CN111652855A - Point cloud simplification method based on survival probability - Google Patents

Point cloud simplification method based on survival probability Download PDF

Info

Publication number
CN111652855A
CN111652855A CN202010427712.7A CN202010427712A CN111652855A CN 111652855 A CN111652855 A CN 111652855A CN 202010427712 A CN202010427712 A CN 202010427712A CN 111652855 A CN111652855 A CN 111652855A
Authority
CN
China
Prior art keywords
points
point
curvature
boundary
survival probability
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010427712.7A
Other languages
Chinese (zh)
Other versions
CN111652855B (en
Inventor
梁晋
赫景彬
刘世凡
李成宏
马金泽
苗泽华
邬宏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN202010427712.7A priority Critical patent/CN111652855B/en
Publication of CN111652855A publication Critical patent/CN111652855A/en
Application granted granted Critical
Publication of CN111652855B publication Critical patent/CN111652855B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Geometry (AREA)
  • Quality & Reliability (AREA)
  • Processing Or Creating Images (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a point cloud simplification method based on survival probability, wherein in the method, original point cloud data are read, a topological relation is established for the original point cloud data based on a kdtree algorithm, and all neighborhood points in the radius r range of each data point are obtained; performing covariance analysis on each data point and neighborhood points thereof by using multi-thread parallel computing based on a principal component analysis method to obtain a covariance matrix; dividing all data points into boundary points or non-boundary points according to whether the data points are boundary points or not, sorting the non-boundary points according to the curvature, and dividing the non-boundary points into high curvature points and low curvature points according to a preset threshold; according to a predetermined compaction ratio to n1、n2、n3Calculating the number of points to be deleted of the boundary point, the high curvature point and the low curvature point; traversing each data point of point cloud based on multi-thread parallel computingAnd randomly generating a random number with the size between 0 and 1 every time, and comparing the random number with the survival probability to obtain the simplified point cloud data.

Description

Point cloud simplification method based on survival probability
Technical Field
The invention belongs to the technical field of high-precision 3D measurement, and particularly relates to a point cloud simplification method based on survival probability.
Background
With the continuous development of the 3D sensor technology, the three-dimensional scanning technology for acquiring the point cloud data of the 3D model is continuously updated, and the measurement accuracy and efficiency of the acquired point cloud data of the object surface are also higher and higher. The point cloud data obtained by scanning with the high-precision structured light scanner has a large amount of redundancy, and in practical application, the scale of the original point cloud data obtained by the high-precision structured light three-dimensional scanner is usually in the order of tens of millions or even hundreds of millions, so that the burden of point cloud data storage, transmission and operation and the difficulty of subsequent processing work are increased. There is a need to condense point clouds while preserving point cloud features and boundary information.
In recent years, a bounding box method, a clustering method, and the like are commonly used as point cloud reduction algorithms. The bounding box method is to generate a three-dimensional grid using octree partitioning. And if the normal vector deviation of the point cloud in the traversal grid is larger than a specified threshold value, subdividing the cells. And after the grid division is finished, selecting representative points for each grid to form a simplified point cloud. This approach can result in high curvature partial feature loss. The core idea of the clustering method is a divide-and-conquer method, and the method can be divided into a bottom-up region growing algorithm and a top-down hierarchical method according to the dividing idea. The partition abort condition is that the number of points in the class reaches a threshold value, which causes problems such as boundary shrinkage. When the traditional method is applied to high-precision industrial measurement, measurement deviation is increased, so that a point cloud reduction algorithm for guaranteeing boundaries and characteristics is important for improving measurement precision.
The above information disclosed in this background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art.
Disclosure of Invention
Aiming at the problems of poor feature preservation effect and difficulty in parallel computing in the prior art, the invention provides a point cloud simplification method based on survival probability.
The invention aims to realize the purpose through the following technical scheme, and the point cloud simplification method based on the survival probability comprises the following steps:
reading original point cloud data, establishing a topological relation for the original point cloud data based on a kdtree algorithm, and acquiring all neighborhood points within a radius r range of each data point;
in the second step, based on a principal component analysis method, performing covariance analysis on each data point and its neighborhood points by using multi-thread parallel computation to obtain covariance matrixes, and respectively computing three eigenvalues lambda of the covariance matrixes1、λ2、λ3The data points correspond to curvatures of
Figure BDA0002498265920000021
In the third step, all data points are divided into boundary points or non-boundary points according to whether the data points are boundary points or not, the non-boundary points are sorted according to the curvature, the data points are divided into high curvature points and low curvature points according to a preset threshold value, and the number of the boundary points, the number of the high curvature points and the number of the low curvature points are respectively n1、n2、n3，
In the fourth step, n is added according to a predetermined reduction ratio1、n2、n3The method comprises the steps of calculating the number of deleted points needed by boundary points, high curvature points and low curvature points, wherein the low curvature points are deleted firstly, then the high curvature points are arranged, finally the boundary points are arranged, determining a survival probability model based on the simplification ratio of point cloud and the number of the boundary points, the high curvature points and the low curvature points, and endowing survival probability to each boundary point, the high curvature points and the low curvature points, wherein the survival probability of the boundary points is the maximum, and the survival probability of non-boundary points is determined from the curvature pointsThe size is decreased progressively from the big to the small,
in the fifth step: and traversing each data point of the point cloud based on multi-thread parallel computing, randomly generating a random number with the size of 0-1 each time, comparing the random number with the survival probability, reserving the data point when the random number is smaller than the survival probability, and deleting the data point to obtain the simplified point cloud data if the random number is not smaller than the survival probability.
In the method, in the first step, the number n of points in the neighborhood of each data point is countediWherein the radius r is 5mm when niIf the number of the data points is less than 5, the data points are judged to be outliers and deleted, and the total number of the points is changed into n after the neighborhood point searching operation is finished.
In the method, in the second step, each data point p and its neighborhood point p are calculated in a multi-thread parallel computing based on a principal component analysis methodiCentralizing the data point p as a center, carrying out covariance analysis on the centralized point to obtain a covariance matrix A,
Figure BDA0002498265920000022
wherein n is the total point number, A is a symmetric semi-positive definite matrix, all the eigenvectors are real numbers, and the eigenvectors are mutually orthogonal.
In said method, λ1Greater than λ2，λ2Greater than λ3。
In the method, in the third step, the determination of each data point boundary point includes projecting the data point and its neighborhood points to the tangent plane where the data point is located, connecting each neighborhood point with the central point located at the center to obtain the maximum included angle α, and determining the data point as the boundary point when α is greater than 90 °.
In the fourth step, in the survival probability model, only the low curvature points are simplified, the corresponding survival probability of the boundary points and the high curvature points is 1, and the survival probability of the low curvature points is (n)3-nr0)/n3The minimum point threshold value after the simplification is n1+n2+kr0n3Wherein n is1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (4) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points, and n is the total point number.
In the fourth step, in the survival probability model, the high curvature point and the low curvature point are simplified, the survival probability of the high curvature point is 1, the survival probability of the boundary point is 1, and the survival probability of the low curvature point is kr0The minimum point threshold value after the simplification is n1+0.5n2(1+kr0)+kr0n3，n1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
In the fourth step, in the survival probability model, the high curvature point and the low curvature point are simplified, and the survival probability of the high curvature point is kr0Boundary point survival probability of 1 and low curvature survival probability of kr0The minimum point threshold value after the simplification is n1+0.5n2(3kr0+kr0)+kr0n3Wherein n is1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
In the fourth step, in the survival probability model, the boundary points, the high curvature points and the low curvature points are simplified, and the survival probability of the high curvature points is kr0To 3kr0The boundary point survival probability is (nr)0-2kn2r0-kn3r0)/n1Low curvature survival probability of kr0The minimum point threshold value after the simplification is 3kr0n1+2kr0n2+kr0n3，n1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
When point cloud is simplified, the points are firstly divided into boundary points, high curvature points and low curvature points. In the case of point cloud reduction, what we prefer to keep is the boundary region and the feature region, and the feature region exists in the high curvature part. As desired, different survival probabilities are given to the boundary points, the high curvature points, and the low curvature points, respectively, and the boundary points are most likely to survive, and the non-boundary points are more likely to survive as their curvatures are larger. The points reduced by the algorithm are a proper subset of the original point cloud, so that the measurement deviation cannot be increased due to the movement of the points, and the characteristics of the original point cloud are retained to the maximum extent. During curvature calculation and point deletion operation, a multithreading parallel acceleration technology can be adopted, and the overall operation efficiency of the algorithm is guaranteed.
The above description is only an overview of the technical solutions of the present invention, and in order to make the technical means of the present invention more clearly apparent, and to make the implementation of the content of the description possible for those skilled in the art, and to make the above and other objects, features and advantages of the present invention more obvious, the following description is given by way of example of the specific embodiments of the present invention.
Drawings
Various other advantages and benefits of the present invention will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. It is obvious that the drawings described below are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort. Also, like parts are designated by like reference numerals throughout the drawings.
In the drawings:
FIG. 1 is a schematic flow chart of the overall process of the present invention;
FIG. 2(a) and FIG. 2(b) are schematic diagrams of boundary point condition determination;
FIGS. 3(a) to 3(d) are graphs of four survival probability models according to the present invention;
FIG. 4 is a diagram of pump body model raw point cloud data;
FIG. 5 is a point cloud data diagram after a pump body model is reduced by 25%;
FIG. 6 is a detail view of point cloud data after a pump body model is subjected to 25% simplification;
FIG. 7 is a diagram of the effect of the pump body model after 25% reduction.
The invention is further explained below with reference to the figures and examples.
Detailed Description
Specific embodiments of the present invention will be described in more detail below with reference to fig. 1 to 7. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention may be embodied in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
It should be noted that certain terms are used throughout the description and claims to refer to particular components. As one skilled in the art will appreciate, various names may be used to refer to a component. This specification and claims do not intend to distinguish between components that differ in name but not function. In the following description and in the claims, the terms "include" and "comprise" are used in an open-ended fashion, and thus should be interpreted to mean "include, but not limited to. The description which follows is a preferred embodiment of the invention, but is made for the purpose of illustrating the general principles of the invention and not for the purpose of limiting the scope of the invention. The scope of the present invention is defined by the appended claims.
For the purpose of facilitating understanding of the embodiments of the present invention, the following description will be made by taking specific embodiments as examples with reference to the accompanying drawings, and the drawings are not to be construed as limiting the embodiments of the present invention.
The point cloud simplification method based on the survival probability comprises the steps of,
reading original point cloud data, establishing a topological relation for the original point cloud data based on a kdtree algorithm, and acquiring all neighborhood points within a radius r range of each data point;
in the second step, based on a principal component analysis method, performing covariance analysis on each data point and its neighborhood points by using multi-thread parallel computation to obtain covariance matrixes, and respectively computing three eigenvalues lambda of the covariance matrixes1、λ2、λ3The data points correspond to curvatures of
Figure BDA0002498265920000051
In the third step, all data points are divided into boundary points or non-boundary points according to whether the data points are boundary points or not, the non-boundary points are sorted according to the curvature, the data points are divided into high curvature points and low curvature points according to a preset threshold value, and the number of the boundary points, the number of the high curvature points and the number of the low curvature points are respectively n1、n2、n3，
In the fourth step, n is added according to a predetermined reduction ratio1、n2、n3The number of the points to be deleted of the boundary points, the high curvature points and the low curvature points is calculated, wherein the low curvature points are deleted firstly, then the high curvature points and finally the boundary points are deleted, a survival probability model is determined based on the simplification ratio of the point cloud and the number of the boundary points, the high curvature points and the low curvature points, the survival probability is given to each boundary point, the high curvature points and the low curvature points, wherein the survival probability of the boundary points is the maximum, the survival probability of the non-boundary points is decreased from the maximum to the minimum along with the curvature,
in the fifth step: and traversing each data point of the point cloud based on multi-thread parallel computing, randomly generating a random number with the size of 0-1 each time, comparing the random number with the survival probability, reserving the data point when the random number is smaller than the survival probability, and deleting the data point to obtain the simplified point cloud data if the random number is not smaller than the survival probability.
In a preferred embodiment of the method, in a first step, the number n of points in the neighborhood of each data point is countediWherein the radius r is 5mm when niIf the number of the data points is less than 5, the data points are judged to be outliers and deleted, and the total number of the data points is counted after the neighborhood point searching operation is finishedBecomes n.
In a preferred embodiment of the method, in the second step, each data point p and its neighborhood point p are calculated in parallel using multiple threads based on principal component analysisiCentralizing the data point p as a center, carrying out covariance analysis on the centralized point to obtain a covariance matrix A,
Figure BDA0002498265920000061
wherein n is the total point number, A is a symmetric semi-positive definite matrix, all the eigenvectors are real numbers, and the eigenvectors are mutually orthogonal.
In a preferred embodiment of said method, λ1Greater than λ2，λ2Greater than λ3。
In a preferred embodiment of the method, in the third step, the determining of each boundary point of the data points includes projecting the data points and neighboring points thereof to a tangent plane where the data points are located, connecting the neighboring points with a central point located at the center to obtain a maximum included angle α, and determining the data points as boundary points when α is greater than 90 °.
In the fourth step, in the survival probability model, only the low curvature points are reduced, the survival probability of the boundary point corresponding to the high curvature point is 1, and the survival probability of the low curvature point is (n)3-nr0)/n3The minimum point threshold value after the simplification is n1+n2+kr0n3Wherein n is1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (4) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points, and n is the total point number.
In the fourth step, in the survival probability model, the high curvature point and the low curvature point are reduced, the survival probability of the high curvature point is 1, the survival probability of the boundary point is 1, and the survival probability of the low curvature point is kr0The minimum point threshold value after the simplification is n1+0.5n2(1+kr0)+kr0n3，n1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
In a preferred embodiment of the method, in the fourth step, in the survival probability model, the high curvature point and the low curvature point are reduced, and the survival probability of the high curvature point is kr0Boundary point survival probability of 1 and low curvature survival probability of kr0The minimum point threshold value after the simplification is n1+0.5n2(3kr0+kr0)+kr0n3Wherein n is1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
In a preferred embodiment of the method, in the fourth step, in the survival probability model, the boundary points, the high curvature points and the low curvature points are reduced, and the survival probability of the high curvature points is kr0To 3kr0The boundary point survival probability is (nr)0-2kn2r0-kn3r0)/n1Low curvature survival probability of kr0The minimum point threshold value after the simplification is 3kr0n1+2kr0n2+kr0n3，n1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
In a preferred embodiment of the method, data points with a curvature of less than 0.02 are low curvature points and data points with a curvature of greater than 0.02 are high curvature points.
For a further understanding of the present invention, in one embodiment, a method includes,
reading original point cloud data, quickly establishing a topological relation on the original point cloud by using kdtree, and acquiring all neighborhood points within the radius r of each point;
by utilizing a PCA method and using a multithreading parallel computing technology to quickly perform covariance on each data point and neighborhood points thereofAnalyzing to obtain covariance matrix, and respectively calculating its three eigenvalues lambda1、λ2、λ3，λ1＞λ2＞λ3. The point corresponding to a curvature of
Figure BDA0002498265920000071
All data points are divided into two types, namely boundary points and non-boundary points according to whether the data points are boundary points or not. And sorting the non-boundary points according to the curvature, and dividing the non-boundary points into high curvature points and low curvature points according to a set threshold value. The number of boundary points, high curvature points and low curvature points is n1、n2、n3；
According to the desired percentage of reduction, with n1、n2、n3The number of points to be deleted for each type of points is calculated. Optimally, the low curvature point is deleted firstly, then the high curvature point is arranged, and finally the boundary point is arranged. Selecting a survival probability model based on the curvature characteristics of the point cloud and combining the number of points to be reduced in each region, and endowing each point with a survival probability, wherein the survival probability of the boundary points is the largest, and the survival probability of the non-boundary points is decreased from large to small along with the curvature;
and traversing each data point of the point cloud by using a multi-thread parallel computing technology, randomly generating a random number with the size of 0-1 each time, comparing the random number with the survival probability, and when the random number is smaller than the survival probability, keeping the point, otherwise, deleting the point. And finally, the simplified point cloud data is obtained.
In one embodiment, as shown in fig. 1, the concrete steps of the compaction method are as follows:
step S01: reading original point cloud data, wherein the number of points in the point cloud is N. Constructing a topological structure of the original point cloud by establishing kdtree, searching and recording all neighborhood points in the neighborhood of each data point radius r according to the established kdtree, and counting the number n of the points in the neighborhoodiWherein the radius r is typically 5 mm. When n isiIf < 5, the point is determined to be an outlier, and the point is deleted. And after the neighborhood point searching operation is finished, the total number of points is changed into n.
Step S02:and solving the curvature of each data point, and improving the operation rate by using a multithreading parallel acceleration technology. Firstly, each data point p and its neighborhood point piCentering is performed with this data point p as the center. And carrying out covariance analysis on the centered points to obtain a covariance matrix A:
Figure BDA0002498265920000072
a is a symmetric semi-positive definite matrix, all eigenvectors of the matrix are real numbers, and the eigenvectors are mutually orthogonal. Respectively calculate three characteristic values lambda thereof1、λ2、λ3(λ1＞λ2＞λ3). The point corresponding to a curvature of
Figure BDA0002498265920000081
S03, counting the number of boundary points, high curvature points and low curvature points, judging the boundary point of each data point, projecting the point and its neighborhood point to the tangent plane of the point, connecting each neighborhood point with the center point to obtain the maximum included angle α, judging the point as the boundary point when α is more than 90 degrees, wherein the point is shown as a non-boundary point in fig. 2(a) and a boundary point in fig. 2(b), and counting the number n of all boundary points1And sorting the non-boundary points from large to small according to the curvature. Data points with curvature c less than 0.02 are low curvature points and data points with curvature c greater than 0.02 are high curvature points. Count the number n of high curvature points2Number of points of low curvature n3。
Step S04: according to the calculated boundary point number n1High curvature point n2Low number of curvature points n3Reduced ratio r to point cloud0And selecting a proper survival probability model according to the lowest sampling ratio k of the low-rate points. Fig. 3(a) shows a probabilistic model mode 1, in which high curvature points and boundary points are not reduced, but only low curvature points are reduced. The minimum point threshold value after the simplification in the mode is n1+n2+kr0n3. The probability modelThe survival probability of the boundary point corresponding to the high curvature point is 1, and the survival probabilities of the low curvature points are unified into (n)3-nr0)/n3. Fig. 3(b) shows a probabilistic model mode 2, in which the boundary points are not reduced, and the high curvature points and the low curvature points are reduced. With the highest probability of survival for high curvature points being 1. The minimum point threshold value after the simplification in the mode is n1+0.5n2(1+kr0)+kr0n3. In the probability model, the corresponding survival probability of the boundary points is 1, and the low-curvature survival probability is unified to kr0. The high curvature part corresponds to the linear function and corresponds to each coefficient as follows:
Figure BDA0002498265920000082
s is the minimum survival probability of the high curvature point, t is the slope corresponding to the high curvature section, and p is the intercept corresponding to the high curvature section.
The survival probabilities of all boundary points are consistent, the survival probabilities of all low-curvature points are consistent, the survival probability of a high-curvature part is changed along with the change of the curvature, and for a point sequence which is well arranged according to the curvature, the relation between the serial number corresponding to the high-curvature point and the survival probability can be expressed by a linear function. And the corresponding survival probability can be obtained according to the sequence number of each high-curvature point.
Fig. 3(c) shows a probabilistic model mode 3, in which the boundary points are not reduced, and the high curvature points and the low curvature points are reduced. Wherein the lowest survival probability of the high curvature point is k r0. The minimum point threshold value after the simplification in the mode is n1+0.5n2(3kr0+kr0)+kr0n3. In the probability model, the corresponding survival probability of the boundary points is 1, and the low-curvature survival probability is unified to kr0. The high curvature part corresponds to the linear function and corresponds to each coefficient as follows:
Figure BDA0002498265920000091
s is the minimum survival probability of the high curvature point, t is the slope corresponding to the high curvature section, and p is the intercept corresponding to the high curvature section.
FIG. 3(d) shows probabilistic model mode 4, in which the boundary points, high curvature points and low curvature points are reduced. Wherein the lowest survival probability of the high curvature point is kr0The highest survival probability is 3kr0In this mode, the minimum point threshold after the reduction is 3kr0n1+2kr0n2+kr0n3. In the probability model, the corresponding survival probability of the boundary point is (nr)0-2kn2r0-kn3r0)/n1The low curvature survival probability is unified as kr0. The high curvature part corresponds to the linear function and corresponds to each coefficient as follows:
Figure BDA0002498265920000092
t is the slope corresponding to the high curvature segment, and p is the intercept corresponding to the high curvature segment.
Step S05: based on the calculated survival probability, the point cloud starts to be condensed. And traversing each data point of the point cloud by using a multi-thread parallel computing technology, randomly generating a random number with the size of 0-1 each time, comparing the random number with the survival probability, and when the random number is smaller than the survival probability, keeping the point, otherwise, deleting the point. Finally, the final simplified point cloud data is obtained.
In order to verify the feasibility of the invention, the pump body original point cloud model shown in fig. 4 is reduced by 25% according to the steps S01 to S05. The simplified point cloud model is shown in fig. 5, and the local detail view is shown in fig. 6. It can be seen from fig. 5 and 6 that when the algorithm is used for reducing the point cloud, the number of points reserved by the high curvature point is more than that of the low curvature point, and the boundary part can be well reserved. The mesh model obtained by triangularizing the simplified point cloud is shown in fig. 7. Further, the method can well keep the characteristics of the original point cloud when the point cloud is simplified.
Although the embodiments of the present invention have been described above with reference to the accompanying drawings, the present invention is not limited to the above-described embodiments and application fields, and the above-described embodiments are illustrative, instructive, and not restrictive. Those skilled in the art, having the benefit of this disclosure, may effect numerous modifications thereto without departing from the scope of the invention as defined by the appended claims.

Claims (9)

1. A method of point cloud reduction based on survival probability, the method comprising the steps of:
reading original point cloud data, establishing a topological relation for the original point cloud data based on a kdtree algorithm, and acquiring all neighborhood points within a radius r range of each data point;
in the second step, based on a principal component analysis method, performing covariance analysis on each data point and its neighborhood points by using multi-thread parallel computation to obtain covariance matrixes, and respectively computing three eigenvalues lambda of the covariance matrixes1、λ2、λ3The data points correspond to curvatures of
Figure FDA0002498265910000011
In the third step, all data points are divided into boundary points or non-boundary points according to whether the data points are boundary points or not, the non-boundary points are sorted according to the curvature, the data points are divided into high curvature points and low curvature points according to a preset threshold value, and the number of the boundary points, the number of the high curvature points and the number of the low curvature points are respectively n1、n2、n3，
In the fourth step, n is added according to a predetermined reduction ratio1、n2、n3The number of the points to be deleted of the boundary points, the high curvature points and the low curvature points is calculated, wherein the low curvature points are deleted firstly, then the high curvature points and finally the boundary points are deleted, a survival probability model is determined based on the simplification ratio of the point cloud and the number of the boundary points, the high curvature points and the low curvature points, the survival probability is given to each boundary point, the high curvature points and the low curvature points, wherein the survival probability of the boundary points is the maximum, the survival probability of the non-boundary points is decreased from the maximum to the minimum along with the curvature,
in the fifth step: and traversing each data point of the point cloud based on multi-thread parallel computing, randomly generating a random number with the size of 0-1 each time, comparing the random number with the survival probability, reserving the data point when the random number is smaller than the survival probability, and deleting the data point to obtain the simplified point cloud data if the random number is not smaller than the survival probability.
2. The method of claim 1, wherein in the first step, the number of points n in the neighborhood of each data point is preferably countediWherein the radius r is 5mm when niIf the number of the data points is less than 5, the data points are judged to be outliers and deleted, and the total number of the points is changed into n after the neighborhood point searching operation is finished.
3. The method according to claim 1, wherein in the second step, each data point p and its neighborhood point p are calculated in parallel using multiple threads based on principal component analysisiCentralizing the data point p as a center, carrying out covariance analysis on the centralized point to obtain a covariance matrix A,
Figure FDA0002498265910000012
wherein n is the total point number, A is a symmetric semi-positive definite matrix, all the eigenvectors are real numbers, and the eigenvectors are mutually orthogonal.
4. The method of claim 1, wherein λ1Greater than λ2，λ2Greater than λ3。
5. The method according to claim 1, wherein in the third step, each data point boundary point determination includes projecting the data point and its neighboring points to a tangent plane where the data point is located, connecting each neighboring point with a central point located at the center to find a maximum included angle α, and determining the data point as a boundary point when α is greater than 90 °.
6. The method according to claim 1, wherein in the fourth step, only the low curvature points are reduced in the survival probability model, the boundary points correspond to the high curvature points with survival probability of 1, and the low curvature points with survival probability of (n)3-nr0)/n3The minimum point threshold value after the simplification is n1+n2+kr0n3Wherein n is1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (4) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points, and n is the total point number.
7. The method according to claim 1, wherein in the fourth step, the high curvature point and the low curvature point are reduced in the survival probability model, the high curvature point survival probability is 1, the boundary point survival probability is 1, and the low curvature survival probability is kr0The minimum point threshold value after the simplification is n1+0.5n2(1+kr0)+kr0n3，n1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
8. The method according to claim 1, wherein in the fourth step, the survival probability model reduces the high curvature point and the low curvature point, and the survival probability of the high curvature point is kr0Boundary point survival probability of 1 and low curvature survival probability of kr0The minimum point threshold value after the simplification is n1+0.5n2(3kr0+kr0)+kr0n3Wherein n is1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
9. The method according to claim 1, wherein in the fourth step, the survival probability model is refined by the boundary points, the high curvature points and the low curvature pointsHigh curvature point survival probability of kr0To 3kr0The boundary point survival probability is (nr)0-2kn2r0-kn3r0)/n1Low curvature survival probability of kr0The minimum point threshold value after the simplification is 3kr0n1+2kr0n2+kr0n3，n1Number of boundary points, n2Is a high number of curvature points, n3To a low number of curvature points, r0And (5) predetermining a reduction ratio for the point cloud, wherein k is the lowest sampling ratio of the low-curvature points.
CN202010427712.7A 2020-05-19 2020-05-19 Point cloud simplification method based on survival probability Active CN111652855B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010427712.7A CN111652855B (en) 2020-05-19 2020-05-19 Point cloud simplification method based on survival probability

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010427712.7A CN111652855B (en) 2020-05-19 2020-05-19 Point cloud simplification method based on survival probability

Publications (2)

Publication Number Publication Date
CN111652855A true CN111652855A (en) 2020-09-11
CN111652855B CN111652855B (en) 2022-05-06

Family

ID=72348338

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010427712.7A Active CN111652855B (en) 2020-05-19 2020-05-19 Point cloud simplification method based on survival probability

Country Status (1)

Country Link
CN (1) CN111652855B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112971756A (en) * 2021-02-07 2021-06-18 佛山科学技术学院 Speckle blood flow imaging method, electronic device, and computer-readable storage medium
CN113155054A (en) * 2021-04-15 2021-07-23 西安交通大学 Automatic three-dimensional scanning planning method for surface structured light
CN113269791A (en) * 2021-04-26 2021-08-17 西安交通大学 Point cloud segmentation method based on edge judgment and region growth
WO2021159838A1 (en) * 2020-10-12 2021-08-19 平安科技（深圳）有限公司 Method and apparatus for simplifying point cloud data, and storage medium and electronic device
JP2023529527A (en) * 2021-05-21 2023-07-11 ベイジン バイドゥ ネットコム サイエンス テクノロジー カンパニー リミテッド Method and apparatus for generating point cloud data

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150220812A1 (en) * 2013-12-20 2015-08-06 Visual Technology Services Limited Point cloud simplification
CN105069845A (en) * 2015-07-29 2015-11-18 南京信息工程大学 Point cloud simplification method based on curved surface change
US20160155264A1 (en) * 2014-11-28 2016-06-02 Fu Tai Hua Industry (Shenzhen) Co., Ltd. Electronic device and method for reducing point cloud
CN106355178A (en) * 2016-07-25 2017-01-25 北京航空航天大学 Method of massive points cloud adaptive simplification based on hierarchical clustering and topological connection model
CN106373118A (en) * 2016-08-30 2017-02-01 华中科技大学 A complex curved surface part point cloud reduction method capable of effectively keeping boundary and local features
CN108198244A (en) * 2017-12-20 2018-06-22 中国农业大学 A kind of Apple Leaves point cloud compressing method and device
CN108830931A (en) * 2018-05-23 2018-11-16 上海电力学院 A kind of laser point cloud compressing method based on dynamic grid k neighborhood search
CN109410342A (en) * 2018-09-28 2019-03-01 昆明理工大学 A kind of point cloud compressing method retaining boundary point
CN110807781A (en) * 2019-10-24 2020-02-18 华南理工大学 Point cloud simplification method capable of retaining details and boundary features

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150220812A1 (en) * 2013-12-20 2015-08-06 Visual Technology Services Limited Point cloud simplification
US20160155264A1 (en) * 2014-11-28 2016-06-02 Fu Tai Hua Industry (Shenzhen) Co., Ltd. Electronic device and method for reducing point cloud
CN105069845A (en) * 2015-07-29 2015-11-18 南京信息工程大学 Point cloud simplification method based on curved surface change
CN106355178A (en) * 2016-07-25 2017-01-25 北京航空航天大学 Method of massive points cloud adaptive simplification based on hierarchical clustering and topological connection model
CN106373118A (en) * 2016-08-30 2017-02-01 华中科技大学 A complex curved surface part point cloud reduction method capable of effectively keeping boundary and local features
CN108198244A (en) * 2017-12-20 2018-06-22 中国农业大学 A kind of Apple Leaves point cloud compressing method and device
CN108830931A (en) * 2018-05-23 2018-11-16 上海电力学院 A kind of laser point cloud compressing method based on dynamic grid k neighborhood search
CN109410342A (en) * 2018-09-28 2019-03-01 昆明理工大学 A kind of point cloud compressing method retaining boundary point
CN110807781A (en) * 2019-10-24 2020-02-18 华南理工大学 Point cloud simplification method capable of retaining details and boundary features

Non-Patent Citations (9)

* Cited by examiner, † Cited by third party
Title
BAO-QUAN SHI等: ""Adaptive simplification of point cloud using k-means clustering"", 《COMPUTER-AIDED DESIGN》 *
HONG JIANG等: ""A study and implementation on the data reduction based on the curvature of point clouds"", 《COMPUTER MODELLING & NEW TECHNOLOGIES》 *
LAN XIAO-QI等: ""An unorganized point cloud simplification based on boundary point extraction"", 《INTERNATIONAL SYMPOSIUM ON LIDAR AND RADAR MAPPING 2011: TECHNOLOGIES AND APPLICATIONS》 *
SHIFAN LIU等: ""An edge-sensitive simplification method for scanned point clouds"", 《MEASUREMENT SCIENCE AND TECHNOLOGY》 *
刘迎等: ""特征提取的点云自适应精简"", 《光学精密工程》 *
史宝全等: ""特征保持的点云精简技术研究"", 《西安交通大学学报》 *
宋大虎等: ""保持特征的点云迭代简化算法"", 《计算机应用研究》 *
常俊飞: ""三维点云数据的精简与配准算法研究"", 《中国优秀硕士学位论文全文数据库 基础科学辑》 *
常俊飞等: ""基于边界保留的点云精简算法研究"", 《测绘与空间地理信息》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021159838A1 (en) * 2020-10-12 2021-08-19 平安科技（深圳）有限公司 Method and apparatus for simplifying point cloud data, and storage medium and electronic device
CN112971756A (en) * 2021-02-07 2021-06-18 佛山科学技术学院 Speckle blood flow imaging method, electronic device, and computer-readable storage medium
CN113155054A (en) * 2021-04-15 2021-07-23 西安交通大学 Automatic three-dimensional scanning planning method for surface structured light
CN113269791A (en) * 2021-04-26 2021-08-17 西安交通大学 Point cloud segmentation method based on edge judgment and region growth
JP2023529527A (en) * 2021-05-21 2023-07-11 ベイジン バイドゥ ネットコム サイエンス テクノロジー カンパニー リミテッド Method and apparatus for generating point cloud data

Also Published As

Publication number Publication date
CN111652855B (en) 2022-05-06

Similar Documents

Publication Publication Date Title
CN111652855B (en) Point cloud simplification method based on survival probability
CN110599506B (en) Point cloud segmentation method for three-dimensional measurement of complex special-shaped curved surface robot
CN111080684B (en) Point cloud registration method for point neighborhood scale difference description
CN103914571B (en) Three-dimensional model search method based on mesh segmentation
CN108388902B (en) Composite 3D descriptor construction method combining global framework point and local SHOT characteristics
CN111723915B (en) Target detection method based on deep convolutional neural network
CN108920765B (en) Hypothetical plane fitting method based on building three-dimensional line segment model
US10580114B2 (en) Methods and systems for real time 3D-space search and point-cloud registration using a dimension-shuffle transform
CN115861397A (en) Point cloud registration method based on improved FPFH-ICP
CN112634457B (en) Point cloud simplification method based on local entropy of Hausdorff distance and average projection distance
Baheti et al. Federated Learning on Distributed Medical Records for Detection of Lung Nodules.
CN111783722B (en) Lane line extraction method of laser point cloud and electronic equipment
CN113963138A (en) Complete and accurate extraction method of three-dimensional laser point cloud characteristic point line
CN105809113A (en) Three-dimensional human face identification method and data processing apparatus using the same
CN117274339A (en) Point cloud registration method based on improved ISS-3DSC characteristics combined with ICP
CN111860359A (en) Point cloud classification method based on improved random forest algorithm
CN115830587A (en) Structural surface rapid automatic identification method based on high-precision point cloud data
CN109035311B (en) Automatic registration and internal fixation steel plate pre-bending modeling method for curved bone fracture
CN106980878B (en) Method and device for determining geometric style of three-dimensional model
CN111862176B (en) Three-dimensional oral cavity point cloud orthodontic front and back accurate registration method based on palatine fold
CN113435479A (en) Feature point matching method and system based on regional feature expression constraint
CN113159103A (en) Image matching method, image matching device, electronic equipment and storage medium
CN114022526B (en) SAC-IA point cloud registration method based on three-dimensional shape context
CN112967333B (en) Complex point cloud skeleton extraction method and system based on grading
CN111967365B (en) Image connection point extraction method and device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant