Automated Alignment of Aircraft Wing Radiography Images Using a Modified Rotation, Scale, and Translation Invariant Phase Correlation Algorithm Employing Local Entropy for Peak Detection
Non-destructive evaluation (NDE) of aircraft wings using radiography involves the individual acquisition of a large number of radiographic images, each covering a small section of the wing. The ability to automatically generate a wing mosaic from these individual images could be a valuable tool, especially given the large amount of time required to manually align a single wing. A registration algorithm using phase correlation and the Fourier-Mellin transform was adapted. Due to the nature of the images, the peak selection process performed during phase correlation needed to be modified. The local entropy of each point on the correlation surface was calculated, generating a new local entropy surface, where points of high entropy correspond to true peaks in the correlation surface. This method is currently successful in aligning 83% of the wing radiography images into an automatically generated mosaic, up from a 40%-50% success rate using the existing algorithm.
School:University of Cincinnati
School Location:USA - Ohio
Source Type:Master's Thesis
Keywords:nde image registration aircraft inspection entropy fourier techniques processing
Date of Publication:01/01/2008