The Performance of 3D Multi-slice Branched Surface Reconstruction on CPU-GPU Platform

Normi Abdul Hadi, Norma Alias

2019

Abstract

In this paper, a CPU-GPU algorithm to generate composite contour for 3D branching surface is presented. The composite contour is generated based on the data points from based and branched contours and located in between the two contours. Distance calculation is one of the processes in composite contour generation which consumes the most CPU time, therefore, this process is chosen to be executed on the GPU. The developed composite contour generation method on the CPU-GPU platform is then applied on CT images of Stanford bunny and human pelvic with three different number of curve points per segment. These samples generate 12 composite contours in total. The performance of the developed algorithm is measured based on the processing time and the speedup. The result shows that the CPU-GPU algorithm has improved the speedup as high as 150 times.

Download


Paper Citation


in Harvard Style

Hadi N. and Alias N. (2019). The Performance of 3D Multi-slice Branched Surface Reconstruction on CPU-GPU Platform.In Proceedings of the Second International Conference on Science, Engineering and Technology - Volume 1: ICoSET, ISBN 978-989-758-463-3, pages 49-54. DOI: 10.5220/0009092700490054


in Bibtex Style

@conference{icoset19,
author={Normi Abdul Hadi and Norma Alias},
title={The Performance of 3D Multi-slice Branched Surface Reconstruction on CPU-GPU Platform},
booktitle={Proceedings of the Second International Conference on Science, Engineering and Technology - Volume 1: ICoSET,},
year={2019},
pages={49-54},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0009092700490054},
isbn={978-989-758-463-3},
}


in EndNote Style

TY - CONF

JO - Proceedings of the Second International Conference on Science, Engineering and Technology - Volume 1: ICoSET,
TI - The Performance of 3D Multi-slice Branched Surface Reconstruction on CPU-GPU Platform
SN - 978-989-758-463-3
AU - Hadi N.
AU - Alias N.
PY - 2019
SP - 49
EP - 54
DO - 10.5220/0009092700490054