Development of concepts for high quality image reconstruction based on adaptive grid sizes

In this work an alternative data processing concept is investigated for the correct reconstruction of slice images acquired by an ultrafast electron beam X-ray tomography scanner mainly used for analyzing multiphase flows. Currently, image reconstruction is performed on regular pixel grids by filtered back projection to achieve rapid data processing performances but leading to non-optimal image qualities. To accomplish an improved image quality the usage of irregular reconstruction grids and iterative reconstruction methods is analyzed considering the geometric arrangement and, thus, the real spatial resolution of the ultrafast CT scanner. Finally, a two-stage reconstruction approach is proposed reducing the required amount of computer memory as well as the computational time and inserting prior knowledge about the object of interest. First simulations of different irregular grids provide a promising basis for further successful implementation of the proposed two-stage reconstruction concept.