Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Point Cloud Types Page 1 of 3

Point Cloud Types

In order to efficiently operate on point data in a variety of forms, the Imageware software distinguishes four basic types
of point clouds based on the internal organization of the points. To determine the type of a cloud, use the
Evaluate|Information|Object menu entry or the Info icon on the Toolbar.

What would you like to know more about?

z Arbitrary point clouds

z Scan line point clouds

z Gridded point clouds

z Polygonized point clouds

z Uses for point cloud types

Arbitrary Point Clouds


Arbitrary clouds are multiple unorganized points, usually obtained from hand digitizing systems and some laser scanners.
These point clouds can be viewed in scatter or polyline modes. Scatter mode shows clouds as individual pixels, while
polyline mode draws lines between the points, connecting them in order.

Arbitrary (Scattered) Arbitrary (Polyline)

Scan Line Point Clouds


Scan Line point clouds are organized in consecutive, singular strings of points. Typically, the scan lines are parallel scans
and the planes for consecutive scans are equally spaced from one another. Scan line data is usually obtained from a
coordinate measuring machine or laser scanner. Multiple scan lines are multiple ordered strings of points in one set (as
shown below).

Multiple Scan Lines

Gridded Point Clouds


A point cloud is said to be gridded if all of its points lie on the vertices of a uniform grid in parameter space and there is

file://C:\SDRC\I-DEAS9\freeform\bin\help\how_to\point_processing\pnt_types.htm 11.6.2007
Point Cloud Types Page 2 of 3

exactly one point for every vertex on that grid. The points are arranged in an equal array (i.e. 100 points x 100 points).
Some of the application’s capabilities can be applied only to gridded point clouds. Multiple gridded clouds usually
overlap slightly.

Note: Data from many laser sensors and Moiré sensors are gridded data sets.

Grid Cloud Multiple Grid Clouds

Polygonized Point Clouds


A point cloud is just a collection of points in space. However, if these points represent the outer skin of some object, it’s
possible to connect these points to create a wrapped polygonal mesh. We refer to such structures as polygonized point
clouds. This application can create, store, and manipulate such structures. A polygonized point cloud is a cloud for which
a polygonal mesh exists, since the cloud data is attached mathematically to its closest neighbors, forming triangles called
facets. It can be read in from external sources or created internally. The mesh vertices are always a subset (possibly all)
of the points in the cloud.

Note: Polygonized point clouds are used for visualization, as well as a polygonal geometric representation of shapes. For
example, the polygonal representation could be used to generate STL files for rapid prototyping.

Polygonized Point Cloud (Polygon Mesh)

Point Cloud Type Uses


Some of the point cloud operations work for specific data types only. In most cases, the more organized the data is, the
more efficient the operations can be. This application provides commands for converting data from one type to another
by reordering, extrapolating, or extracting portions of data sets. Applying a command to a given point cloud may result in
point clouds of different types. The Imageware software always attempts to preserve the highest organization possible for
new or modified point clouds.

Tip: The "Guide to Commands" section of this help file mentions the types of clouds produced by the different
commands and warns about point cloud operations that may reduce the ordering (modify the point cloud type).

Some operations, such as free-form fitting of surfaces or z-gridding of point clouds, assume single-valued clouds. Single-
valued cloud in Z means that for any x and y value, there’s only one possible associated z value. In other words, imagine
replacing the points with an opaque sheet. Observe the sheet from a location far on the z axis (seeing the xy plane flat
ahead). If no portion of the sheet (point cloud) is obscured by other portions of the sheet, the cloud is single valued in z.
In general, a single-valued cloud is a cloud for which there exists some direction for which the cloud is single-valued.

file://C:\SDRC\I-DEAS9\freeform\bin\help\how_to\point_processing\pnt_types.htm 11.6.2007
Point Cloud Types Page 3 of 3

For example, a cloud representing the skin of a sphere is not single-valued.

An example of single valued and non-single-valued clouds in z follows:

To determine whether a cloud is single-valued in z, observe it in Standard View #1 in the viewport. This is the view of
the xy plane, down the z axis. Single-valued point clouds in z will not wrap around in this view.

file://C:\SDRC\I-DEAS9\freeform\bin\help\how_to\point_processing\pnt_types.htm 11.6.2007
Curve Types Page 1 of 3

Curve Types

The curves found in this product are B-splines, a particular kind of mathematical entity. B-splines are one type of spline
used to represent geometry. Bezier splines are another similar type - often encountered in computer graphics. While
several varieties of splines exist, B-splines have a broad range of capabilities and characteristics that make them popular.
The broadest form, Non-Uniform Rational B-splines (NURB) is used by Imageware modules to represent geometric
entities.

z Bezier curves

z NURB curves

z Curve continuity

z Comparison of both curve types

Bezier Curves
B-spline curves are actually built on the methods of another curve type: Bezier curves. Bezier curves are simple entities
that are easy to manipulate. The most commonly used Bezier curves are fully defined by four points: two endpoints and
two control points that do not lie on the curve itself, but define its shape.

A Bezier curve, or a single span curve, is composed by order and control points. This means that the order of the curve
reflects the number of control points. In order to make the curve shape more flexible, the order must be raised. Bezier
curves represent a specific family of basis functions, defined in terms of the element degree. Each higher degree
generates more complex basis functions and requires another control point. Every control point in a Bezier curve affects
the entire length of the curve. However, as the degree increases, a single control point plays a diminishing role in the
entire curve.

Bezier curves are important for two reasons:

z They're easy to compute, as they are based on fairly simple mathematical formulas. This allows them to be
readily implemented in hardware and software applications.

z They have predictable behavior. Due to their single span construction, Bezier curves tend to have good curvature
characteristics, which minimize inflections and irregularities in curvature flow. The first and last control points
determine the end locations of the curve – the internal shape being ‘suspended’ from the control frame. Their
predictable nature makes for controlled and logical editing since the control points act to ‘pull’ the curve to the
desired shape.

Despite their advantages, Bezier curves are limited. Their biggest limitation is a consequence of blending each control
point into the entire curve. As more points are used, the degree of the basis functions increases, taking longer to compute.
Plus, each new control point lessens the impact of any individual point on the curve. Complex shapes are generally
modeled as a series of lower-degree Bezier segments.

NURB Curves
B-splines were designed to address the limitations of Bezier curves while keeping their advantages. Spline curves are
piecewise parametric curves. This means that a splice curve is essentially a connected series of parametric curve
segments. B-splines may be thought of much like a series of connected Bezier segments. The B-spline basis functions are
very similar to their corresponding Bezier basis functions, except they don't use every control point at once. B-splines are
easy to compute, predictable, and have local control. In addition, any Bezier curve may be exactly represented by a B-
spline curve.

The broadest form of B-spline is the NURB curve. NURB, Non-Uniform Rational B-spline, is a mathematical
representation of 3-D geometry that can accurately describe any shape from a simple 2-D line, circle, arc, or curve to the
most complex 3-D organic free-form surface or solid. A NURB curve is simply a collection of single span curves that

file://C:\SDRC\I-DEAS9\freeform\bin\help\get_start\working_with_entities\crv_types.... 11.6.2007
Curve Types Page 2 of 3

are tied together with knots to form one entity. Because of their flexibility and accuracy, NURB models can be used in
any process from illustration and animation to manufacturing.

One of the advantages of NURB curves is that they offer a way to represent arbitrary shapes while maintaining
mathematical exactness and resolution independence. You can also use them for interpolating or approximating data or
as a tool to design and control the shapes of three dimensional surfaces for surfaces of revolution, trimming, and
extruding. Among their useful properties are the following:

z They can represent virtually any shape, from points, straight lines, and polylines to conic sections to freeform
curves with arbitrary shapes.

z They give you great control over the shape of a curve. A set of control points and knots, which guide the curve's
shape, can be directly manipulated to control its smoothness and curvature. You can define a curve with no kinks
or sudden changes in direction or you can make localized changes to the shape of the curve by moving individual
control points, without affecting the overall shape of the curve.

z They can represent very complex shapes with remarkably little data. For instance, approximating a circle three
feet across with a sequence of line segments would require tens of thousands of segments to make it look like a
circle instead of a polygon. Defining the same circle with a NURB representation takes only seven control
points.

Continuity
It's important to see how a curve relates to the same properties of neighboring elements. Typically, we want two elements
to have matching (or continuous) values of some sort where they meet. If the elements touch, there is positional
continuity; if they share the same tangent normal at their endpoints, there is tangent continuity; and if they have the
same curvature values (identical radius values), there is curvature continuity. Continuity is dictated by the order of the
curve.

Each continuity level proceeds the next. That is to say, if two curves don't touch (not positional matched), then they can’t
have tangent continuity. Equally, if they aren't tangent to each other, then logically, they can't have curvature continuity.

Order Continuity
Order 2 Position Continuity (C0)
Order 3 Tangent Continuity (C1)
Order 4 Curvature Continuity (C2)
Order >4 Curvature Continuity + a repeated knot (C3)

Curve Knots (NURB only)

To delineate the sections of a NURB curve, specific parameter values are assigned where the pieces will join. These
parameter values are known as knots, since they are located where the curve sections are tied together. As we pass from
one section to the next, a new control point is used, while another is no longer needed. The knot value specifies the
parameter value at which the change will take place.

Knots have an added advantage of achieving continuity. Each span (or section) is intimately associated with the spans on
either side of it, since they share most of their control points. Spans along a B-spline of a given degree can have an order
of continuity equal to one less than the degree of the curve. A cubic curve will therefore have curvature continuity
between each of its spans. If the spans were made from Bezier curves, they would have to be carefully constructed to
achieve this continuity.

Multiple knots occur when the curve has an order greater than four. Double knots are two knots on top of each other. A
double knot can cause curvature discontinuity across the knot. Triple knots are three knots on top of each other. A triple
knot can cause tangent discontinuity across the knot.

Number of Knots Continuity

file://C:\SDRC\I-DEAS9\freeform\bin\help\get_start\working_with_entities\crv_types.... 11.6.2007
Curve Types Page 3 of 3

Single Knot Curvature Continuity


Double Knot Tangent Continuity
Triple Knot Position Continuity

Comparison
The similarities and differences between NURB and Bezier entities are summarized in the table below. It is
recommended that the Bezier method be used when creating surfaces that will be edited, since editing a NURB surface
can result in poor surface quality (controlling the curvature flow across NURB entities demands a high level of user
understanding. With the greater flexibility that NURB offers, comes the increased scope for user error). However, when
using ‘auto-creation’ commands (that is to say, computer generated, as opposed to user derived entities), such as fillets
and flanges, it may be preferable to use NURB creation since it can result in potentially fewer surfaces of higher
accuracy.

Bezier NURB
Order Yes Yes
Control Points Control points are equal to the order Control points may be greater than or equal to the
order. If control points are greater than the order,
then there are knots in the geometry.
Knots No Yes
Advantages z Simple and easy to edit z Greater flexibility for complex shapes

z Clean curvature z Can create continuous complex shapes


characteristics
Disadvantages z Multiple entities required for z Difficult to achieve Curvature flow
complex shapes
z Require high level of user understanding
z Lengthy creation for complex
shapes
When to Use Surfaces that will be edited Auto-creation tools (fillets, flanges etc.)

file://C:\SDRC\I-DEAS9\freeform\bin\help\get_start\working_with_entities\crv_types.... 11.6.2007
Surface types Page 1 of 2

Surface types

B-Spline Surfaces
The methods used to represent and evaluate B-spline curves can be easily extended to describe surfaces as well. Any
position on a parametric curve may be referenced in terms of the variable u, In this sense, the curve is one-dimensional
(the u dimension). Another parameter, v, is added to describe B-spline surfaces. B- spline surfaces, therefore may be seen
as having two parametric dimensions, u and v. Just as a position on the Earth’s surface may be specified by latitude and
longitude, u and v parameter values can describe all positions on a B-spline surface.

Because u and v are parameters, they will have minimum and maximum values. The surface edges are found at these
extreme parameter values. The interior of the surface is completely described by all uv coordinates between these
extremes. Many systems also allow “trimming” of surfaces by specifying additional uv curves which limit the surface to
specific parametric regions.

If there are m control points to describe u and n to describe v, each of the surface’s control points are part of an m x n
array of points. These points contain and describe the surface just as their one-dimensional counterparts describe a curve.
Each point has local control over a region of the surface, just as with curves.

Like curves, the regions of a B-spline surface are defined by knots. Each region (or patch) is bounded by u and v values.
Many systems draw curves across the surface at knot values. These curves are isoparameter curves, which show the
curve of a single parameter value (the knot) across a surface. This special kind of isoparameter curve is also called a flow
line. Therefore, patches are singular surface regions bounded by flow lines, controlled by a selected set of control points.

The normal to a surface is the direction perpendicular to it, like “up” and “down” are relative to our position on the Earth.
The tangent plane to a surface is “level” with that surface (and perpendicular to the normal). Surface normals and
tangents are local information about surfaces, so they vary with uv position.

Surface curvature is more complex than with curves, because curvature at any point on a surface can be evaluated in u, v,
or in some combination. To understand this, consider a saddle. From side to side, the saddle curves away from you; from
front to back, the saddle curves toward you. Somewhere in between, the two cancel each other out. Surface curvature is
more complex, but still an everyday concept. Several ways of quantifying and categorizing curvature exist for
performing analysis.

For C0 continuity, the above is true. For G0, this need not be the case. The edges will be positionally continuous even if
the knots and control points do not match.

For C1 continuity, the last pairs of control points will be aligned. For G1, this need not be the case. G1, called tangent-
plane continuity, requires that the normals of the two surfaces at the edges be in the same direction. Tangent-plane
continuity is illustrated below:

If two surfaces must be curvature continuous (G2/C2), the conditions for tangent continuity must be met, plus the next

file://C:\SDRC\I-DEAS9\freeform\bin\help\get_start\working_with_entities\surfaces.htm 11.6.2007
Surface types Page 2 of 2

rows of control points on either side must be in just the right place. For cubic B-spline surfaces, curvature continuity is
the highest order of continuity available. Internally, cubic B-spline surface patches are normally C2 continuous with each
other.

file://C:\SDRC\I-DEAS9\freeform\bin\help\get_start\working_with_entities\surfaces.htm 11.6.2007

You might also like