Fusional laminography: A strategy for exact reconstruction on CL and CT information complementation

Abstract

Computed laminography (CL) is a non-destructive imaging technique for plate-like samples (PLS). Benefiting from the isotropic projections, CL provides clearer structure details compared to computed tomography (CT). However, the projections of CL suffer from the mapping defects in the Fourier domain, which makes the exact reconstruction theoretically impossible. This paper introduces the fusional laminography (FL) to achieve exact reconstruction, whose strategy is combining the Fourier information of CL and CT. In detail, the Fourier Slice Theorem is used to establish the mapping between the 2-D Fourier transform of the spatial projection and the 3-D Fourier transform of the objective function. The projections of CL and CT are respectively mapped with interpolation frameworks, achieving the completeness of 3-D Cartesian Fourier domain. Experimental results indicate that FL outperforms the latest iterative algorithm in terms of artifact removal and edge preserving. FL is particularly suitable for the testing of the plate-like samples with irregular structure, as it does not rely on any specific smoothing strategies or sparseness assumptions. Additionally, it demonstrates the possibility of facilitating information integrity through multiple scanning geometries.