Two Local FBP Algorithms for Helical Cone-beam Computed Tomography
Résumé
When dealing with X-ray inspection of large industrial objects, we are generally interested in the 3D reconstruction of a region-of-interest potentially containing flaws or other features rather than on the reconstruction of the whole object. This local reconstruction is related to a reduction in the field-of-view which consequently truncates the projections during data acquisition and produces severe artifacts when using the Feldkamp-Davis-Kress algorithm. To reduce these artifacts caused by transversely truncated projections, we can replace the global ramp filtering approach of the Feldkamp-Davis-Kress method into either the local first derivative operator combined with the Hilbert transform or the 1D Laplace operator followed by a non-local filtering step using a logarithmic kernel. In this paper, our contribution is the extension of the two algorithms from circular to helical scanning geometry. For performance evaluation of our implementations, we present a numerical comparison of the two algorithms with the standard helical FDK algorithm using both complete and truncated data generated by CIVA (the simulation platform for non-destructive testing techniques developed at CEA).
Origine : Fichiers produits par l'(les) auteur(s)
Loading...