CN110837873A - Three-dimensional point cloud clustering algorithm - Google Patents
Three-dimensional point cloud clustering algorithm Download PDFInfo
- Publication number
- CN110837873A CN110837873A CN201911125618.XA CN201911125618A CN110837873A CN 110837873 A CN110837873 A CN 110837873A CN 201911125618 A CN201911125618 A CN 201911125618A CN 110837873 A CN110837873 A CN 110837873A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- clustering
- cloud data
- euclidean distance
- value
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an algorithm for three-dimensional point cloud clustering, which specifically comprises the following steps: s1, setting a cluster center set, importing a large amount of point cloud data collected by equipment into a clustering center set setting unit, selecting cluster centers Z1 and S2, calculating threshold values T and S3, and searching all cluster centers. The three-dimensional point cloud clustering algorithm can search a second clustering center Z2 in a large amount of point cloud data collected by equipment, traverse all mode samples obtained by given point cloud data to quickly obtain and determine a threshold value, form a communication region according to the distance, traverse the Euclidean distance between each sample in the point cloud and each clustering center sample in the current clustering center set through the point cloud data by the intra-clustering center positioning unit to quickly obtain the maximum minimum value to search a clustering center point set, provide a basis for realizing clustering and complete clustering of the point cloud data.
Description
Technical Field
The invention relates to the technical field of three-dimensional point cloud clustering, in particular to an algorithm for three-dimensional point cloud clustering.
Background
When point cloud data is obtained, some noise inevitably appears in the point cloud data due to the influence caused by equipment precision, experience and environment factors of an operator, the diffraction characteristics of electromagnetic waves, the surface property change of a measured object and the influence of the data splicing and registering operation process.
Before reconstructing a curved surface, the point cloud data needs to be clustered to ensure that the subsequent 3D reconstruction work is accurate and efficient, and the K-means algorithm is a method widely used at present, but has some limitations, for example, different initial value centers may cause the algorithm to fall into local optimum to cause the clustering result to be unstable.
The invention provides an algorithm for clustering according to a maximum and minimum distance value based on a set threshold, aiming at the problems of calculation and storage of a large amount of point cloud data, the novel clustering algorithm is adopted, the purpose is to ensure that the subsequent 3D reconstruction work is accurately and efficiently carried out, and because noise interference occurs to the collected point cloud data in the early stage, the accuracy is not high, and the difference with the actual result is large, therefore, the algorithm which is in line with the current business needs to be researched and developed aiming at business design.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a three-dimensional point cloud clustering algorithm, which solves the problems that the subsequent 3D reconstruction work is accurate and efficient by clustering point cloud data before reconstructing a curved surface, and the K-means algorithm is a method widely used at present but has some limitations, for example, different initial value centers can cause the algorithm to be trapped in local optimization to cause unstable clustering results and the like.
(II) technical scheme
In order to achieve the purpose, the invention is realized by the following technical scheme: an algorithm of three-dimensional point cloud clustering specifically comprises the following steps:
s1, setting a clustering center set: importing a large amount of point cloud data collected by equipment into a gathering center set setting unit, and selecting a clustering center Z1;
s2, calculating a threshold value T: before calculating the threshold, searching a second clustering center Z2 in a large amount of point cloud data collected by equipment, traversing the given point cloud data to obtain all pattern samples, calculating to obtain Euclidean distance data sets between all pattern samples and the clustering center Z1 samples to obtain the maximum Euclidean distance value, wherein an bounding box containing the point cloud data is a range in a spherical bounding box taking the value as the radius, and multiplying the value by a given parameter through a threshold T calculating unit to obtain a threshold T;
s3, finding all cluster centers: calculating Euclidean distance between each sample in the point cloud and each sample in the current clustering center set through traversing the point cloud data by the intra-clustering center positioning unit to obtain a group of Euclidean distance values, taking out the minimum Euclidean distance value in the group of data to obtain the minimum Euclidean distance value between all samples in the group of point cloud data and the current clustering center set, finding out the maximum Euclidean distance value from the group of data to obtain the maximum minimum value and a sample corresponding to the maximum minimum value, comparing the maximum Euclidean distance value with the threshold value T calculated by the threshold value T calculating unit before, if the maximum Euclidean distance value is greater than the threshold value, obtaining a new clustering center point, adding the corresponding sample into the clustering center set, continuing to search the maximum minimum value through the process of searching the maximum minimum value, then comparing the obtained new distance value with the threshold value, and stopping searching the clustering center until the maximum minimum value is smaller than the threshold value, finally, a clustering center set can be obtained;
s4, point cloud data classification: clustering the cluster center point set obtained in the step S3 by using the cluster center points in the cluster center point set and the threshold value obtained by calculation through a point cloud data classification unit, obtaining a group of Euclidean distance values by traversing the Euclidean distance value of each point and each cluster center in the point cloud data, and finding the minimum Euclidean distance value larger than the threshold value and the corresponding cluster center point thereof so as to realize clustering operation;
s5, algorithm detection: through the determination of the threshold in the step S2, a cluster central point set is searched in the step S3, and a result finally obtained by clustering the point cloud data is detected to be correct through a result detection output unit and then output.
Preferably, the algorithm comprises an intra-cluster center set setting unit, a threshold T calculating unit, an intra-cluster center positioning unit, a point cloud data classifying unit and a result detection output unit.
Preferably, the output end of the intra-cluster center set setting unit is connected with the input end of the threshold T calculation unit, and the output end of the threshold T calculation unit is connected with the input end of the intra-cluster center positioning unit.
Preferably, the output end of the intra-cluster central positioning unit is connected with the input end of the point cloud data classification unit, and the output end of the point cloud data classification unit is connected with the input end of the result detection output unit.
Preferably, the clustering center Z1 in S1 is the first pattern sample in the point cloud data.
Preferably, the second cluster center Z2 in S2 is the maximum euclidean distance value between all pattern samples in S1 and the cluster center Z1 sample.
Preferably, the maximum minimum value in S3 is the maximum euclidean distance value among the minimum euclidean distance values between all samples in the point cloud data and the current cluster center set.
Preferably, the clustering data in S3 is a set of all minimum euclidean distance values greater than a threshold.
(III) advantageous effects
The invention provides an algorithm for three-dimensional point cloud clustering. Compared with the prior art, the method has the following beneficial effects:
(1) the three-dimensional point cloud clustering algorithm can search a second clustering center Z2 in a large amount of point cloud data collected by equipment, traverse all mode samples obtained by given point cloud data to quickly obtain and determine a threshold value, form a communication region according to the distance, traverse the Euclidean distance between each sample in the point cloud and each clustering center sample in the current clustering center set through the point cloud data by the intra-clustering center positioning unit to quickly obtain the maximum minimum value to search a clustering center point set, provide a basis for realizing clustering and complete clustering of the point cloud data.
(2) The algorithm of the three-dimensional point cloud clustering is characterized in that actual data testing is carried out, comparison and optimization are carried out for multiple times, the accuracy of the algorithm is calculated and tested to reach commercial conditions through the current product landing, and the value of a clustering center can be accurately obtained.
(3) The three-dimensional point cloud clustering algorithm determines and searches a clustering center point set by providing a threshold, a connected region formed by clustering is beneficial to calculation, graph construction and the like of various subsequent point cloud data with characteristics of the three-dimensional point cloud, and a better clustering effect can be finally determined by continuously testing the weight and the detection effect of each parameter.
Drawings
FIG. 1 is a schematic block diagram of the system of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, an embodiment of the present invention provides a technical solution: an algorithm of three-dimensional point cloud clustering specifically comprises the following steps:
s1, setting a clustering center set: importing a large amount of point cloud data collected by equipment into a gathering center set setting unit, and selecting a clustering center Z1;
s2, calculating a threshold value T: before calculating the threshold, searching a second clustering center Z2 in a large amount of point cloud data collected by equipment, traversing the given point cloud data to obtain all pattern samples, calculating to obtain Euclidean distance data sets between all pattern samples and the clustering center Z1 samples to obtain the maximum Euclidean distance value, wherein an bounding box containing the point cloud data is a range in a spherical bounding box taking the value as the radius, and multiplying the value by a given parameter through a threshold T calculating unit to obtain a threshold T;
s3, finding all cluster centers: calculating Euclidean distance between each sample in the point cloud and each sample in the current clustering center set through traversing the point cloud data by the intra-clustering center positioning unit to obtain a group of Euclidean distance values, taking out the minimum Euclidean distance value in the group of data to obtain the minimum Euclidean distance value between all samples in the group of point cloud data and the current clustering center set, finding out the maximum Euclidean distance value from the group of data to obtain the maximum minimum value and a sample corresponding to the maximum minimum value, comparing the maximum Euclidean distance value with the threshold value T calculated by the threshold value T calculating unit before, if the maximum Euclidean distance value is greater than the threshold value, obtaining a new clustering center point, adding the corresponding sample into the clustering center set, continuing to search the maximum minimum value through the process of searching the maximum minimum value, then comparing the obtained new distance value with the threshold value, and stopping searching the clustering center until the maximum minimum value is smaller than the threshold value, finally, a clustering center set can be obtained;
s4, point cloud data classification: clustering the cluster center point set obtained in the step S3 by using the cluster center points in the cluster center point set and the threshold value obtained by calculation through a point cloud data classification unit, obtaining a group of Euclidean distance values by traversing the Euclidean distance value of each point and each cluster center in the point cloud data, and finding the minimum Euclidean distance value larger than the threshold value and the corresponding cluster center point thereof so as to realize clustering operation;
s5, algorithm detection: through the determination of the threshold in the step S2, a cluster central point set is searched in the step S3, and a result finally obtained by clustering the point cloud data is detected to be correct through a result detection output unit and then output.
The algorithm comprises an intra-cluster central set setting unit, a threshold T calculating unit, an intra-cluster central positioning unit, a point cloud data classifying unit and a result detection output unit, wherein the output end of the intra-cluster central set setting unit is connected with the input end of the threshold T calculating unit, the output end of the threshold T calculating unit is connected with the input end of the intra-cluster central positioning unit, the output end of the intra-cluster central positioning unit is connected with the input end of the point cloud data classifying unit, and the output end of the point cloud data classifying unit is connected with the input end of the result detection output unit.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (8)
1. An algorithm of three-dimensional point cloud clustering is characterized in that: the method specifically comprises the following steps:
s1, setting a clustering center set: importing a large amount of point cloud data collected by equipment into a gathering center set setting unit, and selecting a clustering center Z1;
s2, calculating a threshold value T: before calculating the threshold, searching a second clustering center Z2 in a large amount of point cloud data collected by equipment, traversing the given point cloud data to obtain all pattern samples, calculating to obtain Euclidean distance data sets between all pattern samples and the clustering center Z1 samples to obtain the maximum Euclidean distance value, wherein an bounding box containing the point cloud data is a range in a spherical bounding box taking the value as the radius, and multiplying the value by a given parameter through a threshold T calculating unit to obtain a threshold T;
s3, finding all cluster centers: calculating Euclidean distance between each sample in the point cloud and each sample in the current clustering center set through traversing the point cloud data by the intra-clustering center positioning unit to obtain a group of Euclidean distance values, taking out the minimum Euclidean distance value in the group of data to obtain the minimum Euclidean distance value between all samples in the group of point cloud data and the current clustering center set, finding out the maximum Euclidean distance value from the group of data to obtain the maximum minimum value and a sample corresponding to the maximum minimum value, comparing the maximum Euclidean distance value with the threshold value T calculated by the threshold value T calculating unit before, if the maximum Euclidean distance value is greater than the threshold value, obtaining a new clustering center point, adding the corresponding sample into the clustering center set, continuing to search the maximum minimum value through the process of searching the maximum minimum value, then comparing the obtained new distance value with the threshold value, and stopping searching the clustering center until the maximum minimum value is smaller than the threshold value, finally, a clustering center set can be obtained;
s4, point cloud data classification: clustering the cluster center point set obtained in the step S3 by using the cluster center points in the cluster center point set and the threshold value obtained by calculation through a point cloud data classification unit, obtaining a group of Euclidean distance values by traversing the Euclidean distance value of each point and each cluster center in the point cloud data, and finding the minimum Euclidean distance value larger than the threshold value and the corresponding cluster center point thereof so as to realize clustering operation;
s5, algorithm detection: through the determination of the threshold in the step S2, a cluster central point set is searched in the step S3, and a result finally obtained by clustering the point cloud data is detected to be correct through a result detection output unit and then output.
2. The algorithm for three-dimensional point cloud clustering according to claim 1, comprising an intra-cluster center set setting unit, a threshold T calculating unit, an intra-cluster center positioning unit, a point cloud data classifying unit and a result detection output unit.
3. The algorithm for three-dimensional point cloud clustering according to claim 2, wherein: the output end of the intra-cluster center set setting unit is connected with the input end of the threshold T calculating unit, and the output end of the threshold T calculating unit is connected with the input end of the intra-cluster center positioning unit.
4. The algorithm for three-dimensional point cloud clustering according to claim 2, wherein: the output end of the intra-cluster central positioning unit is connected with the input end of the point cloud data classification unit, and the output end of the point cloud data classification unit is connected with the input end of the result detection output unit.
5. The algorithm for three-dimensional point cloud clustering according to claim 1, wherein: the clustering center Z1 in S1 is the first pattern sample in the point cloud data.
6. The algorithm for three-dimensional point cloud clustering according to claim 1, wherein: the second cluster center Z2 in S2 is the maximum euclidean distance value between all pattern samples in S1 and the cluster center Z1 sample.
7. The algorithm for three-dimensional point cloud clustering according to claim 1, wherein: and the maximum minimum value in the step S3 is the maximum euclidean distance value among the minimum euclidean distance values of all samples in the point cloud data and the current clustering center set.
8. The algorithm for three-dimensional point cloud clustering according to claim 1, wherein: the cluster data in S3 is a set of all minimum euclidean distance values greater than a threshold.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911125618.XA CN110837873A (en) | 2019-11-18 | 2019-11-18 | Three-dimensional point cloud clustering algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911125618.XA CN110837873A (en) | 2019-11-18 | 2019-11-18 | Three-dimensional point cloud clustering algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110837873A true CN110837873A (en) | 2020-02-25 |
Family
ID=69576608
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911125618.XA Pending CN110837873A (en) | 2019-11-18 | 2019-11-18 | Three-dimensional point cloud clustering algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110837873A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103870845A (en) * | 2014-04-08 | 2014-06-18 | 重庆理工大学 | Novel K value optimization method in point cloud clustering denoising process |
| CN104376057A (en) * | 2014-11-06 | 2015-02-25 | 南京邮电大学 | Self-adaptation clustering method based on maximum distance, minimum distance and K-means |
| CN105548066A (en) * | 2015-12-11 | 2016-05-04 | 贵州中烟工业有限责任公司 | Method and system for distinguishing colloid types |
| CN109359679A (en) * | 2018-10-10 | 2019-02-19 | 洪月华 | Distributed traffic big data parallel clustering method suitable for wide area network |
| CN110147815A (en) * | 2019-04-10 | 2019-08-20 | 深圳市易尚展示股份有限公司 | Multiframe point cloud fusion method and device based on K mean cluster |
-
2019
- 2019-11-18 CN CN201911125618.XA patent/CN110837873A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103870845A (en) * | 2014-04-08 | 2014-06-18 | 重庆理工大学 | Novel K value optimization method in point cloud clustering denoising process |
| CN104376057A (en) * | 2014-11-06 | 2015-02-25 | 南京邮电大学 | Self-adaptation clustering method based on maximum distance, minimum distance and K-means |
| CN105548066A (en) * | 2015-12-11 | 2016-05-04 | 贵州中烟工业有限责任公司 | Method and system for distinguishing colloid types |
| CN109359679A (en) * | 2018-10-10 | 2019-02-19 | 洪月华 | Distributed traffic big data parallel clustering method suitable for wide area network |
| CN110147815A (en) * | 2019-04-10 | 2019-08-20 | 深圳市易尚展示股份有限公司 | Multiframe point cloud fusion method and device based on K mean cluster |
Non-Patent Citations (1)
| Title |
|---|
| CSLINBO: "《聚类算法——最大最小距离算法(python实现)》", 《HTTPS://BLOG.CSDN.NET/HA_HHA/ARTICLE/DETAILS/79108137》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111061821B (en) | Low-voltage distribution network topology verification method and system based on improved k-value clustering algorithm | |
| KR20160019897A (en) | Fast grouping of time series | |
| CN107179310B (en) | Raman spectrum characteristic peak recognition methods based on robust noise variance evaluation | |
| CN111028016A (en) | Sales data prediction method and device and related equipment | |
| CN108667684A (en) | A kind of data flow anomaly detection method based on partial vector dot product density | |
| CN116522268B (en) | Line loss anomaly identification method for power distribution network | |
| CN109521725A (en) | The method, apparatus and equipment and machine readable media of detection abnormal data | |
| CN111046532A (en) | Coherent cluster recognition method based on elbow criterion | |
| CN114978877B (en) | Abnormality processing method, abnormality processing device, electronic equipment and computer readable medium | |
| CN108154173A (en) | A kind of oil-water interface measuring device of crude oil storage tank and method | |
| CN109405979B (en) | A kind of Michelson's interferometer image width of fringe detection method and system | |
| CN114898118A (en) | Automatic statistical method and system for power transmission line house removal amount based on multi-source point cloud | |
| CN108254038B (en) | A kind of crude oil storage tank oil-water interfaces data go the calculation method of puppet and level gauging | |
| CN113191224A (en) | Unmanned aerial vehicle signal extraction and identification method and system | |
| CN117495891A (en) | Point cloud edge detection method and device and electronic equipment | |
| CN109284409A (en) | Picture group geographic positioning based on extensive streetscape data | |
| CN110837873A (en) | Three-dimensional point cloud clustering algorithm | |
| CN111343664B (en) | User positioning method, device, equipment and medium | |
| CN116226468A (en) | Service data storage management method based on gridding terminal | |
| CN109739840A (en) | Data processing empty value method, apparatus and terminal device | |
| CN115934699A (en) | Abnormal data screening method and device, electronic equipment and storage medium | |
| CN115413026A (en) | Base station selection method, system, equipment and storage medium based on clustering algorithm | |
| CN103853817A (en) | Method for detecting space singular point of mass statistical data based on GIS (Geographic Information System) | |
| CN110045354B (en) | Method and device for evaluating radar performance | |
| CN112882822B (en) | Method, apparatus, device and storage medium for generating load prediction model |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200225 |
|
| RJ01 | Rejection of invention patent application after publication |