A Topology Structure Repair Algorithm
for Triangular Mesh Model
Quan Wang, Pengfei Yang*, Yuanyuan Jiang, Ling Huang, Geyi Zhang
School of Computer, Xidian University
Xi’an, China
*Corresponding Author e-mail: pfyang@xidian.edu.cn
Abstract—Triangular mesh model built by existing surface
reconstruction methods still have issues such as dislocation, serial
hole, stitching and multilateral collinear. The model existing
above-mentioned defects cannot generate standard STL file for
3D printing. To solve these problems, this paper propose a
topology structure repair method for triangular mesh model,
which containing four types detailed solutions targeting different
topological error. Moreover, in order to improve the effectiveness
and superiority, we design a global error connected region and
correlative data structure, which can achieve the aim of rapid
detection and repair for different topology error. The experiment
results demonstrate that the method is effective in repairing
irrational triangular mesh models.
Keywords—topology structure repair; triangular mesh; surface
reconstruction; hole filling
I. INTRODUCTION
Surface reconstruction based on point cloud data is an
important research topic in reverse engineering, which has
been widely applied in industry, computer vision and graphics
etc.[1][2][3]. However, holes and topology errors, due to a
variety of reasons, are usually present in models built from
surface reconstruction. Some holes and topology errors are
caused by the intrinsic limitations of the sensors or existing
surface reconstruction algorithms. Others are the result of
object self-occlusion, insufficient view coverage, and shallow
grazing angles. The triangular mesh model generated by
existing surface reconstruction methods have serious topology
errors that cannot satisfy the requirements of the standard STL
file. And the models with topology errors cannot be directly
applied to the 3D printer, which limit the rapid development of
3D printing [4][5].
In this paper, we present a repair method based on the
topology information of triangular facets, which can be
applied to the mesh model built from surface reconstruction.
Moreover, we also propose the concept of the global error
connected region, aiming to optimize the detection efficiency
of our method. After detecting and repairing serious topology
structure errors, including dislocation, serial hole, stitching
crack and multilateral collinear, the method generates standard
STL file which can be directly applied to 3D printing. In
addition, our method has good reusability which can be
combined with existing surface reconstruction methods, and
fast and precisely repair topology error issues that cannot be
solved in the surface reconstruction methods or the self-repair
surface reconstruction methods.
The organization of the paper is as follows: In section
Ⅱ,we give a brief overview of existing methods. In Section Ⅲ,
we introduce the construction method of topology structure
and present the concept of global error connected region. In
Section Ⅳ, we propose our method framework that repair four
different topology errors for mesh model. Section V evaluates
the proposed method with contrast experiment and contains a
detailed parameter analysis. Finally, the paper is concluded is
Section Ⅵ.
II. RELATED WORK
The crucial issues of surface reconstruction is to ensure the
integrity of the data and the model. Many methods of repair
reconstruction model are proposed by scholars at home and
abroad recently. Existing methods of hole-filling and model
repair can be approximately divided into two classes: repair
method based on point cloud, and repair method based
triangular mesh model.
The repair methods based on point cloud [6][7][8]are
suitable for uniform distribution and small amount of point
cloud data, and it is difficult to obtain the ideal result for a
massive amount of input points.
Model repair methods based on triangular mesh [9][10]
directly process input mesh data and intuitively repair holes
and topology errors. Southern R et al. [11], who used Least
square mesh (LSM) model to reconstructed the geometry
structure of mesh by computing the Laplacian matrix of the
grid. But the method is based on the assumption that the
topological information of holes is obtained in advance. Qiu N
et al. [12] present an improved hole-filling algorithm by using
boundary projection on constrained Delaunay triangulation.
The method is applicable to reconstruct the missing part of
teeth and gums, but has accuracy problems in the case of
boundary grid normal rapid change. In [13][14], the authors
propose repairing methods based on region growth, and the
geometric properties of the hole feature region are used for
asynchronous growth. But the accuracy of the added vertices
remains to be improved. Radial basis function (RBF) neural
network method is used in paper[15] to improve the accuracy
and speed of the repair method, but it not works on the data
without training. In [16][17], these papers introduce a hole-
filling strategy for triangular meshes and at the same time
recover the sharp features of the original 3D model at the hole
areas. Liu et al. [18] achieves hole filling in the high curvature
region of the mesh model, through using the average curvature
of the hole neighborhood in the grid subdivision process. The
This paper is supported by National Natural Science Foundation of China
under Grant 61572385, Science & Technology Program of Shaanxi Province,
China (Grant NO.2015KTCXSF-01).