Automated object Positioning from Ground Penetrating Radar Images

An automated procedure has been developed for the analysis of images generated by ground penetrating radar. In particular, the hyperbolic arcs produced, for example, as the probe is scanned across buried pipesor cables, are located and automatically analysed to deduce the depthand position of the buried object. There are two novel steps to this process, which is computationally much faster than the Synthetic Aperture Focusing Technique to which it is related. In the first step, the arc-like reflections from defects are located by evaluating the correlations between a series of range signals. In this way information fromthe full 'ringing' waveform of the radar pulse may be utilised, and it becomes possible to follow arcs ev en where they cross other arcs inthe image originating from other defects or from random electrical impedance fluctuations in the ground. The method thus mimics the eye in following the arcs from an assimilation of the patterns within a broad area of the image. In the second step the object positions corresponding to random pairs of points along the arc are analysed statisticallyto find the most likely position of the object. The method is demonstrated using real radar images.