WO2015110162A1

WO2015110162A1 - Method for processing a scatter plot

Info

Publication number: WO2015110162A1
Application number: PCT/EP2014/051346
Authority: WO
Inventors: Philippe Le Goff; Catherine HUGEL - LE GOFF
Original assignee: Hygie-Tech Sa
Priority date: 2014-01-23
Filing date: 2014-01-23
Publication date: 2015-07-30

Abstract

The concept that underlies the invention is based on the idea that all the elements of the scatter plot that constitute the digital representation of a chosen environment, i.e. the points, do not necessarily need to belong to the same single set (the scatter plot), but that it would be wiser to organise them by families. The method according to the invention thus adds at least one and preferably two additional dimensions or characteristic components to the three conventional Cartesian dimensions assigned to the points of the initial scatter plot. Said new dimensions or characteristic components are preferably two in number and provide information for each point respectively on the identity particularity P and on the weighting coefficient C thereof. A point of the scatter plot to which said additional coordinates or characteristic components have been added is thus referred to as "metapoint" or significant point.

Description

Point cloud processing method

The invention relates to a method for processing a scatter plot obtained by a three-dimensional scanning technique of an environment.

The topographic survey, three-dimensional cartography and the activities of measurement of the infrastructures applied in particular in the trades of the architecture, the civil engineering, the scientific computation on the structures and the aerial flows, the surveillance by digital imagery and the Telemetry has evolved with the advent of technologies based on the digitization of the environment and increased computing capabilities of large files.

Among the technologies enabling the digitization of the environment, we will note more particularly the equipment using:

• a laser beam, whether these are fixed or mobile, such as for example:

o laser - portable scanner

o laser - terrestrial scanner

o LIDAR system (Light Detection and Ranging)

• a sound wave,

· A series of photogrammetric shots.

The principle is based on the interpretation by a collector coupled to a digital analyzer of the emissions coming directly or indirectly, after simple or multiple reflections on one or more obstacles, from a source compatible with the nature of the information sought.

These technologies all have in common to deliver the result of their measurements in the form of computer files providing the environment measured in the form of a cloud of points. Each point represents a transmitting element in the case of a direct acquisition or an obstacle element in the case of an indirect acquisition. As a general rule, the point is identified by its Cartesian coordinates in an orthonormal coordinate system that can originate at the location of the measuring equipment possibly associated with a characteristic of the measurement. Depending on the nature of the technology implemented and the model of the equipment used, the digital format of the output file may be different and require prior decoding before it can be read in the clear.

However, in general, the size of the files generated by the scanning equipment of the environment is most often particularly great if the level of precision required for the measurement of the environment is maximum.

This constant prompted software developers and publishers to develop and offer solutions that increase user comfort by constantly considering the duality of the size of the file to be processed (which increases with the level of precision) and the performance of the software. the computer environment available from the user. In addition, the manufacturers of scanning equipment themselves have often promoted the production of software that minimizes the ability to read generated files in order to increase the diffusion of their technology.

In this context, it is logical to note that supply and innovation in the treatment of digitized point clouds has generally followed the evolution of offers and material innovations, that is to say computer systems.

In a non-exhaustive manner, the following references consisting of patents and patent applications illustrate various point cloud processing methods: US6608913, US7746341, US7925109, US20130028482A1.

These references deal mainly with methods to read and visualize large point clouds. The three-dimensional point cloud inspection functions by the user in the computing environment are enhanced to increase usability and visual comfort. In As an extension of these basic functions, some solutions offer complementary treatments that provide a mesh of the point cloud. However, the mesh methods used derive from classical techniques (Delaunay triangulation, pivoting lease, ...) and can by construction only deliver continually differentiable solutions when the reality is not necessarily. While in most cases these mesh methods have no significant impact, there are situations in which such approaches can not be satisfactory. Thus, the use of such solutions to give the physical limits of a computational fluid dynamics domain will not be suitable.

The inventive method is thus born in particular from the need to have a tool that can meet the requirements imposed by the boundary conditions for the physical boundaries of a computational fluid dynamics domain.

The purpose of the present invention is therefore to overcome the aforementioned drawbacks by providing a point cloud processing method that allows to obtain after processing a computer file of reasonable size, not requiring a computer computing capacity or display too much. important and can be used and viewed on a wider variety of computer systems by a wider group of users, without losing precision and significant information about the environment studied and contained in the point cloud representing said environment.

The present invention relates to a method of treating a scatter plot according to claim 1.

The method according to the invention gives the person skilled in the art new solutions to effectively assist him in trades in sectors such as architecture, urban planning, engineering or any other activity requiring work on a three-dimensional representation of the environment. The method makes it possible to obtain from a raw or initial point cloud (generated by appropriate environmental measurement means) a reworked point cloud forming a Computer file less bulky and less greedy in computing resources and therefore more easily usable and viewable by the user, without losing essential information about the environment or limit the possibilities of use of said reworked point cloud. The method of the invention further comprises steps for selectively and automatically isolating an object contained in the digitized environment. It also allows to merge, add or subtract digital records in the form of point clouds highlighting the common objects and differences between two successive surveys. This invention also automatically reduces the number of aberrant points and artifacts present on the initial raw point cloud, the presence of this "noise" in the digitized environment is well known to users and considerably reduces operating comfort. . The examples given below illustrate some of these possibilities.

The appended figures describe schematically and by way of example the method according to the invention as well as examples of implementation of said method.

Figure 1 schematically illustrates the different steps of the method of processing a scatter plot according to the invention.

Figure 2 illustrates the set of metapoints corresponding to a cloud of points digitizing a certain environment.

FIG. 3 illustrates the set of significant metapoints corresponding to FIG. 2 after implementation of the method according to the invention.

FIG. 4 illustrates a set of significant metapoints obtained using the method according to the invention.

FIG. 5 illustrates the result of the assimilation operation applied in FIG. 4.

FIG. 6 illustrates a set of significant metapoints obtained using the method according to the invention. FIG. 7 illustrates the result of the aggregation operation applied in FIG.

FIG. 8 illustrates the import of an external object into a set of significant metapoints generated by the method according to the invention.

FIG. 9 illustrates an example of use of the set of significant metapoints generated by the method according to the invention, the result of the method having been converted into a suitable computer file format and then able to be visualized and / or reworked on other computing platforms.

After scanning a selected environment using appropriate measurement means as described above, a digital survey is obtained in the form of a scatter plot constituting a computer file. At this point, each point of the point cloud is identified by its Cartesian coordinates considered with respect to an origin which may be for example at the positioning point of the measuring means. The computer file obtained is generally very large and requires significant computer resources for its reading and visualization.

The concept at the origin of the invention is based on the idea that all the elements of the point cloud constituting the digital reading of a chosen environment, ie the points themselves, should not not necessarily all belong to one and the same set (the cloud), but it would be better to organize them into families. This idea is to apply the theory of sets to the collection of points contained in a cloud of points.

The method according to the invention thus adds at least one but preferably two additional dimensions or characteristic components to the three conventional Cartesian dimensions assigned to the points of the initial point cloud and uses these new dimensions as described below. Preferably, these new dimensions or characteristic components are two in number and provide respectively for each point on its identity characteristic P and on its weighting coefficient C. A point of the cloud of points to which have been added these additional coordinates or characteristic components is then named "metapoint" or significant point.

As illustrated in FIG. 1, in a first step of the method according to the invention, the points of the digitized initial point cloud are first converted into metapoints.

The metapoints are then positioned in a repository independent of the means of measuring the environment used and their positioning constructed by the method. This repository, which is an orthonormal reference, is taken as absolute reference. The spatial coordinates of any given metapoint in the absolute repository can, however, be expressed in any other repository by conventional vector operations using rotation matrices and translation matrices. During such a transfer operation from one repository to another, the first and second characteristic components (P, C) of the metapoints are kept unchanged.

A metapoint M is thus represented by a quintuplet of type (X; Y; Z;

P; C) in which:

• X is the coordinate of the metapoint M along the Ox axis;

• Y is the coordinate of the metapoint M along the axis Oy;

• Z is the coordinate of the metapoint M along the Oz axis;

· P is the first characteristic component or identity characteristic of the metapoint M;

• C is the second characteristic component or weighting coefficient of the metapoint M;

where Ox, Oy, Oz designates the orthonormal coordinate system, considered as the absolute reference.

The X, Y and Z components of the quintuplet are classic and do not require any particular comment. The first characteristic component P is obtained by numerical or manual indexing of the point cloud. It may for example be constituted by the value of a reflectance, a phase, a temperature, a color, an intensity recorded with the digital data during the measurement of the environment.

On this first characteristic component P, the method according to the invention builds in a second step a partition of the initial point cloud in equivalence classes according to the theory of sets in Boolean or Lindenbaum algebra. The properties of equivalence relations being satisfied, it follows that:

• The meeting of all equivalence classes covers the entire space, that is, the initial scatter plot;

• Equivalence classes, which are three-dimensional objects in the digitized environment, are disjointed; a metapoint M can belong to only one class.

The method according to the invention then makes it possible to modify the partition initially obtained, which in the absence of characteristic information becomes the limit case according to which a single equivalence class exists and then corresponds to the initial point cloud in its entirety. The invention makes it possible to modify the partition of the set (the cloud of points or a collection of points clouds) according to criteria chosen by the user.

The second characteristic component C assigns each point of the point cloud a weighting coefficient. By default, each point receives an equal weight worth one unit. The advantage of the second characteristic component C lies in the fact that it allows a rational reduction of the size of the raw computer file as digitized by the measuring equipment as described below.

In a new step, the method according to the invention constructs a mesh of the space of the point cloud according to the axes of the absolute reference frame Ox, Oy, Oz. The mesh consists of cubes called meshes and filling the whole space. Each mesh is identified by a point, called reference point A (a vertex, a center face, the center) and by the length a of its edge. This length can be adapted at will by the user.

Then, by the application of an iterative algorithm, the method according to the invention determines a new set of metapoints, called significant metapoints M, for which the existence is controlled by the condition:

• The second characteristic component C of the significant metapoint M on the square mesh A and of edge a is greater than a limit value d; · The second characteristic component C of the significant metapoint M on the mesh A and of edge is obtained by summing the weights of the set of metapoints Mi contained in the mesh A and satisfying the condition of belonging to the same class equivalence.

The limit value d is set by the user and corresponds to the minimum density of points per unit area desired.

This approach can significantly reduce the computer size of files without loss of information compared to the original data. This reduced set of metapoints is called together significant metapoints. Each significant metapoint contains the set of characteristics of the metapoints belonging to the same equivalence class and found in the mesh A that it characterizes.

For each significant metapoint M, the method can determine a significant metapoint M 'adjoint. The adjoint M 'is defined by its components M': (Χ ', Υ', Ζ ', P, C) whose P and C values of the first and second characteristic components are identical to that of the significant metapoint M: (X , Y, Z, P, C). The adjoint M 'of M is the result of a spatial transformation obtained by a succession of rotations and translations. These properties of the method according to the invention are essential because they make it possible to carry out the basic algebraic operations on two sets of significant metapoints. These basic operations are called merge, addition and subtraction.

Be two digital readings of the same environment. These records may have been obtained at different times, from different points of view, and so on. By applying the method according to the invention to each of these readings, first and second clouds of (meta) significant points are obtained. The use of an absolute reference that does not depend on the measurement means makes it possible to compare and perform certain operations on the significant metapoints. So that two significant metapoints M1 (X1, Y1, Z1, P1, C1) and M2 (X2, Y2, Z2, P2, C2), M1 belonging to the first significant metapoints cloud and M2 belonging to the significant second metapoint cloud, can to be candidates for one of the basic operations, it is first necessary to check the general condition:

X1 = X2 and Y1 = Y2 and Z1 = Z2,

where X1, Y1, Z1, and X2, Y2, Z2 respectively denote the coordinates of M1 and M2 in the absolute frame of reference.

The metapoint M3 (X3, Y3, Z3, P3, C3) resulting from the operation on M1 and M2 will also be a significant metapoint belonging to a third cloud of significant points resulting from the operation on the first and second scatterplots significant and will have for coordinates in the absolute reference common to the three clouds of points:

X3 = X1 = X2 and Y3 = Y1 = Y2 and Z3 = Z1 = Z2.

For merge, add, and subtract operations, the following specific condition must also be true:

P1 = P2,

where P1 and P2 respectively denote the first characteristic component of M1 and M2. The second characteristic component C3 of M3 for the merging operation is given by:

C3 = max (C1, C2).

The second characteristic component C3 of M3 for the addition operation is given by:

C3 = C1 + C2.

The second characteristic component C3 of M3 for the operation of subtraction of M1 by M2 is given by:

C3 = C1 - C2.

In contrast, the second characteristic component C3 of M3 for the M2 subtraction operation by M1 is given by:

C3 = C2 - C1.

The merger and addition operations are in their nature quite close. The collection of significant metapoints M3 resulting from the operation is differentiated only by the value of the second characteristic component C3. In the case of fusion, the second characteristic component C3 is taken equal to the largest between C1 and C2. In the case of addition, the second characteristic component C3 is taken equal to the algebraic sum of C1 plus C2.

Physically, the difference between fusion and addition can be interpreted as follows. Either the observation of a volume element from two distinct points of view: in the case of fusion, we keep the collection of significant metapoints of the element recorded from the best point of view at the risk of underestimating the density C3; in the case of addition, we sum the collection of significant metapoints of the element recorded from each point of view at the risk of overestimating the density C3.

Unlike the merge and add operations, the subtraction operation is not commutative. Also, we will have M3 ≠ M3 'if M3 = M1 - M2 and M3' = M2 - M1. A practical use of the subtraction operation, as made possible by the method according to the invention, consists in comparing two collections of significant metapoints of the same spatial domain but digitized at different times. The comparison by subtraction will have the effect of "extinguishing" all the significant metapoints present on the two digitizations and "illuminating" only the significant metapoints belonging to new or missing objects according to whether it is M3 or M3 ' .

This feature is specific to the method according to the invention. It finds many applications, especially in the field of surveillance.

By its innovative approach, the method according to the invention also allows complementary operations very useful to those skilled in the art. Thus, by the aggregation and assimilation operations, the method makes it possible to obtain respectively:

• The set of significant metapoints that belong to the same object but could not be

o neither determined during the digitization,

o neither manually marked on the display of the cloud of significant points by the graphical interface;

• The set of metapoints that belong to the same plane in the absolute repository.

The aggregation operation is performed by assigning, in a first step, the first characteristic component a value P common to all the metapoints present in the spatial domain considered. This step consists of placing the set of metapoints in the same equivalence class. The second step is to select from among the metapoints a series of metapoints (at least one) and assign them for the first characteristic component a value P 'different from P. In a third step, the metapoints located outside the domain of work are characterized by the assignment for their first characteristic component of a value P "different from P 'and P. These initial conditions being determined, the method according to the invention uses an iterative approach to assign the metapoints of the equivalence class P the value P 'subject to satisfying the inclusion conditions given by:

• The distance from a metapoint of class P to a metapoint of class P 'is less than a certain chosen value;

• The requirement of continuity on the mesh.

The assimilation operation is in principle similar to that of the aggregation operation except that the requirement of continuity on the mesh is replaced by the double requirement of continuity on the mesh and flatness of the metapoints. The set of metapoints of the class P 'obtained by the assimilation operation can then be placed in different subclasses. This makes it possible to index the 3D objects belonging to the plane constructed by assimilation.

The method according to the invention can be implemented by a computer system comprising a program and thus makes it possible, from an initial point cloud generated by appropriate measurement means, to obtain a cloud of significant points whose computer size is smaller but retains all the environmental information contained in the initial point cloud and furthermore can be easily manipulated and viewed by the user who can perform various operations on the reworked cloud.

Example 1

Generate a single class of meaningful metapoints from a point cloud assembly

The files of the initial point clouds are for example loaded one by one in the computer program that includes a computer system and giving the user interface for the method according to the invention. Each point cloud is oriented so that its own axes coincide with the axes of the absolute repository. A remarkable point visible on each cloud is positioned on the origin of the absolute reference. P values of metapoints are transiently forced to 0 to allow the complete merging of the metapoint files into a single global file.

The user defines a unit mesh by choosing the value of its edge "a". The user defines the domain in which the meshes will be generated. The mesh is then automatic, it is built according to the axes of the absolute reference.

The global file of the metapoints is loaded on this mesh, the calculation of the significant metapoints is realized automatically by activating the corresponding menu.

Figures 2 and 3 respectively show the set of metapoints corresponding to the initial point clouds and the set of significant metapoints.

Example 2

Generation of single surfaces by assimilation operation from a set of significant metapoints

The file of the significant metapoints is loaded into the computer program the computer program that includes a computer system and giving the user interface for the method according to the invention. The user activates the menu for automatically searching active plans and performing the assimilation operation. Figures 4 and 5 respectively show the significant metapoints and the result of the assimilation operation.

Example 3

Tracing pipe elements by aggregation operation from a set of significant metapoints

The file of the significant metapoints is loaded into the computer program the computer program that includes a computer system and giving the user interface for the method according to the invention. The P values are all set equal to 0 for all significant metapoints. A sample of the three-dimensional view pipe is selected, the P values of significant metapoints in the selection are all set equal to 1. Significant metapoints with values P = 0 and P = 1 are grouped in a single file that is reloaded in the computer program. The end condition of the aggregation operation is defined by the user and the calculation is activated by the corresponding menu. At the end of the operation, all the points belonging to the pipe are saved in a specific file with for all metapoints a value of P = 1.

Figures 6 and 7 respectively show the significant metapoints and the result of the aggregation operation highlighting the set of significant metapoints pertaining to the pipe.

Example 4

Export in dxf format for use in another DAO application of a staging of three-dimensional objects from various origins

A simple surface file, as described in Example 2, is loaded into the computer program the computer program that includes a computer system and giving the user interface for the method according to the invention. A three-dimensional vector file representing a certain object in the dxf format is imported into the program. The values of the P component of the significant metapoints of the imported object are set equal. The group forming the object of the significant metapoints is positioned by the control interface of the computer program in the configuration desired by the user. The composite scene is saved and exported in dxf format for use on other platforms.

Figures 8 and 9 show respectively the configuration of a composite scene and the result of the export in the dxf format when opened by another software. Example 5

Reduce the "background noise" present on the raw digital acquisition file of a three-dimensional environment

The procedure of Example 1 is carried out in its entirety. Assuming that the result obtained contains significant points visibly corresponding to "background noise" that its origin is of the "lost points" type or of the "ghost point" type, the user sets a filter value d (corresponding then to the determination of the limit value of the component C to give a significant metapoint) higher than that proposed by default. The user executes again the request for determination of the significant points by activating the menu of "fine filtration".

Claims

claims

Method of computer-implemented processing of at least a first initial point cloud generated by measuring means and constituting a first digital survey of an environment

three-dimensional characterized by the fact that it comprises the following steps:

at) . Positioning each of the points of the first initial point cloud obtained by measuring the chosen environment with the aid of appropriate measuring means, in a first referential called absolute reference, independent of said measuring means, each of said points of the first point cloud initial being identified by its coordinates in said absolute reference;

b). Associate with each of the points of the first initial point cloud considered in the absolute referential a first characteristic component (P) and a second characteristic component (C), which initially is the same for all the points, and which do not depend on the absolute reference frame ;

vs) . Partition the first initial point cloud into classes

equivalence according to the first characteristic component (P); d). Constructing a mesh of the space of the first initial point cloud according to the axes of the absolute referential, each mesh being defined by a reference point (A) and a length of edge (a) determined by the user;

e). Determine for each mesh (A) a significant point, said significant point being determined by the coordinates of the reference point (A) of the mesh in the absolute reference frame, the first characteristic component (P) of the significant point being equal to the first component characteristic (P) of the point of reference (A) of the mesh and the second component

characteristic (C) of the significant point being the sum of the second characteristic components (Ci) of all the points (Mi) of the first initial point cloud considered in the absolute reference and contained in the mesh and belonging to the equivalence class of the point reference point (A) of said mesh, the existence condition of such a significant point being that its second component (C) is greater than or equal to a determined weight value (d), the set of significant points thus determined. forming a first cloud of significant points of reduced size with respect to the first initial point cloud.

Method according to Claim 1, characterized in that the first characteristic component (P) may consist of a reflectance, phase, temperature, color, intensity value recorded by the measuring means during the measurement of

the chosen three-dimensional environment.

Method according to one of the preceding claims, characterized in that it further comprises the following steps:

f). Apply steps a) to e) of the method to at least a second initial point cloud generated by the measurement means and constituting a second digital survey of the three-dimensional environment digitized by the first initial point cloud to obtain a second cloud significant points;

boy Wut) . Determine a third cloud of significant points whose significant points are characterized as follows:

For any pair formed of a significant point M1 (X, Y, Z, P, C1) of the first significant point cloud and a significant point M2 (X, Y, Z, P, C2) of the second cloud of significant points, such that the coordinates (X, Y, Z) of said points of the first and second significant point clouds in the absolute referential are the same and such that the first characteristic component (P) of said points is equal, a new significant point M3 (X, Y, Z, P, C3) is determined belonging to the third significant point cloud and characterized by the same coordinates (X, Y, Z) in the absolute reference and the same first characteristic component (P) as the points M1, M2 of the first and second significant point clouds and whose second characteristic component (C3) is a function of the second characteristic components (C1, C2) of the points M1, M2 of the first and second clouds of significant points: C3 = f (C1, C2).

Method according to claim 3, characterized in that said function is given by f (C1, C2) = C3 = C1 + C2.

Method according to claim 3, characterized in that said function is given by: f (C1, C2) = C3 = max (C1, C2).

Method according to claim 3, characterized in that the said function is given by: f (C1, C2) = C3 = C1 - C2.

Method according to claim 3, characterized in that the said function is given by: f (C1, C2) = C3 = C2 - C1.

8. Method according to claims 1 or 2, characterized in that it further comprises the following steps: f. assigning to the first characteristic component (P) a first value common to all the significant points obtained a value;

g ', select according to a predefined criterion at least one point

significant M among the set of significant points and assign to it for its first characteristic component a second value, different from the first value;

h ', assigning at any significant point the second value for first characteristic component if the following conditions are satisfied:

i) the distance of a significant point, the first

characteristic component is the first significant point value whose first component

characteristic is the second value is less than a predetermined distance;

ii) the continuity on the mesh is satisfied.

9. Method according to claim 8, characterized in that in

step h 'the following condition must still be satisfied:

iii) Significant points of the resulting subset

belong to the same plane.

Computer system comprising a program for implementing the method according to one of claims 1 to 9.

1 1. Computer system according to the preceding claim, further comprising display means for displaying the cloud of significant points obtained by the method according to one of claims 1 to 9.