Back to search

University of Ioannina

Algorithms and tools for deriving editable models from cross-sectional data sets

Abstract

dc:description

In this thesis, we present a novel method for reconstructing the surface of a 3D object using as input only a point cloud of its surface scan. The objective is to obtain an editable CAD model that is manufacturable and describes accurately the structure and topology of the point cloud. This model will provide a high level representation of the actual object, which can also be modified with the use of a set of editing operators we have defined. The entire method is performed in three phases: the segmentation of the point cloud, the reconstruction of the model and the editing process.In the segmentation phase, the point cloud is sliced into cross sections, which are later on treated as 2D point sets. Several parameters of the slicing procedure influence the resulting model, and have to be set properly to ensure high quality output. One essential parameter is the slicing direction. It has to correspond to the principal axis of the point cloud, either locally or globally, so that the basic features of the point cloud are properly distributed across the cross sections. Another key parameter in the segmentation process is the thickness of the cross sections. The cross sections should contain sufficient information about the features of the point cloud, so their thickness cannot be very slim. But it should not be very thick either, as we do not want to have many features tangled together in the same cross section.The reconstruction phase includes several procedures that take as input the cloud points of the cross sections and provide as output a model of the 3D object, with high level features and properties that can be used for further processing. First, the boundary of each cross section is represented as a set of line segments called the feature poly-line. This poly-line is computed with the use of computational geometry methods, i.e. the convex hull and the Voronoi diagram. The feature poly-line of each cross section is then recomputed as an interpolating B-Spline curve, providing a smooth continuous representation of the model. The curves of the neighboring cross sections are combined with each other to form the surface of the final 3D model. To reconstruct the surface between two adjacent cross sections, a new set of feature points is computed on the B-Spline curves, using arc length parameterization, and a contour triangulation method provides the surface of the model. The resulting model can be subject to high level modifications that provide variations of the initial object with additional user specified properties.For the editing process, we have implemented a set of editing operators, which can be applied either to the entire model, or on a part of it, to deform its surface in various ways. As a point of reference for these editing operators, a given point may be used on the interpolating curve computed from the centroids of the cross sections. We call it the curve of centroids and it is a form of skeleton curve for the model. The editing operators vary from general purpose transformations to high level editing operators addressed to cross sectional data sets used in medical research. The general purpose transformations can be used to modify free form models arbitrarily, and can be applied to parts of the model according to user specified parameters. The high level editing operators allow modifications such as bending or stretching the model of an artery, allowing medical experts to study and control the behavior of tissues during a simulation of a surgical operation such as angioplasty or stent insertion.The novelty of this study includes: (a) the introduction of an efficient method for computing a boundary representations of cross-sectional point sets, using the convex hull and the Voronoi diagram of the points, (b) the use of a technique for inserting artificial cross sections between existing slices, to improve the rendering quality of the model, and (c) the definition of a set of editing operators that may be used in everyday examples, varying from purely aesthetic purposes to simulation of surgical operations in medical data sets.

Degree

thesis:*
Grantor dc:publisher
University of Ioannina
Year dc:date
2017

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Kyriazis, Ioannis
  • Κυριαζής, Ιωάννης

Subjects

dc:subject × 16

Rights

dc:rights
Statement dc:rights
  • BY
Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/43269

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Kyriazis, Ioannis; Κυριαζής, Ιωάννης. Algorithms and tools for deriving editable models from cross-sectional data sets. University of Ioannina, 2017. http://hdl.handle.net/10442/hedi/43269