Title:
|
RECONSTRUCTION OF GEOMETRICALLY PROCESSED SURFACE FROM COMPUTED TOMOGRAPHY DATA |
Author(s):
|
Mikhail Novozhilov and Mariya Dubrovskaya |
ISBN:
|
978-989-8533-79-1 |
Editors:
|
Katherine Blashki and Yingcai Xiao |
Year:
|
2018 |
Edition:
|
Single |
Keywords:
|
Surface Reconstruction, Marching Cubes, CT-scans, Segmentation, Graph |
Type:
|
Full Paper |
First Page:
|
221 |
Last Page:
|
228 |
Language:
|
English |
Cover:
|
|
Full Contents:
|
click to dowload
|
Paper Abstract:
|
Within this work we consider a problem of isosurface polygonizing by it`s reconstruction from scalar field, obtained from computed tomography data. Proposed method allows to get polygonal surface, convenient for further geometric processing. Firstly, for the problem solving, we build a minimal (this allows to reduce ambiguities) graph, nodes of which represent target surface`s vertices. Secondly, we triangulate the graph by it`s faces traversal. Thirdly, during geometric postprocessing we process the graph in order to improve the surface quality (resolving ambiguities, surface smoothing, holes triangulation). Declared problem can also be solved by Marching Cubes algorithm. However, our approach allows to obtain surface of higher quality and less time spent. Similarly to Marching Cubes algorithm, in our approach we test a condition that is aimed to check for presence of the surface between two nodes with known scalar field values. In our heart CT-scans experiments we examined various forms of our conditions and reconstructed more surfaces than we did using classical Marching Cubes implementation. In addition, our method is fully automatic and doesn`t demand usage of supercomputers, but runs normally on desktop computers. Finally, our method is parallelized at all stages and possesses linear acceleration, relative to the number of processors, that is confirmed by theoretical estimates and experiments. |
|
|
|
|