Options
2002
Conference Paper
Titel
New approach for a fast reconstruction of limited data from x-ray tomography and laminography
Abstract
Laminographic measurement systems are widely used for the examination of flat components like e. g. printed circuit boards. In such systems, X-ray data are only gathered from a certain angular width perpendicular to the flat object plane. In contrast, computed tomography (CT) normally requires a complete data set because the object is inspected from all angular directions over 360°. In the case of large components or fixed installed objects this condition cannot always be satisfied and data from a certain angular region are missing. This leads to incomplete data sets which are common in reconstruction methods like laminography. Fast and commonly used CT reconstruction algorithms, like e. g. the filtered back-projection method, fail in reconstructing such incomplete data. Nevertheless there exist several other methods which allow the reconstruction of object slices or the object volume. These algorithms are very time consuming or have a low contrast resolution, restricting the possible field of applications. This problem can be surmounted by including a priori information about the object itself or the measurement procedure into an iterative algebraic reconstruction technique (ART). Using such a priori information the reconstruction time can be reduced up to a factor of hundred and the contrast resolution and the depth of focus is concurrently increased. This enabled us to examine flat and large objects with defects exhibiting small contrast, such as welding seams or plastic components within reasonable measurement and reconstruction times, and corresponding examples will be presented.