Suitability of a new alignment correction method for industrial CT

Exact knowledge of the geometrical configuration of an industrial CT scanner is essential for high quality CT image reconstruction. Geometrical misalignment can result in severe misalignment artifacts like blurring and the loss of spatial resolution. In computed tomography, redundantly measured rays (i.e. multiple measurements of a line integral through an object) pose a computationally efficient possibility to quantify the rawdata quality in a cost function and thus to reduce many kinds of artifacts. A general downside of such a cost function is that the portion of redundant rays is generally small and depends on the specific data acquisition geometry. The authors have previously proposed [1] to use information associated to plane integrals instead of line integrals in order to tremendously increase the rawdata utilization when formulating a cost function that quantifies the rawdata quality. However, so far the approach was evaluated on simulated data only. Therefore, based on a large database of industrial CT datasets acquired on a DAGE XD7600NT Diamond FP scanner, the suitability of this approach for industrial CT applications is now evaluated and proven.