Method for clearing of virtual representations of objects

0Associated
Cases 
0Associated
Defendants 
0Accused
Products 
0Forward
Citations 
0
Petitions 
1
Assignment
First Claim
1. Method for clearing, in particular for removing, unwanted data from optically detected virtual representations of objects, in particular teeth and intraoral structures, the method comprising:
 a. Defining an extension line of the representation,b. Generating a projection plane at one point of the extension line, the extension line at this point being perpendicular to the generated projection plane,c. Projecting all known points in space of the representation from one region corresponding to the projection plane onto the projection plane, the corresponding point in space being stored for each projected point,d. Generating a twodimensional curve on the projection plane from the projected points,e. Determining maxima, minima and a center of the curve,f. Identifying projected points of the curve that—
viewed from the center of the curve—
lie outside of the minima or maxima,g. Removing the points in space that correspond to the projected points that were identified in f.,h. Optionally, repeating starting from b. for one further point of the extension line.
1 Assignment
0 Petitions
Accused Products
Abstract
A method for clearing unwanted data from optically detected virtual representations of objects includes: a. Defining an extension line of the representation; b. Generating a projection plane at one point of the extension line, which is perpendicular to the generated projection plane; c. Projecting all known points in space of the representation from one region on the projection plane onto the projection plane, the corresponding point in space being stored for each projected point; d. Generating a twodimensional curve on the projection plane; e. Determining maxima, minima and a center of the curve; f. Identifying projected points of the curve that—viewed from the center of the curve—lie outside of the minima or maxima; g. Removing the points in space that correspond to the projected points that were identified in step f.; and h. Optionally, repeating starting from step b. for one further point of the extension line.
16 Citations
No References
METHOD, APPARATUS, SIGNALS, AND MEDIA FOR PRODUCING A COMPUTER REPRESENTATION OF A THREEDIMENSIONAL SURFACE OF AN APPLIANCE FOR A LIVING BODY  
Patent #
US 20110115791A1
Filed 07/18/2008

Current Assignee
Vorum Research Corporation

Sponsoring Entity
Vorum Research Corporation

MEASUREMENT METHOD  
Patent #
US 20110184695A1
Filed 10/29/2009

Current Assignee
Renishaw PLC

Sponsoring Entity
Renishaw PLC

SYSTEM AND METHOD FOR LOCATING ANATOMIES OF INTEREST IN A 3D VOLUME  
Patent #
US 20090079738A1
Filed 09/24/2007

Current Assignee
Dolpin Imaging Systems LLC

Sponsoring Entity
Dolpin Imaging Systems LLC

Method and system for automatic classification of objects  
Patent #
US 20090184865A1
Filed 05/16/2006

Current Assignee
Telefonaktiebolaget LM Ericsson

Sponsoring Entity
Telefonaktiebolaget LM Ericsson

INSPECTION SYSTEM AND METHODS WITH AUTOCOMPENSATION FOR EDGE BREAK GAUGING ORIENTATION  
Patent #
US 20080281543A1
Filed 05/07/2007

Current Assignee
General Electric Company

Sponsoring Entity
General Electric Company

Method and system for enhanced orthodontic trearment plenning  
Patent #
US 20070072144A1
Filed 09/23/2006

Current Assignee
Orametrix Incorporated

Sponsoring Entity
Orametrix Incorporated

Automated centerline detection algorithm for colonlike 3d surfaces  
Patent #
US 20050117787A1
Filed 12/18/2002

Current Assignee
The United States of America represented by the Secretary of Health and Human Services

Sponsoring Entity
THE GOVENMENT OF THE UNITED STATES OF AMERICA AS REPRESENTED BY THE SECERETARY OF THE DEPARTMENT

MANIPULATING A DIGITAL DENTITION MODEL TO FORM MODELS OF INDIVIDUAL DENTITION COMPONENTS  
Patent #
US 20020037489A1
Filed 05/14/1999

Current Assignee
Align Technology Incorporated

Sponsoring Entity
Align Technology Incorporated

METHOD AND SYSTEM FOR TOOTH SEGMENTATION IN DENTAL IMAGES  
Patent #
US 20130022255A1
Filed 05/09/2012

Current Assignee
Carestream Dental Technology Topco Limited

Sponsoring Entity
CareStream Health Incorporated

Medical image processing apparatus, method, and program  
Patent #
US 8,538,105 B2
Filed 10/18/2011

Current Assignee
Fujifilm Corporation

Sponsoring Entity
Fujifilm Corporation

VOXEL LEVEL NEW INFORMATION UPDATES USING INTELLIGENT WEIGHTING  
Patent #
US 20150024337A1
Filed 07/18/2014

Current Assignee
A.Tron3D Gmbh

Sponsoring Entity
A.Tron3D Gmbh

Dental occlusion analysis tool  
Patent #
US 9,017,071 B2
Filed 11/19/2013

Current Assignee
Bruce W. Hultgren

Sponsoring Entity
Bruce W. Hultgren

METHOD OF ALIGNING INTRAORAL DIGITAL 3D MODELS  
Patent #
US 20160070821A1
Filed 09/08/2014

Current Assignee
3M Innovative Properties Company

Sponsoring Entity
3M Innovative Properties Company

TOMOGRAPHIC IMAGE GENERATION DEVICE AND METHOD, AND RECORDING MEDIUM  
Patent #
US 20160081645A1
Filed 09/15/2015

Current Assignee
Fujifilm Corporation

Sponsoring Entity
Fujifilm Corporation

LARGE INTESTINE CONTENTS' DIAGNOSIS AND ACOUSTIC CHARACTERIZATION  
Patent #
US 20170056451A1
Filed 04/26/2016

Current Assignee
HyGIeaCare Inc.

Sponsoring Entity
HyGleaCare Inc.

XRAY IMAGING FOR ENABLING ASSESSMENT OF SCOLIOSIS  
Patent #
US 20170340268A1
Filed 02/13/2017

Current Assignee
Prismatic Sensors AB

Sponsoring Entity
Prismatic Sensors AB

20 Claims
 1. Method for clearing, in particular for removing, unwanted data from optically detected virtual representations of objects, in particular teeth and intraoral structures, the method comprising:
a. Defining an extension line of the representation, b. Generating a projection plane at one point of the extension line, the extension line at this point being perpendicular to the generated projection plane, c. Projecting all known points in space of the representation from one region corresponding to the projection plane onto the projection plane, the corresponding point in space being stored for each projected point, d. Generating a twodimensional curve on the projection plane from the projected points, e. Determining maxima, minima and a center of the curve, f. Identifying projected points of the curve that—
viewed from the center of the curve—
lie outside of the minima or maxima,g. Removing the points in space that correspond to the projected points that were identified in f., h. Optionally, repeating starting from b. for one further point of the extension line.  View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
1 Specification
The invention relates to a method for clearing, in particular for removing, unwanted data from optically detected virtual representations of objects, in particular teeth and intraoral structures.
Many systems for the optical detection of the threedimensional geometry of objects are known in particular in the area of dental treatments. They are used in, for example, the production of prostheses, crowns, inlays or the like, serve for support in the monitoring of orthodontic treatments and/or help in the observation or detection of intraoral structures in general. On the one hand the major advantage of these optical systems is that they are neither invasive nor unpleasant, such as, for example, the dental impression that is often used in conventional dentistry, nor do they constitute a potential risk to patients, as can be the case, for example, in radiationbased methods, such as the xray. On the other hand, the data are in electronic form after acquisition and can be easily stored, for example for later comparisons, or else transmitted, for example from a dentist to a dental laboratory.
One problem that arises constantly in optical methods for detection of the threedimensional geometry of objects, in particular teeth, is that soft parts that are present in the oral cavity, such as the inside of the cheeks or the tongue, are unintentionally acquired. Later correction of these faulty recordings is usually difficult since even in systems that provide several pictures of the same region, the faulty pictures are included in the detected or computed geometry too and corrupt it. Furthermore, unintentionally photographed surfaces constitute an unnecessary additional data volume that under certain circumstances can slow various processes, such as, for example, the visualization of the detected surface geometry.
The approaches to this problem that have been undertaken so far in the state of the art follow mainly two basic strategies. In one strategy, the surfaces that have been defectively acquired are identified as such and removed. One example of this first approach is shown by WO 2013/010910 A1. In the second strategy, empty spaces are defined or identified in which there can be no surfaces, and surfaces that are consequently measured as located in these empty spaces are either removed by the system when identification takes place after measurement, or are ignored from the start. One example of this approach is shown in EP 2 775 256 A1.
It is common to the two systems that during or after scanning, either incorrectly detected surfaces or empty spaces must be actively acquired or recognized as faults; this, on the one hand, requires computer resources and, on the other hand, is susceptible to errors.
Therefore, the object of the invention is to overcome the abovedescribed disadvantages and to make available a simplified method for clearing unwanted surface regions. Preferably, it should also be possible for it to be executed independently of a surface that has been detected at the instant of clearing. This means even without the fault being able to be referenced to an at least partially “finished” surface.
This object is achieved according to the invention by a method of the initially described type, which is characterized in that the method includes the following steps:
 a. Defining of an extension line of the representation,
 b. Generating of a projection plane at one point of the extension line, the extension line being perpendicular to the generated projection plane, in that point,
 c. Projecting of all known points in space of the representation from a region by the projection plane onto the projection plane, whereby for each projected point, at least one corresponding point in space is stored,
 d. Generating a twodimensional curve on the projection plane from the projected points,
 e. Determining of maxima, minima and a center of the curve,
 f. Identifying of projected points of the curve that viewed from the center of the curve lie outside of the minima or maxima,
 g. Removing of the points in space, which correspond to the points identified in step f,
 h. Optionally, repeating starting from step b. a further point of the extension line.
The extension line which is defined in step a) essentially follows the mandibular arch in this case. Possible ways to generate various exact extension lines are explained in later sections.
The plane that has been generated perpendicular to the extension line at a point of the extension line in step b) can also be regarded as a section through the representation.
The projecting from step c) consequently shows essentially the profile of the representation in the section or in the plane of step b). The region can be variously selected in doing so, as is further explained below.
In step d), a twodimensional curve is generated from the projected points of step c). It can contain various substeps, for example for smoothing the curve or for closing gaps. Some possible intermediate steps from step d) are likewise further explained below.
In step e), the maxima and minima as well as a center of the curve are determined. Depending on whether the object in the region of the plane or of the section is a buccal tooth or an incisor, and whether it is located in the upper or lower jaw, the center will be roughly in the area of one or two largest maxima or minima of the curve. This center then lies essentially in the center of the tooth, and the minima or maxima lie on the tips of the teeth. The minima or maxima that are farther away from the center consequently correspond ordinarily to a transition between the gums and other soft tissue, such as, for example, the tongue or the inside of the cheek. If a center cannot be defined in this way, the arithmetic mean between the two end points of the curve can be defined as the center instead.
If, as provided in step f), all points are identified that lie outside of the outer maxima or minima viewed from the center, the unwanted regions are also automatically identified without active recognition of these structures being necessary for this purpose. Of course, the border for identification can also be stipulated to be somewhat outside of the maxima or minima in order not to unintentionally remove desired data.
In step g), all corresponding points in space that correspond to the projected and identified points can then be removed. Consequently, a cleared representation is obtained without the need to actively determine incorrect or correct surfaces in a complicated method for this purpose.
This can then take place in steps for any number of points of the extension line as is defined in step h). The individual planes or sections along the extension line are preferably spaced in this case such that each part of the representation lies in at least one (of the) region(s) from step c) and has been projected onto one of the planes from step b).
In order to computationally simplify the dividing, the planes or sections can preferably be generated equidistantly on the extension line.
Other preferred embodiments of the invention are the subject matter of the remaining dependent claims.
Preferred exemplary embodiments of the invention are described in more detail below using the drawings. Here:
Then, for each brick, the information as to whether the voxels of the brick contain surface information is retrieved (step 12).
If it is ascertained that at least one voxel of the brick contains a surface, a center point of the brick is notated as a location vector. Here, the location vector corresponds to a connection of an origin of a coordinate system, in which the TSDF is notated, to the center point of the brick (step 13).
If a brick does not contain any voxels that contain surface information, it is marked, for example, as “empty” (step 14).
Then, all empty bricks and location vectors are combined into a common point cloud. However, for each location vector it is stored, to which voxels it corresponds (step 15).
In the next step 32, an extension line for the representation is chosen. A highly simplified extension line is a straight line along the representation. One example of such a straight line can be the yaxis of the principal axes determined in
Examples for possible determination of curved extension lines are found in
In a next step 33, planes of intersection of the (optionally simplified) representation are generated. They are each aligned perpendicular to the extension line. If the extension line is a straight line, the planes of intersection are consequently parallel. Preferably, the planes of intersection along the extension line are equidistantly generated. If the representation in step 31 has been simplified, “slices” in the thickness of one or two bricks at a time (for example, 8 or 16 voxels thick) are suitable. A stipulation of the distances of the slices can also be based upon the actual sizes of the represented object regardless of the voxel subdivision. Thus, for example, a distance of 2 mm can be selected. Here, one “slice” corresponds to the region in front of and/or behind the plane of intersection, preferably to the region in front of each plane of intersection viewed in the direction of the extension line. However, for example, several “slices” can also together form the region. In doing so, “slices” in front of and behind the plane (viewed along the extension line) can also be chosen. For curved extension lines, consequently “wedges” form that can build the regions around the extension lines.
In a following step 34, the points in the regions are projected onto the plane. In very simple applications of the invention, for example, all points within the region (which are therefore located in the “slice”) can be mapped along perpendiculars onto the plane. Alternatively, a projection can also take place along perpendiculars of an adjacent plane.
In step 35, a twodimensional curve on the plane is determined. To do this, for example, all imaged points can simply be joined. One preferred and advantageous method with different optional variants for generating a twodimensional curve is shown by
With the curve that has been generated in this way (see, for example, 71, 72 in
In order to be able to actually indicate criteria for distinguishing between teeth and artefacts, such as, for example, parts of a cheek, in this process, as provided according to the invention, the orientation of the coordinate system to the (optionally simplified) model must be considered, depending on whether a “hanging” tooth or a “standing” tooth is being examined; either “maxima” or “minima” are selected as criteria.
In general, all considerations, inasmuch as they relate to minima or maxima, can accordingly also be used reversed. For the sake of clarity, described below is only the procedure for a model in which the teeth and surrounding intraoral structures are oriented such that the tips of the teeth point down. All considerations can be easily transferred by one skilled in the art to models with tips of the teeth that point up.
If the teeth, as in the illustrated example from
In the then following step 38, the points of the curve that lie outside of the established maxima (or minima) (see 77 and 77′ in
In step 39, the points in space or data in the voxels of the TSDF that correspond to the points of the curve that were marked in the preceding step are then erased or set to “unknown” and thus are removed from the representation. With this, the clearing is completed. The process of marking voxels as “unknown” or “unseen” is described in more detail in, for example, US 2015/0024337 A1.
In the preferred method for generating points of the curve that is shown in
The points that are obtained in step 52 are then entered in step 35 on the plane from step 33 as points of the curve.
In doing so, it can happen that the curve is very “serrated”; this can be disadvantageous for different analyses. That is why the curve, as already mentioned, can be smoothed in step 43.
In one preferred alternative further embodiment of the invention, a common distance for all centers can also be set after determining the center of the curve. It can correspond in particular to half of the entire thickness of a molar, in particular ⅔ of the thickness of a molar. For this purpose, for example, a measured thickness can be added. In case there is still too little data for such statements about the object to be measured, for example, statistical data can, however, also be used to choose a corresponding distance. Conventionally, however, a distance of from 5 mm to 7 mm will be suitable.
In order, however, to avoid distortion of the representation in this method, it is useful to optimize the centers before applying the distance to the centers with respect to the probability that the centers lie on the actual tooth centers. For this purpose, the centers that were determined beforehand in the curves on the planes or sections are projected onto a center projection plane that has been spanned between the x and the yaxis. Then a center curve is formed on the center projection plane at these points. Methods of forming these curves and additionally optimizing them in amelioration are explained for
If the center curve has been formed, the components of the representation that are outside of the distance that is to be stipulated can be removed. For this purpose, it is not necessary to use these new centers in the sections or planes. The distance can be much more easily applied directly in the entire model (regardless of the sections). To do this, two parallel curves to the center curve are simply produced at a predetermined or stipulated distance. These parallel curves are then spanned perpendicularly to the center projection plane (therefore along the zaxis) to (parallel) surfaces. The region between the surfaces is then left in the representation. The region outside is removed.
The resulting boundary surfaces can be applied in another further development of the method that is independently of the invention advantageous in order to avoid future faulty data when the representation is being acquired. To do this, the regions outside of the parallel surfaces are blocked from the start and spatial information that is being acquired within these regions is simply not considered, for example when the representation is being generated.
After the projection region has been defined, in step 114, all points of the projection region are projected vertically (therefore following the zaxis of the coordinate system) onto the extension projection plane. This yields a 2D point cloud (shown symbolically and highly schematically in
Within each strip, in step 116, the largest and the smallest xvalues are then determined, and in step 117, the arithmetic mean is formed. From the arithmetic mean from step 117 and the center of the strip on the yaxis, in step 118, a point that is assigned to one strip at a time and that is shown black in
Then, in step 119, a curve can be determined from the points from step 118. One especially suitable and preferred method for this purpose is the method of least squares. Other approximation methods can also be used, however. One possible approximated curve 172 that originated according to the method shown in
Furthermore,
It has been shown that an approximation to a thirddegree polynomial fits especially well to the shape of a dental arch in the anterior region (incisors) of said arch. However, in the posterior region (thus in the direction of the molars) the curve deviates farther from the shape of the mandibular arch than a simple straight extension line. In order to maintain the advantages of the good approximation of the polynomial in the anterior region and to still avoid the major deviation in the posterior region, in an advanced embodiment of the method from
In the subsequent step 122, the normal vectors from step 121 are projected onto a unit sphere (Gaussian projection). The origin of the coordinate system in which the representation or its simplification is notated can be simply used as the center point of the unit sphere. Alternatively, the center of gravity of the representation can be used. Both variants are covered, for example, at the same time when the coordinate system in which the representation or its simplification is notated has been produced according to the method that is shown in
The Gaussian image that has been formed in step 122 can then be examined for free surfaces in the following step 123. In doing so, it is assumed that even if the model has gaps in which no data could have been acquired, in any case no data can be acquired in the region of the jawbone itself. Therefore to identify a larger region in which nothing has been imaged on the sphere at the same time means to identify the jaw or the “origin” of the represented tooth. If then in step 124, a center of this region is determined and then in step 125 a connection is drawn from the center point of the sphere to the center point of the region, it can be assumed that this connection corresponds essentially to the alignment of the represented teeth. Consequently, the connection that was generated in step 125 is stipulated as the direction of the zaxis. In this way, an optimum alignment of the representation to the coordinate system is effected.
One method for determining the (approximate) center of the empty region in step 124 could, for example, consist in that first the center of gravity of all imaged points on the Gaussian sphere is determined. This center of gravity of the imaged points on the Gaussian sphere is then offset somewhat from the center point and will be exactly opposite the empty region. If then a connection is drawn from the center of gravity of the imaged points on the Gaussian sphere to the center point, it points automatically in the direction of the center of the empty region. It must then only still be set to length 1 (while retaining the direction), and the abovedescribed vector that is then stipulated as the zaxis in step 125 is obtained.
In step 126, first the largest eigenvector of the representation is determined for the determination of the other axes of the coordinate system. It will generally not be orthogonal to the abovedefined zaxis and is therefore not suited to be used itself as the axis. Therefore, in step 127, first of all a first crossproduct of the largest eigenvector and the zaxis is determined. The direction of the resulting vector is then defined as the direction of the xaxis. To form the direction of the yaxis, in step 128, the crossproduct of the defined zaxis from step 125 and the defined xaxis from step 127 is then simply formed.
Alternatively, in step 128, the crossproduct of the xaxis that was formed in step 127 and the largest eigenvector that was determined in step 126 can be formed in order to determine a new zaxis. The largest eigenvector is then preserved as the yaxis.
If the method shown in
The method shown in
The advanced embodiment, which was explained for
In general, the described technology can be used both after scanning and also during scanning. If the latter should be desired, for example, an image (clone) of the representation can be produced, processed parallel to detection and can be joined together with the representation that is just being detected at a later time. A method that is suitable for this purpose is shown, for example, by the Austrian utility model with application number GM 50210/2016.
Then, in a step 182, socalled features within the representation are determined. Features are characteristics that stand out in the surface topography of the representation. They can be, for example, edges and in particular peaks, corners or even depressions of the model. Features are generally determined by identifying extreme changes in the surface curvature. To do this, all points of the model and their spatial relationship to adjacent points are examined individually. If all direct neighbors of a point lie essentially in one plane, the point also lies in one plane. If all neighbors of a point lie essentially in two planes, the point lies on an edge. If the neighbors of a point lie in three or more planes, the point lies on a peak or depression. The manner in which the features are determined is irrelevant to the invention. By way of example, but not limiting, the following methods known from the state of the art are mentioned at this point: “Harris Feature Detector”, CenSurE (“Centre Surround Extremas”), ISS (“Intrinsic Shape Signatures”), NARF (“Normal Aligned Radial Feature”), SIFT (“Scale Invariant Feature Transform”), SUSAN (“Smallest Univalue Segment Assimilating Nucleus”), and AGAST (“Adaptive and Generic Accelerated Segment Test”).
If the represented objects are teeth, the features can be, for example, protuberances, tips and/or fissures. Aside from teeth with an unusual malposition, it can usually be assumed that these features follow essentially the mandibular arch. They can therefore be used especially advantageously for construction of an extension line.
Analogously to the method that is shown in
In step 184, the determined features of the representation are projected orthogonally, viz. along the zaxis, into the extension projection plane. As also already described for
The twodimensional point cloud that was generated in step 184 can then be used as a basis for an extension line. In step 185, the latter can be produced, for example, by the application of the Least Squares Method to the points. As already explained for
 11 Breaking down the model into bricks
 12 Voxels in bricks contain surface information?
 13 Determining a common location vector for all voxels of the brick
 14 Marking of the brick as “empty”

15 Joining all location vectors and empty bricks together into a simplified point cloud
FIG. 2 
21 Determining of the covariance matrix of the point cloud from steps 11 to 15 (
FIG. 1 ) or step 31 (FIG. 3 )  22 Determining of the three eigenvectors of the covariance matrix from step 21
 23 Defining of the smallest eigenvector as the direction of the zaxis
 24 Defining of the largest eigenvector as the direction of the yaxis
 25 Defining of the middle eigenvector as the direction of the xaxis
 26 Determining of the center of gravity of the point cloud

27 Defining the center of gravity as the origin of a coordinate system with the axes from steps 23, 24 and 25
FIG. 3 
31 (optional) Simplifying the model (see steps 11 to 15 from
FIG. 1 ) 
32 Determining of an extension line (for example, the yaxis of the principal major axes, see steps 21 to 27 from
FIG. 2 or steps 121 to 128 fromFIG. 12 ; or a curved extension line, see steps 111 to 119 fromFIG. 11 or steps 131 to 135 fromFIG. 13 )  33 Generating of planes of intersection perpendicular to the extension line
 34 Projecting onto the planes

35 Determining of the twodimensional curve (see steps 41 to 43 from
FIG. 4 )  36 Determining of a center of the curve from step 35
 37 Determining of the minima and maxima to the left and right of the center
 38 Marking the points of the curve outside of the outer minima or maxima from step 37

39 Removing the points in space or voxels belonging to the marked points of the curve (in step 38)
FIG. 4 
41 Determining of the projected points (see steps 51 to 53 from
FIG. 5 )  42 (optional) Adding of other points by interpolation between two adjacent points at a time

43 (optional) Smoothing (see steps 61 to 63 from
FIG. 6 )FIG. 5  51 Defining of strips
 52 Determining the centers of gravity of the strips

53 Entering of the centers of gravity as points of the curve
FIG. 6  61 Executing of a Fourier transform onto the points from step 41
 62 Removing of the high frequencies (all set to zero except the base frequency and first harmonic frequency)

63 Executing of an inverse Fourier transform
FIG. 7 
71 A smoothed curve (see also steps 61 to 63 from
FIG. 6 ) 
72 An unsmoothed curve (see also steps 51 to 53 from
FIG. 5 )  73 A middle of the curve
 74 A first maximum of the curve (left)
 75 A second maximum of the curve (right)

76 A region around the middle
FIG. 9 
91 A symbolic curved extension line
FIG. 11 
111 Loading of the representation with a coordinate system (see
FIG. 2 orFIG. 12 )  112 Spanning of an extension projection plane that is spanned from the x and yaxis of the coordinate system
 113 (optional) Defining an extension projection region
 114 Projecting of the points of the extension projection region onto the extension projection plane and generating of a 2D point cloud
 115 Breaking down the 2D point cloud from step 114 into strips
 116 Determining the respective largest and smallest xvalues per strip from step 115
 117 Forming the arithmetic mean of the two values from step 116 for each strip from step 115
 118 Generating of one point per strip from step 115 with the arithmetic mean from step 117 and the center of the strip on the yaxis

119 Creating a curve that is defined as an extension line from the points of step 118
FIG. 12 
121 Making available a vector representation, for example from the method from
FIG. 1  122 Projecting the vectors of the representation from step 121 onto a unit sphere (Gaussian imaging)
 123 Checking of the sphere for a larger free region (without projected vectors)
 124 Determining of a center point of the region from step 123
 125 Defining the direction of the zaxis of the coordinate system as the direction of the connection from the center point of the unit sphere to the center point of the region from step 124
 126 Determining the largest eigenvector of the representation
 127 Forming a first crossproduct of the zaxis from step 125 and the largest eigenvector from step 127 and defining the first crossproduct as the xaxis

128 Forming of a second crossproduct from the zaxis and xaxis and defining the second crossproduct as the yaxis
FIG. 13 
131 Loading of the representation with a coordinate system (see
FIG. 2 orFIG. 12 )  132 Spanning of an extension projection plane between the x and yaxis of the coordinate system
 133 Orthogonal mapping of all points of the (if applicable simplified) representation on the extension projection plane
 134 Applying the method of least squares to the mapping from step 133 and noting of the resulting curve

135 Defining of the curve from step 134 as an extension line
FIG. 17  171 Extension line approximated as a thirddegree polynomial

172 Alternative course of the extension line starting from the inflection point according to the inflection tangent of the polynomial 171
FIG. 18  181 Loading the (if applicable simplified) representation
 182 Determining of features
 183 Generating an extension projection plane
 184 Forming a twodimensional point cloud by projecting the features from step 182 orthogonally onto the extension projection plane from step 183
 185 Forming of a graph along the twodimensional point cloud from step 184