and Van Genderen [4] and Van Dongen et al.[5] applied the back-propagation and conjugate-gradient inversiontechniques to form two-dimensional (2D) and three-dimensional(3D) images for a borehole radar. However, these techniques havea significant computational burden. Therefore, there is a need forobtaining images with good range and cross-range resolution witha fast algorithm.We previously developed a synthetic-aperture algorithm forimaging antenna–platform interactions based on multifrequency,multispatial scattered-field data [6–8]. In this paper, we extend ouralgorithm to generate 3D GPR images of scattered data fromburied objects underground. This technique is based on the ap-proximate Fourier transform relationship between the frequency-spatial variables and the distance-angle information of the buriedscatterer. The algorithm is quite attractive, since it forms 3Dimages by using a fast Fourier transform (FFT) followed by asimple transformation from the distance-angle domain to the im-age domain. It is computationally fast. Furthermore, the cross-range resolution can be made as good as the range resolution bycontrolling the size of the collection aperture.