A technique is described that enables virtual reconstruction of three-dimensional images of steel reinforcing bars embedded within concrete. The method is based on the systematic characterization of an inductive sensor response, using appropriate curvilinear models; this characterization is then applied to real, two-dimensional scan data. The first phase involves formulating a mathematical description of the line scan response for a single bar, taken at several depths, using a Pearson VII model. The data in this case are termed the depth response image, with the critical properties of interest being the peak position, peak intensity, and full width at half height. The next phase requires application of this model to actual scan data, and a description of the sensor's depth response using a Bleasdale power-law regression model. In the final phase, different bar layers are isolated using a polynomial-based layer separation technique. Data provided by these means enable a three-dimensional image of the bar mesh to be constructed and visualized. Given a known bar size, the scan depth can be calculated to an accuracy of ± 2 mm or ± 5%, whichever is greater. When neither the depth nor the size is available, the method allows estimates to be made with accuracies of ± 1 bar size (DIN 488) and ±10% cover depth (BS1881).