Iterative Region-of-Interest Reconstruction from Limited Data Using Prior Information
Iterative Region-of-Interest Reconstruction from Limited Data Using Prior Information
复制标题
DOI：
10.1007/s11220-017-0165-8
复制
发表时间：
2017-04
影响因子：
2.2
通讯作者：
Jonas Vogelgesang;C. Schorr
中科院分区：
文献类型：
--
作者：
Jonas Vogelgesang;C. Schorr
在实际应用中，计算机断层扫描和计算机层析成像存在数据不完整的问题。特别是，当检测纵向和横向直径差异极大的大型物体时，或者当需要高分辨率重建时，扫描系统的物理条件导致数据受限和投影截断，也称为内部或兴趣区域（ROI）问题。为了恢复被检测对象的搜索密度函数，我们推导了ROI问题的半离散模型，该模型本质上允许将几何先验信息合并到抽象Hilbert空间设置的有界线性算子中。假设物体内部的衰减近似为常数，对于有兴趣定位裂纹或孔隙等缺陷的纤维增强塑料零件或均匀物体，我们采用半离散Landweber-Kaczmarz方法从测量数据中恢复ROI内物体的内部结构，从而得到半离散迭代方法。最后，给出了具有固有限制源和ROI问题的三维层析成像应用的数值实验，验证了所提出的ROI重建方法。
In practice, computed tomography and computed laminography applications suffer from incomplete data. In particular, when inspecting large objects with extremely different diameters in longitudinal and transversal directions or when high resolution reconstructions are desired, the physical conditions of the scanning system lead to restricted data and truncated projections, also known as the interior or region-of-interest (ROI) problem. To recover the searched-for density function of the inspected object, we derive a semi-discrete model of the ROI problem that inherently allows the incorporation of geometrical prior information in an abstract Hilbert space setting for bounded linear operators. Assuming that the attenuation inside the object is approximately constant, as for fibre reinforced plastics parts or homogeneous objects where one is interested in locating defects like cracks or porosities, we apply the semi-discrete Landweber–Kaczmarz method to recover the inner structure of the object inside the ROI from the measured data resulting in a semi-discrete iteration method. Finally, numerical experiments for three-dimensional tomographic applications with both an inherent restricted source and ROI problem are provided to verify the proposed method for the ROI reconstruction.