Construction of X-ray Microtomography for Cultural Relics Study
Tung-Ho Chen (陳東和)1*, Chien-Chi Huang (黃千奇)1
1Scientific Research and Analysis Laboratory, Department of Conservation, National Palace Museum, Taipei, Taiwan
* presenting author:Tung-Ho Chen,
X-ray computed tomography (CT) is an effective analytical technique for investigating non-destructively the 3D-structure of an object. Through the 3D-image reconstruction process, the inner structure and morphology of the examined object can be explored in detail. With the aim of studying different types of artifacts housed in the National Palace Museum (NPM), a versatile X-ray cone-beam micro-CT has recently been constructed. With more suitability and flexibility than most expensive commercial CT scanners which are normally designed for medical or industrial application under the constraint of analyzing specific materials in a limited dimension, this NPM homemade cost effective CT system, dedicated to relics investigation, with 11 controlling axes, permits of scrutinizing various kinds of materials, including ceramics, bronzes, lacquers, woods, etc., in different sizes.
Equipped with an open microfocus X-ray tube of 240kV, the micro-CT system contains two imaging sensors: a CsI scintillator-mirror-CCD detector and a flat panel detector. After processing the scanned 2D images by median filter and flat-field correction for better image quality, the 3D image reconstruction is processed using Feldkamp, Davis and Kress (FDK) algorithm and iterative algorithm, which provide tomograms allowing of exploring the 3D structure of relics. As the CT scanned images contribute to big data which may paralyze the computation, a new parallel dynamic computing technology – the compute unified device architecture (CUDA) with graphic processing unit (GPU) is thus employed for accelerating the CT image reconstruction process. In this work, several artifacts of the NPM are shown for demonstration of examining their inner structure through tomograms.

Keywords: microtomography, CT, image reconstruction, FDK algorithm, GPU, CUDA