11436

X-ray CT on the GPU

Ravikiran Tadepalli
Iowa State University
Iowa State University, 2013

@article{tadepalli2013x,

   title={X-ray CT on the GPU},

   author={Tadepalli, Ravikiran},

   year={2013}

}

Download Download (PDF)   View View   Source Source   

669

views

Nondestructive testing (NDT) is a collection of analysis techniques used by scientists and technologists as a way of analyzing the interior of an object without damaging the object. Since the analysis is done without damaging the object, NDT is an extremely valuable technique used in various industries for troubleshooting and research. CNDE has a long history of working with a variety of industrial sectors which include Aerospace (commercial and military aviation) and Defense Systems (ground vehicles and personnel protection); Energy (nuclear, wind, fossil); Infrastructure and Transportation (bridges, roadways, dams, levees); and Petro-Chemical (offshore, processing, fuel transport piping) to provide cost-effective tools and solutions. X-ray tomography is the procedure of using X-rays for generating tomographic slices of the required object. The object is bombarded with X-rays and the scanned image intensity values are collected on a detector. A significant drawback in X-ray tomography is the amount of data collected. It is generally huge in the order of gigabytes and hence the processing of data presents a big challenge. One way to speed up the processing of data is to run the programs on a cluster. CNDE uses a 64 node Beowulf cluster to do the reconstruction of an image. However with the advent of the GPU (Graphic Processing Unit) we have a far more cost efficient and time efficient hardware to run the reconstruction algorithm. The GPU can be fitted into a single PC, costs 10 times less than the cluster and also has a longer life time. This thesis has two major components to it. One of it is the development of new preprocessing and post processing techniques (includes filters, hot pixel removal etc.) to improve the quality of the input data and the other is the implementation of these techniques as well as the reconstruction program on the GPU using CUDA. Speedup on the GPU is not just a matter of porting the developed algorithms in parallel onto the hardware like in a cluster. GPU architecture is extremely complex and involves the usage of many different types of memory each with its own advantages and disadvantages and also many other optimization techniques for accessing and processing the data. These new techniques as well as the introduction of GPU are a significant addition to X-ray program here at CNDE.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2017 hgpu.org

All rights belong to the respective authors

Contact us: