LNLS, Campinas, Brazil
Three-dimensional image reconstruction in X-ray computed tomography (XRCT) is a mathematical process that entirely depends on the alignment of the object of study. Small variations in pitch and roll angles and translational shift between center of rotation and center of detector can cause large deviations in the captured sinogram, resulting in a degraded 3D image. Most of the popular reconstruction algorithms are based on previous adjustments of the sinogram ray offset before the reconstruction process. This work presents an automatic method for shift and angle adjust of the center of rotation (COR) before the beginning of the experiment removing the need of setting geometrical parameters to achieve a reliable reconstruction. This method correlates different projections using Scale Invariant Feature Transform algorithm (SIFT) to align the experimental setup with sub-pixel precision and fast convergence.
