|
3D X-Ray Computed Tomography (3DCT) is a well established method in medicine which recently gained more and more importance and acceptance in the examination and inspection of industrial specimens. Computed Tomography is a method for imaging specimens in three dimensions in order to detect and measure hidden errors, e.g. shrinkage pores, cracks, contaminations, pores, material voids, material changes etc., in the material. So 3DCT offers complete 3D volume information and therefore a unique insight into the inner structure of a specimen. Compared to medical CT, in industrial 3DCT the specimen is rotated on a rotary plate, taking a 1024² projection image at each angular position. These projection images are subsequently reconstructed to a 3D volumetric dataset using the filtered back projection reconstruction algorithm.
Our goal is to do the reconstruction process on the GPU in order to create a fast and efficient reconstruction method on high end graphic cards hardware. Concerning computation speed modern GPUs are outperforming even top of the line CPUs. Further more, every (half a) year a new generation of GPUs is published which doubles the performance and which can additionally be enhanced by usage of new technologies like SLI. As GPUs are getting increasingly flexible in terms of programming and also the equipped memory is sufficient, the interest in porting common algorithms to fast and comparably lower cost GPU hardware is increasing.
|
