Parallel Iterative Methods with A Priori Information for Robust Computed Laminography of Low Contrast, Difficult-To-Measure Objects

点击翻译按钮获取中文摘要

The goal of this project is to develop a novel 3D tomographic reconstruction method to make computed laminography (CL) a viable technique for non-destructive inspection scenarios, where standard computed tomography (CT) yields unsatisfying results, for example inspection of multilayered printed circuit boards, layer structure analysis of fibre-reinforced plastic, or inspection of large unwieldy components used in lightweight construction. To improve the reconstruction quality and performance every aspect of the pipeline from calibration of the scanner over the mathematical modelling to fast parallel algorithms on modern hardware has to be addressed. We focus on four key research areas: 1) consistent use of a priori information across the whole reconstruction pipeline, 2) optimization of basis functions and iterative convergence criteria, 3) reduction of the reconstruction time and resource usage by utilizing novel parallel hardware architectures, and 4) robust and efficient calibration of the data acquisition routines. Such a cross-disciplinary, whole-pipeline optimization requires the cooperation of researchers from the fields of mathematics, computer science, and engineering.