CN101447030A - Method for quickly querying scattered point cloud local profile reference data - Google Patents
Method for quickly querying scattered point cloud local profile reference data Download PDFInfo
- Publication number
- CN101447030A CN101447030A CNA2008101597476A CN200810159747A CN101447030A CN 101447030 A CN101447030 A CN 101447030A CN A2008101597476 A CNA2008101597476 A CN A2008101597476A CN 200810159747 A CN200810159747 A CN 200810159747A CN 101447030 A CN101447030 A CN 101447030A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- hollow ball
- node
- reference data
- point
- 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
Abstract
The invention provides a method for quickly querying scattered point cloud local profile reference data used for product reverse engineering. The method is characterized in that firstly, the scattered point cloud file output by product digital equipment is read in a memory and a linear table memory structure therefore is established; a scattered point cloud dynamic space index structure is established on the basis of R<*>-tree; an initial hollow ball is constructed according to a leaf node where the target point is arranged; the leaf node crossed with the hollow ball is queried, thus querying the local profile reference data of the target point; if the queried point number is less than a valve value k set by the user, the internal radius and the external radius of the hollow ball are extended and the local profile reference data of the target point is re-gained, otherwise k local profile reference data used as target points closely to the target point are taken. The experiment result indicates that the method can query the local profile reference data of various complex profile scattered point cloud, has the advantages of strong stability and high query efficiency, and is easy to be extended as a method for querying local profile reference data of other space objects.
Description
Technical field
The invention provides a kind of method for quickly querying of product reverse-engineering scattered point cloud local profile reference data, belong to product reverse Engineering Technology field.
Background technology
Extensively adopt dispersion point cloud surface-type feature analytical technology that dispersion point cloud is analyzed in the reverse-engineering, and with analysis result as curve reestablishing, simplify, the reference data of technology such as segmentation.Continuous development along with measuring technique, can high-level efficiency, collection model shape data accurately, but the data of gathering do not have tangible topological relation, and data volume is big, therefore, scattered point cloud local profile reference data can be obtained rapidly and accurately and quality and the speed that the dispersion point cloud surface-type feature is analyzed will be directly influenced.
Existing technical literature retrieval is found, Xiong Bangshu etc. are at academic journal " computer-aided design (CAD) and graphics journal " 2004,16 (7), the paper of delivering on the P909-911 " k of three-dimensional scattered data being nearest-neighbor fast search algorithm " proposes the scattered point cloud local profile reference data querying method based on the space lattice division, Zhou Rurong etc. are at academic journal " software journal " 2001,12 (2), the paper of delivering on the P249-255 " the curve reestablishing algorithm research of mass data point " proposes the scattered point cloud local profile reference data querying method based on hash table, and Shi Guirong pitches the local profile reference data of subdivision tree query dispersion point clouds based on space eight in its PhD dissertation " reverse engineering Geometric Modeling gordian technique ".These methods are pressed predetermined space subdivision rule cut-point cloud space, the spatial sub-regions that generates is carried out encoding process, set up static data directory structure, when data are carried out query manipulation, the spatial sub-regions coded message of direct computational data point correspondence in index structure, according to coded message, can reduce the hunting zone of data in a cloud static data structure significantly, realize impact point location and local profile reference data inquiry fast.The defective that said method exists is: based on the disposal route of static data structure, the data structure that is adopted is not supported operations such as Data Dynamic insertion, deletion, adjustment, and too rely on the spatial distribution state of data, when data along a certain dimension over-tilting, even when deteriorating to two dimension or one dimension state, can cause the data structure rapid deterioration, have a strong impact on the search efficiency of scattered point cloud local profile reference data.
In sum, existing Data Access Technology is applied in the product reverse-engineering data processing, only be fit to the inquiry distribution local profile reference data of dispersion point cloud comparatively uniformly, the dispersion point clouds different for data scale, topological structure and profile complexity lack versatility, cause the stability of follow-up data query manipulation and search efficiency can't satisfy engineering demand.
Summary of the invention
For overcoming the existing deficiency of scattered point cloud local profile reference data inquiring technology in product reverse-engineering data processing, the object of the present invention is to provide a kind of method for quickly querying of scattered point cloud local profile reference data, make it in time to respond for the query manipulation of all kinds of complex profile data, have the advantages that stability is strong, search efficiency is high, and easily expand to the querying method of the local profile reference data of other spatial object.Its technical scheme is as follows:
A kind of method for quickly querying that is used for the scattered point cloud local profile reference data of product reverse-engineering, comprise following steps successively: 1) the dispersion point cloud file with the output of product digitizer is read in the memory device, and set up the linear list storage organization for it, based on R
*-tree sets up dispersion point cloud dynamic space index structure; 2) adopt depth-first traversal method traversal dispersion point cloud dynamic space index structure, query aim point place leaf node, make up initial hollow ball according to this leaf node, the centre of sphere is an impact point, internal diameter is zero, and external diameter is MBR (Minimum Boundary Rectangular) the circumsphere radius of its place leaf node; 3) calculate minor increment and the ultimate range of node MBR to the centre of sphere, if minor increment less than hollow ball external diameter and ultimate range greater than the hollow ball internal diameter, show that this node and hollow ball intersect, the crossing leaf node of inquiry and hollow ball is gathered, with it as query region; 4) data point that each leaf node comprises in the traversal queries zone and calculate its distance to impact point p is if this distance less than the hollow ball external diameter and greater than the hollow ball internal diameter, is then added this data point among neighbour's point sequence L; 5) if sequence L in number of data points n less than k, then expand the interior external radius of hollow ball and return step 3) continuation inquiry, add among neighbour's point sequence L otherwise from sequence L, choose, finish the inquiry of local profile reference data apart from k-n nearer data point of impact point.
For realizing goal of the invention, the method for quickly querying of described scattered point cloud local profile reference data, in step 1), the index node of definition dispersion point cloud dynamic space index structure comprises root node, inner node and leaf node, adopt the k-means algorithm that dispersion point cloud is carried out cluster analysis, set up dispersion point cloud dynamic space index structure, method is: 1. choose the initial sub-clustering center of k node MBR center as index node arbitrarily; 2. each non-sub-clustering axial cable is drawn node and be inserted in the nearest sub-clustering in its MBR center, and the sub-clustering result specification is turned to four-dimensional some object p
i(x
i, y
i, z
i, r
i), (x wherein
i, y
i, z
i) be the MBR centre coordinate, r is the circumsphere radius of MBR; 3. for N index node in the same bunch node, its four-dimensional normalized coordinates is p
i(x
i, y
i, z
i, r
i) (i=1 ..., N), will be with r
iFor weight factor calculates gained node center as new sub-clustering center; 4. new sub-clustering center and last sub-clustering center are compared, if identical then finish sub-clustering, otherwise replace original sub-clustering center with new sub-clustering center, return step and 2. continue sub-clustering.
For realizing goal of the invention, the method for quickly querying of described scattered point cloud local profile reference data, calculate minor increment minDist and the ultimate range max Dist of each node MBR to impact point p respectively by formula (1) and formula (2), formula (1) is: and minDist (p, R)=‖ p
i-r
i‖, wherein
p
iBe a p i dimension coordinate value, (u
i, v
i) be MBR minimum, the maximum vertex of leaf node R;
Formula (2) is: and max Dist (p, R)=‖ p
i-r
i‖, wherein
p
iBe a p i dimension coordinate value, (u
i, v
i) be MBR minimum, the maximum vertex of leaf node R.
Be the realization goal of the invention, the method for quickly querying of described scattered point cloud local profile reference data, the interior external radius of expansion hollow ball in step 5), specifically: if the number of data points n among neighbour's point sequence L is less than k, then with current hollow ball S (p, r
1, r
2) expand to hollow ball S ' (p, r
1', r
2'), its internal diameter r
1'=r
2, external diameter
Realize the self-adaptation expansion of hollow ball.
The present invention compared with prior art has the following advantages:
1) makes up its interior external radius of hollow ball and dynamic expansion, accurately obtained the local profile reference data of impact point;
2) adopt the depth-first traversal method to obtain the leaf node that intersects with hollow ball, therefrom the local profile reference data of query aim point has effectively dwindled query context, has improved search efficiency, on average improves 20%-50%;
3) replace local point set to organize the dispersion point cloud topological relation with minimum area-encasing rectangle, can conveniently expand to the local profile reference data querying method of other spatial objects such as triangle gridding.
Description of drawings
Fig. 1 is the method for quickly querying program flow diagram of scattered point cloud local profile reference data of the present invention.
Fig. 2-Fig. 6 is each layer of dynamic space index structure node MBR illustraton of model that the present invention sets up the Venus dispersion point cloud.
Fig. 7 makes up initial hollow ball synoptic diagram in the step 2 of the present invention.
Fig. 8 obtains the interior neighbour of query region to put synoptic diagram in the step 3 of the present invention.
Fig. 9 is step 4 a dynamic expansion hollow ball synoptic diagram of the present invention.
Figure 10 is the product profile dispersion point cloud that is adopted in the invention process case.
Figure 11 and Figure 12 are the method for quickly querying efficiency analysis figure of scattered point cloud local profile reference data of the present invention.
Specific implementation method
The present invention adopts c programming language to realize the method for quickly querying of scattered point cloud local profile reference data, and the invention will be further described below in conjunction with accompanying drawing.
Fig. 1 is the method for quickly querying program realization flow figure of scattered point cloud local profile reference data of the present invention.Local profile reference data polling routine based on dispersion point cloud dynamic space index structure comprises dispersion point cloud dynamic space index structure construction procedures 1; Make up initial hollow ball program 2; Determine query region program 3; The neighbour who obtains in the query region puts program 4; Self-adaptation dynamic expansion hollow ball program 5.Wherein, dispersion point cloud dynamic space index structure construction procedures 1 is expressed as four-dimensional some object with each node unification of index structure, adopts the k-means algorithm that dispersion point cloud is carried out cluster analysis, sets up dispersion point cloud dynamic space index structure.Make up initial hollow ball program 2 and make up initial hollow ball according to impact point place leaf node.Determine the leaf node set that 3 inquiries of query region program and hollow ball intersect, it is put query region as the neighbour.Obtaining neighbour in the query region puts program 4 and it is added in neighbour's point sequence to the distance of impact point according to each point in the query region.If the neighbour who inquires counts less than threshold value, then call self-adaptation dynamic expansion hollow ball program 5, the interior external radius of expansion hollow ball, return information 3.
Fig. 2-Fig. 6 is each layer node MBR illustraton of model that the present invention calls the dynamic space index structure that Venus dispersion point cloud that 1 pair of dispersion point cloud dynamic space index structure construction procedures records by laser scanner sets up.The number of data points of testing used some cloud is 2.6218 ten thousand, and the minimum child node of the indexing parameter node that is adopted is counted m=8, maximum child node is counted M=20, and node inserts number of times R=17 again, and the dispersion point cloud dynamic space index structure structure time is 5.247 seconds.Wherein Fig. 2 has shown the MBR of the root node of Venus dispersion point cloud dynamic space index structure, Fig. 3 has shown the MBR of the inner node of the first floor, Fig. 4) shown the MBR of the inner node of the second layer, Fig. 5 has shown the MBR of leaf node, Fig. 6 has shown the Venus dispersion point cloud, promptly original Venus model outer profile data.This illustraton of model shows, adopts the dynamic space index structure can accurately realize the space clustering of extensive dispersion point cloud in the product reverse engineering is divided, and has the higher data adaptivity.
Fig. 7 is that the present invention makes up initial hollow ball synoptic diagram.Adopt depth-first traversal method traversal dispersion point cloud dynamic space index structure, query aim point place leaf node N, making up the centre of sphere is impact point p, internal diameter r
1Be zero, external diameter r
2Initial hollow ball S (p, r for the MBR circumsphere radius of leaf node N
1, r
2).
Fig. 8 is that the present invention determines the query region synoptic diagram.Adopt the depth-first traversal method to obtain the leaf node of dispersion point cloud space index structure, calculate minor increment min Dist and the ultimate range max Dist of each node MBR respectively to impact point p by formula (1) and formula (2).If the MBR of leaf node to the minor increment of impact point less than hollow ball external diameter and ultimate range greater than the hollow ball internal diameter, i.e. min Dist＜r
2And max Dist 〉=r
1, showing that then this leaf node and hollow ball intersect, will gather as query region with the leaf node that hollow ball intersects.
min?Dist(p，R)＝‖p
i-r
i‖ (1)
Wherein
p
iBe a p i dimension coordinate value, (u
i, v
i) be MBR minimum, the maximum vertex of leaf node R.
max?Dist(p，R)＝‖p
i-r
i‖ (2)
Wherein
p
iBe a p i dimension coordinate value, (u
i, v
i) be MBR minimum, the maximum vertex of leaf node R.
Calculate the distance of interior each data point of query region, if this distance value less than the hollow ball external diameter and greater than the hollow ball internal diameter, then adds this point among neighbour's point sequence L to the hollow ball centre of sphere.
Fig. 9 is a self-adaptation expansion hollow ball synoptic diagram of the present invention.If the neighbour who inquires counts less than user's preset threshold, then with current hollow ball S (p, r
1, r
2) expand to hollow ball S ' (p, r
1', r
2'), its internal diameter r
1'=r
2, external diameter
Determine and hollow ball S ' (p, r
1', r
2') crossing query region, obtain S ' (p, r
1', r
2') in data point, and the k-n that the distance impact point is a nearer data point adds among neighbour's point sequence L, is k as if number of data points among the L, then stops the expansion of hollow ball, finishes the inquiry of local profile reference data; Otherwise continue the expansion hollow ball.Experiment showed, that usually hollow ball being expanded 1-2 time just can satisfy search request.By external radius in the expansion hollow ball, can effectively guarantee the accuracy of local profile reference data search algorithm.
The dispersion point cloud that Figure 10 a-Figure 10 f is in the implementation case to be adopted, these clouds there are differences on data scale or DATA DISTRIBUTION, and its mid point cloud a is a scan-line data, and counting is 666; The data point of point cloud b and c evenly distributes, and counting is respectively 5,070 and 10,073; The data point of point cloud d is the uneven distribution state, and counting is 20,6218; Point cloud e is the scan-line data of closure profile, and counting is 34,834; Point cloud f has endoporus, the data state that is evenly distributed, and counting is 45,075.These dispersion point clouds are set up the dynamic space index structure respectively, and the improve parameter unification of its dynamic space index structure is taken as m=8, M=20.
Figure 11 obtains the used time diagram of point cloud local profile reference data shown in Figure 10 e for this paper adopts this paper algorithm, and counting of its mid point cloud is 34834, and the span of k is 10-50.Figure 12 inquires about the used time for the dispersion point cloud that difference is counted carries out local profile reference data, and wherein the k value gets 20, and some cloud point number is respectively 666,5070,10073,206218,34834,45075.As can be seen from the figure, the local profile reference data of this paper algorithm when different the counting of identical count different value of K and identical k value obtain efficient only with the linear growth of the growth of cloud data amount, irrelevant with the cloud data distribution situation.
Claims (4)
1, a kind of method for quickly querying of scattered point cloud local profile reference data is characterized in that comprising following steps: 1) the dispersion point cloud file with the output of product digitizer is read in the memory device, and sets up the linear list storage organization for it, based on R
*-tree sets up dispersion point cloud dynamic space index structure; 2) adopt depth-first traversal method traversal dispersion point cloud dynamic space index structure, query aim point place leaf node makes up initial hollow ball according to this leaf node, and the centre of sphere is an impact point, and internal diameter is zero, and external diameter is the MBR circumsphere radius of its place leaf node; 3) calculate minor increment and the ultimate range of node MBR to the centre of sphere, if minor increment less than hollow ball external diameter and ultimate range greater than the hollow ball internal diameter, show that this node and hollow ball intersect, the crossing leaf node of inquiry and hollow ball is gathered, with it as query region; 4) data point that each leaf node comprises in the traversal queries zone and calculate its distance to impact point p is if this distance less than the hollow ball external diameter and greater than the hollow ball internal diameter, is then added this data point among neighbour's point sequence L; 5) if sequence L in number of data points n less than k, then expand the interior external radius of hollow ball and return step 3) continuation inquiry, add among neighbour's point sequence L otherwise from sequence L, choose, finish the inquiry of local profile reference data apart from k-n nearer data point of impact point.
2, the method for quickly querying of scattered point cloud local profile reference data as claimed in claim 1, it is characterized in that: in step 1), the index node of definition dispersion point cloud dynamic space index structure comprises root node, inner node and leaf node, adopt the k-means algorithm that dispersion point cloud is carried out cluster analysis, set up dispersion point cloud dynamic space index structure, method is: 1. choose the initial sub-clustering center of k node MBR center as index node arbitrarily; 2. each non-sub-clustering axial cable is drawn node and be inserted in the nearest sub-clustering in its MBR center, and the sub-clustering result specification is turned to four-dimensional some object p
i(x
i, y
i, z
i, r
i), (x wherein
i, y
i, z
i) be the MBR centre coordinate, r is the circumsphere radius of MBR; 3. for N index node in the same bunch node, its four-dimensional normalized coordinates is p
i(x
i, y
i, z
i, r
i) (i=1 ..., N), will be with r
iFor weight factor calculates gained node center as new sub-clustering center; 4. new sub-clustering center and last sub-clustering center are compared, if identical then finish sub-clustering, otherwise replace original sub-clustering center with new sub-clustering center, return step and 2. continue sub-clustering.
3, the method for quickly querying of scattered point cloud local profile reference data as claimed in claim 1, it is characterized in that: in step 3), calculate minor increment min Dist and the ultimate range max Dist of each node MBR respectively by formula (1) and formula (2) to impact point p
Formula (1) is: min Dist (p, R)=|| p
i-r
i||, wherein
p
iBe a p i dimension coordinate value, (u
i, v
i) be MBR minimum, the maximum vertex of leaf node R;
Formula (2) is: max Dist (p, R)=|| p
i-r
i||, wherein
p
iBe a p i dimension coordinate value, (u
i, v
i) be MBR minimum, the maximum vertex of leaf node R.
4, the method for quickly querying of scattered point cloud local profile reference data as claimed in claim 1 is characterized in that: in step 5), if the number of data points n among neighbour's point sequence L is less than k, then with current hollow ball S (p, r
1, r
2) expand to hollow ball S ' (p, r
1', r
2'), its internal diameter r
1'=r
2, external diameter
Realize the self-adaptation expansion of hollow ball and return step 2), continue inquiry; If count greater than k among the L, then that the distance impact point is a nearer k-n data point is added among neighbour's point sequence L; When number of data points equals k among the L, stop the expansion of hollow ball, finish the inquiry of local profile reference data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008101597476A CN101447030A (en) | 2008-11-12 | 2008-11-12 | Method for quickly querying scattered point cloud local profile reference data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008101597476A CN101447030A (en) | 2008-11-12 | 2008-11-12 | Method for quickly querying scattered point cloud local profile reference data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101447030A true CN101447030A (en) | 2009-06-03 |
Family
ID=40742702
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2008101597476A Pending CN101447030A (en) | 2008-11-12 | 2008-11-12 | Method for quickly querying scattered point cloud local profile reference data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101447030A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102508973A (en) * | 2011-11-10 | 2012-06-20 | 山东理工大学 | Rapid intersection method for STL (stereo lithography) models of products |
| CN103701466A (en) * | 2012-09-28 | 2014-04-02 | 上海市政工程设计研究总院（集团）有限公司 | Scattered point cloud compression algorithm based on feature reservation |
| CN105593907A (en) * | 2013-08-16 | 2016-05-18 | 界标制图有限公司 | Generating representations of recognizable geological structures from a common point collection |
| CN109299109A (en) * | 2018-11-09 | 2019-02-01 | 南京天辰礼达电子科技有限公司 | A kind of storage of detail design point cloud data, quickly load, the method inquired |
-
2008
- 2008-11-12 CN CNA2008101597476A patent/CN101447030A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102508973A (en) * | 2011-11-10 | 2012-06-20 | 山东理工大学 | Rapid intersection method for STL (stereo lithography) models of products |
| CN102508973B (en) * | 2011-11-10 | 2014-07-02 | 山东理工大学 | Rapid intersection method for STL (stereo lithography) models of products |
| CN103701466A (en) * | 2012-09-28 | 2014-04-02 | 上海市政工程设计研究总院（集团）有限公司 | Scattered point cloud compression algorithm based on feature reservation |
| CN105593907A (en) * | 2013-08-16 | 2016-05-18 | 界标制图有限公司 | Generating representations of recognizable geological structures from a common point collection |
| US10261217B2 (en) | 2013-08-16 | 2019-04-16 | Landmark Graphics Corporation | Generating representations of recognizable geological structures from a common point collection |
| CN109299109A (en) * | 2018-11-09 | 2019-02-01 | 南京天辰礼达电子科技有限公司 | A kind of storage of detail design point cloud data, quickly load, the method inquired |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tsai | Delaunay triangulations in TIN creation: an overview and a linear-time algorithm | |
| Shi et al. | Adaptive simplification of point cloud using k-means clustering | |
| CN104616349B (en) | Scattered point cloud data based on local surface changed factor simplifies processing method | |
| Doraiswamy et al. | Efficient algorithms for computing Reeb graphs | |
| JP2014002519A (en) | Spatiotemporal data management system, spatiotemporal data management method, and spatiotemporal data management program | |
| CN108932347B (en) | Spatial keyword query method based on social perception in distributed environment | |
| Schön et al. | Octree-based indexing for 3D pointclouds within an Oracle Spatial DBMS | |
| CN112765405B (en) | Method and system for clustering and inquiring spatial data search results | |
| CN104731984B (en) | Automobile wheel hub surface sampling point R tree overflow node incremental clustering optimization method | |
| CN105550368A (en) | Approximate nearest neighbor searching method and system of high dimensional data | |
| CN101447030A (en) | Method for quickly querying scattered point cloud local profile reference data | |
| CN103817938A (en) | Rapid bridging method of polygonal area containing plurality of holes | |
| CN114970216A (en) | Electromagnetic environment visual expression method based on subdivision grids | |
| CN114693855B (en) | Point cloud data processing method and device | |
| CN101510315B (en) | Method for establishing space index structure of product STL model | |
| CN108563915B (en) | Vehicle digital simulation test model construction system and method, and computer program | |
| CN107818338A (en) | A kind of method and system of building group pattern-recognition towards Map Generalization | |
| Li et al. | A new extracting algorithm of k nearest neighbors searching for point clouds | |
| CN105138607B (en) | A kind of KNN querying methods based on combination grain distributed memory grid index | |
| Zhao et al. | A new k nearest neighbours algorithm using cell grids for 3d scattered point cloud | |
| Zhou et al. | A multi-representation spatial data model | |
| CN116720632A (en) | Engineering construction intelligent management method and system based on GIS and BIM | |
| CN115408382B (en) | Vector space information gridding storage management method for large-scale satellite data | |
| CN106096618A (en) | The acquisition methods of three dimensional practicality local profile feature reference data | |
| CN110297874B (en) | Multi-scale road network skyline query method based on Voronoi |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Open date: 20090603 |