3-DIMENSIONAL FLAW RECONSTRUCTION AND DIMENSIONAL ANALYSIS USING A REAL-TIME X-RAY-IMAGING SYSTEM

We have implemented a three-dimensional point reconstruction method based on stereo radiography techniques utilizing a real-time radiography (RTR) laboratory. The continuous radiographic image coupled with computerized position control and feature point determination eliminates problems associated with film detectors (development time, occluded/crossed disparities due to missing data between views). Furthermore, the RTR laboratory geometry affords precise control over sample movements using a stepper motor-controlled positioner. We used a simple image-based calibration method to accurately determine the X-ray source and sample position relative to the detector, and 'pincushion' distortion inherent in the electrostatic inverting image intensifier tube was reduced to less than +/-0.5 pixels using a two-dimensional third-order polynomial-based image warp. These calibrations resulted in an overall depth accuracy of +/-0.2 mm, making this system competitive with computed tomography-based measurements. Moreover, this system achieves good results using two views of a sample rotated 10-30-degrees between views, so it works equally well with symmetric and plate-like geometries. The stereo radiography equations were generalized to support arbitrary sample manipulation (translation, rotation) and an arbitrary number of views using a fixed source/detector geometry. The overall system performance was determined by measuring lengths and orientations of simulated cracks (0.3 mm diameter copper wires) embedded at different depths within an aluminium plate, and by measuring the locations of coolant tube axes within a jet engine turbine blade sample.