Drawing curves onto a cloud of points for point-based modelling Phillip N. Azariadis a,b, * , Nickolas S. Sapidis a a Department of Product & Systems Design Engineering, University of the Aegean, Ermoupolis, Syros 84100, Greece b ELKEDE—Technology & Design Centre SA, Research & Technology Department, 14452 Metamorphosis, Greece Received 18 December 2003; received in revised form 30 April 2004; accepted 4 May 2004 Abstract Point-based geometric models are gaining popularity in both the computer graphics and CAD fields. A related design/modelling problem is the focus of the reported research: drawing curves onto digital surfaces represented by clouds of points. The problem is analyzed and solved, and a set of ‘design tools’ are proposed which allow the user/designer to efficiently perform ‘product development’ (alternative name: ‘detail design’) tasks which require efficient processing of a ‘digital surface’. The primary tool is a robust and efficient point projection algorithm combined with a smoothing technique for producing smooth ‘digital curves’ lying onto the cloud surface. The new design tools are tested on a real-life industrial example with very satisfactory results, which are thoroughly presented in the paper. q 2004 Elsevier Ltd. All rights reserved. Keywords: Digital curves; Digital surfaces; Point-based representation; Point projection algorithm; Polylines; Smoothing polylines 1. Introduction During the last 5–6 years, an increasing trend has been developing in Computer-Graphics, as well as in the CAD community, towards surface models based on discrete elements like polygons, triangles and, very recently, points. Regarding polygons and triangles, these have long been the standard ‘basic element’ in graphics, and recently their importance has also been increasing in CAD/CAE/CAM. Indeed, modern design-technologies like Reverse Engineering (RE) [7], Virtual Engineering [15] and Rapid Prototyping [37] are using, almost exclusively, polygon meshes. ‘Collaborative design’ methods are also using meshes [28]; e.g. Ref. [36] contributes to network-based integrated design and reports that "faceted models have become the de facto standard model for describing complex geometry over the Internet". Finally, ‘concept development’ systems most often use a polygon mesh [28,30] rather than analytic geometric models. Very recently, the above trend, towards ‘discrete models’, has been culminating by bringing to the foreground the plainest of all geometric elements, the point! The reasons are very convincing: regarding modern computer graphics, ‘in complex models the triangle size is decreasing to pixel resolution’ [3], thus, points, and more specifically ‘connec- tivity-free points’, are the obvious choice for a primary surface model. Many recent papers (see Refs. [2,40] and references therein) adopt this approach, and develop efficient algorithms to solve related problems. For CAD, since the design process starts with points (user defined or imported using RE) and ends with points (VE/simulation/ analysis models or NC data), why use as primary model something else? This argument is developed in Ref. [10], which updates the related discussion first presented, long time ago, by McLaughlin [23]. Cripps [10] presents a comprehensive ‘point-based CAD’/CAM system for design/visualization and tool-path generation using subdivision-surfaces for model refinement, when this is needed. The pioneering work [16] reviews geometric modelling methods for NC machining and proposes point-based approximations as the most appro- priate. Vergeest et al. [31] focus on ‘shape reuse’ in modern design and proposes a novel methodology combining a primary CAD model with many auxiliary shape models in a corporate library. This family of shape models includes feature-, geometric-, CAD-, as well as point cloud models used as descriptors of library items. 0010-4485//$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.cad.2004.05.004 Computer-Aided Design 37 (2005) 109–122 www.elsevier.com/locate/cad * Corresponding author. Address: Department of Product and Systems Design Engineering, University of the Aegean, Ermoupolis, Syros, 84100, Greece. Tel.: þ30-228-109-7129; fax: þ30-228-109-7009. E-mail addresses: [email protected] (P.N. Azariadis), [email protected](N.S. Sapidis).
14
Embed
Drawing curves onto a cloud of points for point-based modelling
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Drawing curves onto a cloud of points for point-based modelling
Phillip N. Azariadisa,b,*, Nickolas S. Sapidisa
aDepartment of Product & Systems Design Engineering, University of the Aegean, Ermoupolis, Syros 84100, GreecebELKEDE—Technology & Design Centre SA, Research & Technology Department, 14452 Metamorphosis, Greece
Received 18 December 2003; received in revised form 30 April 2004; accepted 4 May 2004
Abstract
Point-based geometric models are gaining popularity in both the computer graphics and CAD fields. A related design/modelling problem is
the focus of the reported research: drawing curves onto digital surfaces represented by clouds of points. The problem is analyzed and solved,
and a set of ‘design tools’ are proposed which allow the user/designer to efficiently perform ‘product development’ (alternative name: ‘detail
design’) tasks which require efficient processing of a ‘digital surface’. The primary tool is a robust and efficient point projection algorithm
combined with a smoothing technique for producing smooth ‘digital curves’ lying onto the cloud surface. The new design tools are tested on
a real-life industrial example with very satisfactory results, which are thoroughly presented in the paper.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Digital curves; Digital surfaces; Point-based representation; Point projection algorithm; Polylines; Smoothing polylines
1. Introduction
During the last 5–6 years, an increasing trend has been
developing in Computer-Graphics, as well as in the CAD
community, towards surface models based on discrete
elements like polygons, triangles and, very recently, points.
Regarding polygons and triangles, these have long been the
standard ‘basic element’ in graphics, and recently their
importance has also been increasing in CAD/CAE/CAM.
Indeed, modern design-technologies like Reverse
Engineering (RE) [7], Virtual Engineering [15] and Rapid
Prototyping [37] are using, almost exclusively, polygon
meshes. ‘Collaborative design’ methods are also using
meshes [28]; e.g. Ref. [36] contributes to network-based
integrated design and reports that "faceted models have
become the de facto standard model for describing complex
geometry over the Internet". Finally, ‘concept development’
systems most often use a polygon mesh [28,30] rather than
analytic geometric models.
Very recently, the above trend, towards ‘discrete
models’, has been culminating by bringing to the foreground
the plainest of all geometric elements, the point! The reasons
are very convincing: regarding modern computer graphics,
‘in complex models the triangle size is decreasing to pixel
resolution’ [3], thus, points, and more specifically ‘connec-
tivity-free points’, are the obvious choice for a primary
surface model. Many recent papers (see Refs. [2,40] and
references therein) adopt this approach, and develop
efficient algorithms to solve related problems. For CAD,
since the design process starts with points (user defined or
imported using RE) and ends with points (VE/simulation/
analysis models or NC data), why use as primary model
something else? This argument is developed in Ref. [10],
which updates the related discussion first presented, long
time ago, by McLaughlin [23].
Cripps [10] presents a comprehensive ‘point-based
CAD’/CAM system for design/visualization and tool-path
generation using subdivision-surfaces for model refinement,
when this is needed. The pioneering work [16] reviews
geometric modelling methods for NC machining and
proposes point-based approximations as the most appro-
priate. Vergeest et al. [31] focus on ‘shape reuse’ in modern
design and proposes a novel methodology combining a
primary CAD model with many auxiliary shape models in a
corporate library. This family of shape models includes
feature-, geometric-, CAD-, as well as point cloud models
used as descriptors of library items.
0010-4485//$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cad.2004.05.004
Computer-Aided Design 37 (2005) 109–122
www.elsevier.com/locate/cad
* Corresponding author. Address: Department of Product and Systems
Design Engineering, University of the Aegean, Ermoupolis, Syros, 84100,
In RE, current systems allow acquisition/processing of
clouds of points so dense that triangulation is meaningless,
in complete analogy to computer-graphics applications.
Some very recent examples of publications, adopting this
approach, are: Woo et al. [34] proposing a new
segmentation method combining unorganized points with
an octree. Benko et al. [6] enhance the standard reverse-
engineering problem with constraints; the whole develop-
ment is based on point clouds. Corbo et al. [11] combine
up to 50 views of objects, measuring 1.2 m £ 0.6 m £ 2 m,
producing clouds with millions of points. This work
proposes novel ‘design analysis tools’ solely based on
point clouds.
Point-based models, enhanced with atomic-level physics,
lead to the so-called ‘particle systems’, which are
extensively used in discrete mechanics models for virtual
claying (see Refs. [17,20] and references therein) competing
against standard methods based on finite elements or
boundary elements. Horvath and his co-workers [15,29]
enhance point-based models and propose ‘vague discrete
modelling’, capable of simultaneously describing multiple
objects, to support concept development in collaborative
virtual environments.
Concluding this short review of ‘point-based surface
modelling’, we note that similar proposals for describ-
ing volumes/solids are constantly appearing, with typical
examples the various ‘volumetric representations’, like
voxels [8], and the ray- and triple-ray representations
[9,24,25].
1.1. Drawing curves on clouds of points
The present paper deals with product design using a point
cloud and focuses on drawing curves on (more accurately:
within) a ‘thick’ or ‘thin’ cloud of points. The core of the
proposed product design approach is shown in Fig. 1:
initially the user constructs a rough shoe design composed
of polylines which only look like they lie onto the cloud
surface (e.g. of a shoe last). The system eventually replaces
these polylines by smoothed ones lying exactly onto the
cloud surface (according to a specific mathematical model)
as it is shown in Fig. 1c. The transformation of the initial
design (Fig. 1a) to the final smooth one (Fig. 1c) requires the
elimination of intermediate side effects, like the line
wrinkles shown in Fig. 1b.
Designing curves onto a cloud of points is a surprisingly
difficult problem, and although no publication on it has been
yet identified, few recent papers deal with 2D versions of the
problem and clearly show the related difficulties. Benko and
Varady [7] deal with segmentation of thin, and thus
triangulated, points, yet curve-fitting to thick 2D clouds of
points does emerge, and heuristics are used to thin a cloud
before curve fitting. Curve-fitting to a thick 2D point set is a
vital subproblem in Ref. [26], too. Again, 2D curve-fitting is
preceded by thinning, based on related tools from image
processing. Polygonal curve-fitting to a 3D point cloud
appears in Ref. [21]; this is heuristically treated by solving
the corresponding 2D problem, which is a reasonable
approach given the paper’s subject (rapid prototyping).
Fig. 1. Product design using a cloud of points. (a) The user defines design polylines in one or more views, (b) the system projects the design polylines onto the
cloud, (c) the system calculates projected smooth cubic B-splines which are finally described as digital curves, i.e., dense polylines.
and smoothing a discretization of a line segment. With
respect to Note 2, we conclude to an alternative definition
for a smooth projection:
Definition 4 (Smooth projection). Given a line segment qi
represented by a set pi ¼ {pi;k ¼ kpi;k;ni;kllk ¼ 0;…;kðiÞ21}
of kðiÞ distinct nodes pi;k; the smooth projection of qi onto
CN is a p-polyline qpi ¼ {pp
i;k} minimizing the energy
function
Ei ¼ ð1 2 gÞPi þ gLi; g [ ½0; 1�; ð13Þ
where
Pi ¼XkðiÞ21
k¼1
Eðppi;kÞ ð14Þ
and
Li ¼XkðiÞ22
k¼0
kppi;k 2 pp
i;kþ1k2
ð15Þ
expresses the length of the p-segment qpi :
Each ppi;k is defined by pp
i;k ¼ pi;k þ tkni;k (compare with
Eq. (2)), thus, the whole p-polyline qpi is defined by the
vector t ¼ ðtkÞ ðk ¼ 1;…; kðiÞ2 1Þ holding the unknown
parameters tk: Eq. (13) is a convex combination of two
functionals, and it is minimized when the parameter vector t
is the solution of the linear system
½ð1 2 gÞI þ gA�t ¼ ð1 2 gÞb þ gc; ð16Þ
Fig. 7. (a) A basic shoe design sketched onto the cloud surface of a last ðN ¼ 6000Þ utilizing d-polylines. (b) The projection of polylines onto the cloud surface
established that curve design on point clouds is not as
complicated as its counterpart in standard ‘continuous
CAD’ systems.
Our approach for designing polynomial-spline
curves onto cloud surfaces uses an interpolation method
[27, Section 9.3.4] to construct a smooth spline of j-degree
pjðuÞ interpolating user-defined vertices of a d-polyline.
Using the same scheme, it is also possible to calculate the
corresponding projection direction njðuÞ: A node-set pj of
distinct nodes is derived from pjðuÞ which is eventually
projected onto the cloud surface. The final result is a
polyline tracing a smooth trajectory onto the cloud of points.
The discretization of pjðuÞ can be achieved using the
method in Ref. [35], which ensures that the topology of the
resulting linear approximation is consistent with that of
the initial curve.
Projecting pj onto the cloud of points is a straightforward
application of either the PointProjection or the Smooth-
SegProjection algorithm provided that at least the two
boundary vertices of pj are fixed. Two alternative
definitions for a digital curve are proposed:
Definition 5 (Digital curve). A digital curve is defined as
the result of the projection of the node-set pj onto CN
according to Definition 1.
Definition 6 (Smooth digital curve). A smooth digital
curve is defined as the result of the smooth projection of the
node-set pj onto CN according to Definition 4.
An example of designing digital curves is shown in
Fig. 10a. The wrinkling effect is present again especially in
the girth curve in the forepart of the shoe last. This problem
is resolved utilizing smooth digital curves with very
satisfactory results as shown in Fig. 10b (see also Fig. 1c).
Fig. 10. Two approaches to construct digital curves onto a cloud surface. The initial d-polylines have been replaced by pointsets derived from the original cubic
B-splines. (a) Digital curves designed onto the cloud surface without smoothing ðj ¼ 3Þ: (b) Smooth Digital Curves designed onto the cloud surface (j ¼ 3;
g ¼ 0:5).
Fig. 11. Application of the proposed methods for designing wearing apparel