CN102831241A - Dynamic index multi-target self-adaptive construction method for product reverse engineering data - Google Patents

Dynamic index multi-target self-adaptive construction method for product reverse engineering data Download PDF

Info

Publication number
CN102831241A
CN102831241A CN2012103325379A CN201210332537A CN102831241A CN 102831241 A CN102831241 A CN 102831241A CN 2012103325379 A CN2012103325379 A CN 2012103325379A CN 201210332537 A CN201210332537 A CN 201210332537A CN 102831241 A CN102831241 A CN 102831241A
Authority
CN
China
Prior art keywords
node
index
bounding box
index structure
ode
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
Application number
CN2012103325379A
Other languages
Chinese (zh)
Inventor
孙殿柱
史阳
刘华东
李延瑞
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.)
Shandong University of Technology
Original Assignee
Shandong University of Technology
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 Shandong University of Technology filed Critical Shandong University of Technology
Priority to CN2012103325379A priority Critical patent/CN102831241A/en
Publication of CN102831241A publication Critical patent/CN102831241A/en
Pending legal-status Critical Current

Links

Images

Abstract

The invention provides a dynamic index multi-target self-adaptive construction method for product reverse engineering data, and the method is characterized by comprising the following steps of firstly reading a product reverse engineering data file, constructing an axial bounding box of each space target, constructing a data node corresponding to each space target according to the center and a circumscribed radius of the axial bounding box, storing the data node into a data node sequence, inserting each data node in the sequence into an index structure through steps such as selection of an insert position, forced re-insertion, fracturing of nodes, adjustment of node axial bonding box and the like, re-inserting the data node with large size of the axial bounding box into the index structure to further optimize the index structure, and realizing the construction of the dynamic index structure of the product reverse engineering data. Due to the adoption of the method, the space index structure of different complicated product reverse engineering data can be constructed, and characteristics of low parameter dependence, strong stability and high inquiring efficiency can be realized.

Description

Product reverse-engineering data dynamic index multiple target self adaptation construction method
Technical field
The present invention provides a kind of product reverse-engineering data dynamic index multiple target self adaptation construction method, belongs to product reverse Engineering Technology field.
Background technology
In product reverse Engineering Technology field, the data forms such as dispersion point cloud, the polygonal grid model that handled initial data is sampled and obtained typical from surface in kind, the core technology that curve reestablishing generation burst continuous curve surface is product reverse-engineering is carried out based on such initial data.Because dispersion point cloud, polygonal mesh and burst continuous curve surface these data forms show as the composite construction of extensive even magnanimity geometric objects, a kind of general and efficient index technology is built for these data types, it is significant for improving product reverse-engineering data-handling efficiency.
Existing literature is retrieved and found, the index technology of existing product reverse-engineering data is generally only applicable to certain specific data type.In dispersion point cloud processing, space Octree and K-D trees are the static index and dynamic index being most widely used respectively.All seas are in its Ph.D. Dissertation's " Research on Modeling Technology of Subdivision Surface "(Nanjing Aero-Space University, 2005)Middle use space Octree as triangle grid model space index structure, tri patch is inserted into the Octree of space by the position according to tri patch bounding box center, set up triangle grid model index structure, organize the neighbor relationships between tri patch, this method represents tri patch with tri patch bounding box center, tri patch position and shared area of space size can not be accurately reflected, accuracy is poor, reduce the quality of index structure and the space querying efficiency based on the structure.Wang Zhanli is in its Ph.D. Dissertation's " nc machining simulation technical research of Virtual manufacture "(Jilin University, 2007)It is middle that triangle grid model is surrounded using one big bounding box, it regard the bounding box as root index node, then tri patch therein is divided into two parts, each section is surrounded with a bounding box, each bounding box recursion is split again, until a bounding box is only comprising a tri patch, set up the non-equilibrium binary tree index structure of triangle grid model, this arrangement enhances the space querying efficiency of triangle grid model, but because the structure is non-equilibrium binary tree, therefore it is only applicable to the more uniform triangle grid model of distribution, when triangle grid model distribution density is uneven, easily there is a certain excessive phenomenon of branch's number of plies set, data structure is caused drastically to deteriorate, serious reduction search index efficiency.Grandson hall post et al. is in its scientific paper " the R*- tree nodes splitting algorithm based on four-dimension cluster "(Mechanical engineering journal, 2009,45(10):180-184)In R*- trees are improved, be allowed to that index dispersion point cloud, the data type such as polygonal mesh can be unified, then in its scientific paper " triangle Bezier curved surface fast algorithm of point of intersection "(Mechanical engineering journal, 2011,47(3):89-94)It is middle using improved R*- trees as the index structure of burst continuous curve surface to improve intersecting triangular Bézier patch search efficiency, but it is due to that improved R- trees employ k- means clustering algorithms in index node fission process, need user mutual setting cluster number of clusters, the difference of cluster number of clusters can cause the very big index node division result of difference, cause index structure and unstable properties, in addition k- means clustering algorithms are a kind of local search algorithms, it is excessively sensitive to initial value, result is divided using the optimal index node of climbing method iterative search, it is easily trapped into local extremum, it is difficult to the index node division result for obtaining global optimum, cause the advantage for failing to give full play to R*- trees.
In summary, the dynamic indexing structure of current product reverse-engineering data has had been provided with certain versatility, the composite construction of various types of geometric objects can be handled based on unified Indexing Mechanism, but still there are problems that data adaptability is poor, index performance is relatively low and system resources consumption relatively, be that product reverse-engineering data build stable, efficient Indexing Mechanism and turned into those skilled in the art's technical problem urgently to be resolved hurrily.
The content of the invention
To overcome the shortcomings of the Indexing Mechanism of existing product reverse-engineering data, present invention aims at provide a kind of product reverse-engineering data dynamic index multiple target self adaptation construction method, make it to index various types of reverse-engineering data, with the characteristics of stability is strong, efficiency data query is high, technical scheme is as follows:
A kind of product reverse-engineering data dynamic index multiple target self adaptation construction method, it is characterised in that comprise the steps of:First, product reverse-engineering data are read, the axial bounding box of each spatial object is set up, its corresponding Data Node are set up according to the center of axial bounding box and bounding polygon, and be stored in Data Node sequence;2nd, Data Node is inserted into index structure, Knots inserting to index structure is comprised the concrete steps that:1)For node selection insertion position;2)By Knots inserting to step 1)In obtained position;3)Knots inserting is made under node n ode, judge whether node n ode child node number is more than the maximum child node number of node, if greater than then to node n ode progress Overflow handlings, if node n ode is non-root index node and layer where the node carries out Overflow handling for the first time during a spatial object is inserted, a part of node is then selectively taken out in node n ode, in the layer that they are reinserted to index structure, node split is otherwise carried out;4)Adjust the axial bounding box of each node;3rd, the excessive axial bounding box of volume is reinserted into index structure, realizes the optimization of index structure;4th, based on product reverse-engineering data dynamic indexing structure, the topological neighbor inquiry of dispersion point cloud, polygonal mesh and burst continuous curve surface is realized.
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method, in step one, read reverse-engineering data file, it is the axial bounding box of unit 1 and each bar rib each parallel to reference axis if setting up the length of side if spatial object is scattered points centered on the point, the axial bounding box for surrounding grid vertex just is set up if spatial object is polygonal mesh, the axial bounding box for surrounding its control vertex just is set up if spatial object is patch surface piece, set up the corresponding Data Node of each spatial object, and it is deposited into Data Node sequence, Data Node includes the information and corresponding axial bounding box information of spatial object.
To realize goal of the invention, Data Node in step 2, is inserted into index structure, method is by described product reverse-engineering data dynamic index multiple target self adaptation construction method:Node includes index node and Data Node, index node includes root index node, internal index node and leaf index node, the superiors' node of index structure is that root index node, orlop node are that leaf index node, remaining node are internal index node, definition
Figure 775570DEST_PATH_IMAGE001
For node maximum child node number (
Figure 989251DEST_PATH_IMAGE001
For the integer more than 2),
Figure 469911DEST_PATH_IMAGE002
For the minimum child node number of node (
Figure 762352DEST_PATH_IMAGE002
For less than or equal to/ 2 integer), in addition to root index node, the child node number of each index node, which is respectively less than, to be equal to
Figure 251419DEST_PATH_IMAGE001
And be more than or equal to
Figure 902980DEST_PATH_IMAGE002
;The axial bounding box of each node surrounds all child nodes of the node just in index structure.
To realize goal of the invention, the step of described product reverse-engineering data dynamic index multiple target self adaptation construction method in step 2 is node selection insertion position is specifically:1)It is current_node to make current node, and sky is returned if index structure is sky, and it is index structure root index node otherwise to make current_node;2)It is level to make the number of plies that node will insert, and level is the leaf layer of index structure if node is Data Node, the insertion of other types node be reinserted by pressure caused by, level for its reinsert before where the number of plies;3) calculate current_node each child node and be inserted into the axial bounding box circumsphere degree of overlapping of node, select degree of overlapping minimum is used as current_node;4)Repeat step 2)Untill the level layers of index structure.
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method in step 2, by Knots inserting to index structure, makes any two node
Figure 948297DEST_PATH_IMAGE003
Figure 281189DEST_PATH_IMAGE004
The bounding polygon of axial bounding box is respectively
Figure 829982DEST_PATH_IMAGE005
Figure 918024DEST_PATH_IMAGE006
, the distance between axial bounding box center is
Figure 122740DEST_PATH_IMAGE007
, two nodes
Figure 724940DEST_PATH_IMAGE004
The computing formula of the circumsphere degree of overlapping of axial bounding box is
Figure 420101DEST_PATH_IMAGE008
, the similitude weighed with the axial bounding box circumsphere degree of overlapping of node between the similitude size between two nodes, the more big then node of degree of overlapping is bigger, otherwise smaller.
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method, in the step 3 of step 2)In, the step of selection reinserts node is specifically:1)To overflowing node n ode'sIndividual child node, calculates the center of their axial bounding box to the distance at the center of node n ode axial bounding box;2)With step 1)The distance value of middle calculating is keyword, descending sort is carried out to node n ode child node, before selectingIndividual child node.
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method, in the step 3 of step 2)In, it is specifically the step of node split:1)Binary coding is carried out to node n ode child node, 0 represents non-cluster center, and 1 represents cluster centre, and constructs and initialize population P (t), the t=1 of given size, calculates its object function and adaptive value;2)Object function according to individual selects non-dominant disaggregation E (t);3)Population P (t) is selected, intersected, mutation operation, obtain population P (t+1) of future generation, make t=t+1;4)Calculate population P (t) target function value and adaptive value;5)Population P (t) non-dominant disaggregation is calculated, non-dominant disaggregation E (t) is then updated;6)Step 7 is jumped to if the evolutionary generation of cut-off is reached), otherwise jump to step 3);7)Non-dominant disaggregation E (t) is decoded, optimal division scheme of the axial bounding box degree of overlapping division scheme minimum with axial bounding box volume sum as node n ode is then therefrom chosen;8)The optimal division scheme for making node n ode is
Figure 372511DEST_PATH_IMAGE011
, by sub-clusteringIt is sub-clustering set as node n ode child node
Figure 778401DEST_PATH_IMAGE013
The newly-built node of difference, calculate the axial bounding box of new node, and by new node
Figure 900258DEST_PATH_IMAGE014
It is inserted into as the child node of node n ode father node in index structure.
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method, in the step 3 of step 2)Node split during, the number of clusters k of object function including sub-clustering and the class cohesion of sub-clustering
Figure 235425DEST_PATH_IMAGE015
, the number of clusters k of sub-clustering is in individual UVR exposure 1 number, calculates class cohesion
Figure 902029DEST_PATH_IMAGE015
Need first to individual decoding, i.e., cluster centre is extracted from individual UVR exposure, the axial bounding box center of other nodes is calculated to the distance of cluster centre, is grouped into away from the sub-clustering representated by its nearest cluster centre, forms sub-clustering set
Figure 478821DEST_PATH_IMAGE011
The decoding of individual is realized, then the definition of class cohesion is
Figure 984889DEST_PATH_IMAGE017
, wherein k represents number of clusters,
Figure 575008DEST_PATH_IMAGE018
For node,
Figure 457513DEST_PATH_IMAGE019
For sub-clustering
Figure 809997DEST_PATH_IMAGE020
Cluster centre;If in the absence of any
Figure 690229DEST_PATH_IMAGE021
So that formula
Figure 128163DEST_PATH_IMAGE022
Set up, and at least one solution is strict, then it is assumed that solution
Figure 548780DEST_PATH_IMAGE023
For non-domination solution, whereinAll feasible solutions are represented,
Figure 806903DEST_PATH_IMAGE025
Represent each object function, i.e. number of clusters k and class cohesion
Figure 732134DEST_PATH_IMAGE015
, non-dominant disaggregation E (t) is constituted by all non-domination solutions;If two solutions in feasible solution
Figure 392660DEST_PATH_IMAGE026
Meet formula
Figure 719736DEST_PATH_IMAGE027
With
Figure 4087DEST_PATH_IMAGE028
Then thinkDominate;Adaptive value is defined as
Figure 298299DEST_PATH_IMAGE031
, wherein
Figure 753551DEST_PATH_IMAGE032
Represent individual
Figure 591057DEST_PATH_IMAGE029
Ranking value, if individual
Figure 157168DEST_PATH_IMAGE029
For non-domination solution, then its ranking value is 1, and otherwise its ranking value adds one to dominate the individual individual amount.
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method is specifically by Knots inserting to index structure, the step of adjustment node axial bounding box:1)If the father node of the new node being inserted into index structure is src_node;2)Adjust father node src_node axial bounding box, all child nodes for making it include father node src_node just;3)If father node src_node is root index node, program is returned, and is otherwise continued executing with;4)It is step 1 to make father node src_node)Middle father node src_node father node, return to step 2).
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method is optimized, step is specifically in step 3 to index structure:1) index structure is traveled through, the average external volume of the axial bounding box of leaf index node layer is calculated;2) each leaf index node of index structure is traveled through, if the volume of leaf index node axial direction bounding box is more than
Figure 757094DEST_PATH_IMAGE034
,
Figure 144213DEST_PATH_IMAGE035
The threshold value set for user, generally takes 3 ~ 5, then the Data Node included is added in interim sequence L, and the Data Node included is deleted from index structure;3) Data Node in sequence L is reinserted into index structure, realizes the global optimization of index structure.
To realize goal of the invention, described product reverse-engineering data dynamic index multiple target self adaptation construction method, in step 4, the contiguous object for inquiring about any spatial object is comprised the following steps that:1)The axial bounding box circumsphere for making spatial object T is S;2) make
Figure 248435DEST_PATH_IMAGE036
The neighbouring object set of the search space object T in using node N as the index structure of root index node is represented, if node N is Data Node and handed over circumsphere S-phase, its spatial object set included is returned to, if node N is not Data Node,
Figure 840828DEST_PATH_IMAGE037
, wherein
Figure 637883DEST_PATH_IMAGE038
Represent the child node handed in node N with circumsphere S-phase;3) current node N is initialized as to the root index node of index structure, then spatial object T neighbouring object set is
Figure 246718DEST_PATH_IMAGE036
The present invention compared with prior art, with following four feature:
1)The similitude of spatial object is weighed according to the axial bounding box circumsphere degree of overlapping of node, spatial object position had not only been reflected but also had reflected its shared area of space size, and improved polymerism of the node in spatial dimension, sub-clustering result is more reasonable;
2)The optimal solution that node split solves node split is carried out with reference to Multi-objective Optimization by Genetic algorithm, the parameter dependence of node split process is reduced, the space index structure for improving spatial object sets up efficiency;
3)The Data Node that the excessive leaf index node of volume is included is reinserted into index structure, it is to avoid the generation of the unusual node of axial bounding box, improves the quality of index structure;
4)The topological neighbor tri patch of tri patch is fast and accurately obtained using depth-first search traversal method, query context can be effectively reduced, the search efficiency of target tri patch topological neighbor dough sheet is improved.
Brief description of the drawings
Fig. 1 is that product reverse-engineering data dynamic indexing structure of the present invention sets up program and realizes flow chart;
Fig. 2 is polygonal mesh and its axial bounding box schematic diagram;
Fig. 3 is the planar structure schematic diagram of index structure of the present invention;
Fig. 4 is the tree-shaped structural representation of index structure of the present invention;
Fig. 5 is the new Knots inserting flow chart of product reverse-engineering data dynamic indexing structure of the present invention;
Fig. 6 is product reverse-engineering data dynamic indexing structure node split flow chart of the present invention;
Fig. 7 is face polygonal grid model of the present invention;
Fig. 8-Figure 11 is the axial bounding box of each layer node of space index structure that the present invention is set up to face polygonal grid model.
Embodiment
The invention will be further described below in conjunction with the accompanying drawings:
Fig. 1 is that product reverse-engineering data dynamic indexing structure of the present invention sets up program and realizes flow chart, product reverse-engineering data dynamic indexing structure sets up program bag product reverse-engineering data file containing reading program 1, Data Node is inserted into index structure program 2, optimum indexing structure program 3 and object space object neighbour's Object Query program 4, wherein, read product reverse-engineering data file program 1 and read product reverse-engineering data file, set up the axial bounding box of each spatial object, its corresponding Data Node is set up according to the center of axial bounding box and bounding polygon, and it is stored in Data Node sequence;Data Node is inserted into index structure program 2 and reads Data Node sequence, select Knots inserting position, judge whether the child node number of node is more than the maximum child node number of node, if the progress node Overflow handling more than if, if node is non-root node and layer where the node carries out Overflow handling for the first time during a spatial object is inserted, a part of node is then selectively taken out in node, in the layer that they are reinserted to index structure, node split is otherwise carried out;The node changed from axial bounding box, along a branch of index structure, the axial bounding box of bottom-up each node of adjustment;Optimum indexing structure program 3 deletes axial bounding box volume excessively or too long and narrow leaf index node, comprising child node be reinserted into index structure, realize the optimization of index structure;The depth-first traversal product reverse-engineering data directory structure of object space object topological neighbor Object Query program 4, obtain the Data Node intersected with the axial bounding box circumsphere of object space object, the topological neighbor object of selection target spatial object from the spatial object that it is included.
Fig. 2 is some polygonal mesh set.The minimum child node number m=3 of node, maximum child node number M=8 in index of definition structure, Fig. 3 is the axial bounding box schematic diagram of node in the space index structure set up to polygonal mesh shown in Fig. 2, Fig. 4 is the tree-shaped structural representation of index structure, node A is root index node, B, C, D are leaf index node, E, F, G, H, I, J, K, L, M, N, O, P, Q are Data Node, and each Data Node includes a polygonal mesh.
Fig. 5 is shown realizes flow chart by Knots inserting to index structure program 2, calls selection Knots inserting location procedure 1)The position that selection node should be inserted, by Knots inserting to program 1)Obtained position, judge whether the child node number of the father node of node is more than the maximum child node number of node, if the progress node Overflow handling more than if, if node is non-root index node and layer where the node carries out Overflow handling for the first time during a spatial object is inserted, selection is then called to reinsert node code program 2) selectively select several nodes to be reinserted into index structure from node, otherwise call node split program 3)Node split is carried out to the father node of new insertion node, call the axial bounding box program 4 of adjustment node) to inserting node from new, along a branch of index structure, the axial bounding box of each node is adjusted, the axial bounding box of each node is surrounded its child node included just.
By Knots inserting into index structure program 2, selection Knots inserting location procedure 1)The step of be specifically:1)If current index structure is sky, return empty;Otherwise root index nodes of the current node current_node as index structure is set;2)Calculate current node current_node each child node
Figure 420211DEST_PATH_IMAGE003
With being inserted into node
Figure 368575DEST_PATH_IMAGE004
Axial bounding box center between distance be, using formula
Figure 432663DEST_PATH_IMAGE040
The circumsphere degree of overlapping of two nodes is calculated, wherein
Figure 816371DEST_PATH_IMAGE005
Figure 947138DEST_PATH_IMAGE006
Respectively nodeWithAxial bounding box bounding polygon, selection degree of overlapping it is minimum is used as current node current_node;3)Repeat step 2)The layer where current node current_node is leaf index node layer.
By Knots inserting into index structure program 2, selection reinserts node code program 2)The step of be specifically: 1)To overflowing node n ode's
Figure 856822DEST_PATH_IMAGE009
Individual child node, calculates the center of their axial bounding box to the distance at the center of node n ode axial bounding box;2)With step 1)The distance value of middle calculating is keyword, descending sort is carried out to node n ode child node, before selecting
Figure 842096DEST_PATH_IMAGE010
Individual child node.
By Knots inserting into index structure program 2, if node n ode child node numbernMore than maximum child node numberM, then call node split program 3) to node n ode carry out node split, its detailed process as shown in fig. 6, in process (1) individual use length for
Figure 151855DEST_PATH_IMAGE009
Binary string represent that each corresponds to node, 1 represents cluster centre, 0 represents non-cluster center, is randomly 0 or 1 by each position in individual, realizes individual initialization, then distribution and initial population_size individual, complete the initialization of population;In process (2) object function include in number of clusters k and class away from
Figure 658797DEST_PATH_IMAGE015
, 1 number is number of clusters k during statistics is individual, calculates in class away from needing first to decode individual, extract the node corresponding to 1 in individual, i.e. then sub-clustering center calculate other nodes to the distance at each sub-clustering center, be classified as the sub-clustering represented by closest sub-clustering center, forms sub-clustering set, calculate class in away from
Figure 970327DEST_PATH_IMAGE041
, wherein k represents number of clusters,
Figure 262768DEST_PATH_IMAGE018
For node,For sub-clustering
Figure 751835DEST_PATH_IMAGE020
Cluster centre;Process (3) calculates adaptive value individual in population, and computing formula is
Figure 403396DEST_PATH_IMAGE031
, wherein
Figure 448713DEST_PATH_IMAGE032
Represent individual
Figure 843922DEST_PATH_IMAGE029
Ranking value, if individualWith
Figure 418440DEST_PATH_IMAGE030
, meet
Figure 685473DEST_PATH_IMAGE042
With
Figure 55012DEST_PATH_IMAGE043
And it is strict, then individual wherein to have a formula
Figure 723891DEST_PATH_IMAGE029
Dominate individual
Figure 982834DEST_PATH_IMAGE030
If dominating individual without any physical efficiency
Figure 409267DEST_PATH_IMAGE029
, then it is individual
Figure 411858DEST_PATH_IMAGE029
For non-domination solution, to the individual distribution sort value 1 of non-dominant in population, other individual ranking values add one equal to the individual individual amount is dominated;Process (4) construction Pareto solution set
Figure 935244DEST_PATH_IMAGE044
, each individual is traveled through, looks for whether, with the presence of the individual of the individual can be dominated, to gather individual addition if in the absence of if for each individual
Figure 99509DEST_PATH_IMAGE044
;Process (5) is solved to Pareto to be gathered
Figure 278817DEST_PATH_IMAGE044
Individual decoding, therefrom select the axial bounding box degree of overlapping division scheme minimum with axial bounding box volume sum as the optimal division scheme of node
Figure 819520DEST_PATH_IMAGE011
, by sub-clustering
Figure 197412DEST_PATH_IMAGE012
It is sub-clustering set as the child node of nodeThe newly-built node of difference
Figure 199183DEST_PATH_IMAGE014
, calculate the axial bounding box of new node, and by new node
Figure 543576DEST_PATH_IMAGE014
It is inserted into as the child node of the father node of node in index structure.
The step of optimum indexing structure program 3 is specifically:1) index structure is traveled through, the average external volume of the axial bounding box of leaf index node layer is calculated
Figure 477772DEST_PATH_IMAGE033
;2)Each leaf index node of index structure is traveled through, if the volume of leaf index node axial direction bounding box is more than
Figure 983840DEST_PATH_IMAGE034
The threshold value set for user, generally takes 3-5), then it is deleted from index structure, and the spatial object included is added in interim sequence L;3)Spatial object in sequence L is reinserted into index structure, the global optimization of index structure is realized.
Fig. 7 is face polygonal grid model, and the model surface-type feature is complex, is made up of 47765 polygonal mesh, sets up its space index structure using the present invention, the axial bounding box effect of each layer node is as shown in figures s-11.Fig. 8 is the axial bounding box of root index node, Fig. 9 is the axial bounding box of internal index node, Figure 10 is the axial bounding box of leaf index node, Figure 11 is the axial bounding box of Data Node, in addition to root index node, the child node number of each index node is in the range of 8 ~ 20, and each Data Node includes a polygonal mesh.
The step of contiguous object for inquiring about any spatial object T is specifically:1)Solution room object T axial bounding box circumsphere S;2) defined function
Figure 82563DEST_PATH_IMAGE036
The neighbouring object set of the search space object T in using node N as the index structure of root index node is represented, if node N is Data Node and handed over circumsphere S-phase, its spatial object set included is returned to, if node N is internal node,
Figure 372730DEST_PATH_IMAGE037
, whereinRepresent the child node handed in node N with circumsphere S-phase;3) current node N is initialized as to the root index node of index structure, then spatial object T neighbouring object set is
Figure 487634DEST_PATH_IMAGE036
The dynamic indexing structure construction method of other products reverse-engineering data is ibid.

Claims (3)

1. a kind of product reverse-engineering data dynamic index multiple target self adaptation construction method, it is characterised in that comprise the steps of:One, read product reverse-engineering data, set up each dispersion point cloud, the axial bounding box of polygonal mesh and burst continuous curve surface, the corresponding Data Node of each spatial object is set up according to the center of axial bounding box and bounding polygon, and it is stored in Data Node sequence, wherein node includes index node and Data Node, index node includes root index node, internal index node and leaf index node, the superiors' node of index structure is root index node, orlop node is leaf index node, remaining node is internal index node, definition
Figure 978DEST_PATH_IMAGE001
For the maximum child node number of node,For the minimum child node number of node, wherein
Figure 259101DEST_PATH_IMAGE001
For the integer more than 2,For less than or equal to
Figure 346323DEST_PATH_IMAGE001
/ 2 integer, in addition to root index node, the child node number of each index node, which is respectively less than, to be equal to
Figure 109617DEST_PATH_IMAGE001
And be more than or equal to
Figure 393968DEST_PATH_IMAGE002
;The axial bounding box of each node surrounds all child nodes of the node just in index structure;2nd, Data Node is inserted into index structure, step is specifically:1)For node selection insertion position, concretely comprise the following steps:(1) it is current_node to make current node, and sky is returned if index structure is sky, and it is index structure root index node otherwise to make current_node;(2) it is level to make the number of plies that node will insert, and level is the leaf layer of index structure if node is Data Node, the insertion of other types node be reinserted by pressure caused by, level for its reinsert before where the number of plies;(3) calculate current_node each child node and be inserted into the axial bounding box circumsphere degree of overlapping of node, selection degree of overlapping it is minimum as current_node, wherein the method for calculating the axial bounding box circumsphere degree of overlapping of two nodes is:Make any two node
Figure 540916DEST_PATH_IMAGE003
Figure 506598DEST_PATH_IMAGE004
Axial bounding box bounding polygon be respectively
Figure 688180DEST_PATH_IMAGE005
Figure 143432DEST_PATH_IMAGE006
, the distance between axial bounding box center is, using formula
Figure 547049DEST_PATH_IMAGE008
Calculate the circumsphere degree of overlapping of the axial bounding box of two nodes;(4) repeat step (2) is untill the level layers of index structure;2)By Knots inserting to step 1)In obtained insertion position;3)Make whether Knots inserting is more than the maximum child node number of node to the child node number under node n ode, judging node n ode, if greater than then Overflow handling is carried out to node n ode, if node n ode is non-root index node and layer where the node carries out Overflow handling for the first time during a spatial object is inserted, calculation overflow node n ode's
Figure 146975DEST_PATH_IMAGE009
The center of the axial bounding box of individual child node, using distance value as keyword, carries out descending sort, before selecting to the distance at the center of node n ode axial bounding box to node n ode child node
Figure 268514DEST_PATH_IMAGE010
Individual child node reinserts them in the layer of index structure, and node n ode child node otherwise is divided into k clusters
Figure 74534DEST_PATH_IMAGE011
, by sub-clustering
Figure 230709DEST_PATH_IMAGE012
It is sub-clustering set as node n ode child node
Figure 27764DEST_PATH_IMAGE013
The newly-built node of difference
Figure 574283DEST_PATH_IMAGE014
, calculate the axial bounding box of new node, and by new node
Figure 482196DEST_PATH_IMAGE014
It is inserted into as the child node of node n ode father node in index structure, realizes the division of node;4)The axial bounding box of each node is adjusted, detailed process is:(1) father node of the new node being inserted into index structure is set as src_node;(2) adjustment father node src_node axial bounding box, all child nodes for making it include father node src_node just;(3) if father node src_node is root index node, program is returned, and is otherwise continued executing with;(4) father node that father node src_node is father node src_node in step (1), return to step (2) are made;3rd, the excessive axial bounding box of volume is reinserted into index structure, realizes the optimization of index structure;4th, based on product reverse-engineering data dynamic indexing structure, the topological neighbor inquiry of dispersion point cloud, polygonal mesh and burst continuous curve surface is realized, wherein inquiring about comprising the following steps that for any spatial object T contiguous object:1)The axial bounding box circumsphere for making spatial object T is S;2) make
Figure 492877DEST_PATH_IMAGE015
The neighbouring object set of the search space object T in using node n as the index structure of root index node is represented, if node n is Data Node and handed over circumsphere S-phase, its spatial object set included is returned to, if node n is internal node,
Figure 398516DEST_PATH_IMAGE016
, wherein
Figure 494648DEST_PATH_IMAGE017
Represent the child node handed in node N with circumsphere S-phase;3) current node N is initialized as to the root index node of index structure, then spatial object T neighbouring object set is
Figure 206252DEST_PATH_IMAGE018
2. product reverse-engineering data dynamic index multiple target self adaptation method for building up as claimed in claim 1, it is characterised in that:In the step 3 of step 2)In, it is specifically the step of node split:1)Binary coding is carried out to node n ode child node, 0 represents non-cluster center, and 1 represents cluster centre, and constructs and initialize population P (t), the t=1 of given size, calculates its object function and adaptive value;2)Object function according to individual selects non-dominant disaggregation E (t);3)Population P (t) is selected, intersected, mutation operation, obtain population P (t+1) of future generation, make t=t+1;4)Calculate population P (t) target function value and adaptive value;5)Population P (t) non-dominant disaggregation is calculated, non-dominant disaggregation E (t) is then updated;6)Step 7 is jumped to if the evolutionary generation of cut-off is reached), otherwise jump to step 3);7)Non-dominant disaggregation E (t) is decoded, optimal division scheme of the axial bounding box degree of overlapping division scheme minimum with axial bounding box volume sum as node n ode is then therefrom chosen;8)The optimal division scheme for making node n ode is
Figure 274702DEST_PATH_IMAGE011
, by sub-clustering
Figure 413560DEST_PATH_IMAGE012
It is sub-clustering set as node n ode child node
Figure 731409DEST_PATH_IMAGE013
The newly-built node of difference
Figure 682922DEST_PATH_IMAGE014
, calculate the axial bounding box of new node, and by new node
Figure 668195DEST_PATH_IMAGE014
It is inserted into as the child node of node n ode father node in index structure.
3. product reverse-engineering data dynamic index multiple target self adaptation method for building up as claimed in claim 1, it is characterised in that:Index structure is optimized in step 3, step is specifically:1) index structure is traveled through, the average external volume of the axial bounding box of leaf index node layer is calculated;2) each leaf index node of index structure is traveled through, if the volume of leaf index node axial direction bounding box is more than
Figure 720782DEST_PATH_IMAGE020
(
Figure 774189DEST_PATH_IMAGE021
The threshold value set for user, generally take 3 ~ 5), then the Data Node included is added in interim sequence L, and the Data Node included is deleted from index structure;3) Data Node in sequence L is reinserted into index structure, realizes the global optimization of index structure.
CN2012103325379A 2012-09-11 2012-09-11 Dynamic index multi-target self-adaptive construction method for product reverse engineering data Pending CN102831241A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2012103325379A CN102831241A (en) 2012-09-11 2012-09-11 Dynamic index multi-target self-adaptive construction method for product reverse engineering data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2012103325379A CN102831241A (en) 2012-09-11 2012-09-11 Dynamic index multi-target self-adaptive construction method for product reverse engineering data

Publications (1)

Publication Number Publication Date
CN102831241A true CN102831241A (en) 2012-12-19

Family

ID=47334376

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2012103325379A Pending CN102831241A (en) 2012-09-11 2012-09-11 Dynamic index multi-target self-adaptive construction method for product reverse engineering data

Country Status (1)

Country Link
CN (1) CN102831241A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105868355A (en) * 2016-03-29 2016-08-17 贵州大学 Large-scale multimedia data spatial index method
CN107294656A (en) * 2017-06-06 2017-10-24 长安大学 A kind of distributed arithmetic code coding/decoding method based on depth-first
CN107391601A (en) * 2017-06-30 2017-11-24 安徽四创电子股份有限公司 A kind of construction method of the high dimensional indexing of face feature vector
CN110048945A (en) * 2019-04-24 2019-07-23 湖南城市学院 A kind of node mobility cluster-dividing method and system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5781906A (en) * 1996-06-06 1998-07-14 International Business Machines Corporation System and method for construction of a data structure for indexing multidimensional objects
US6389424B1 (en) * 1998-10-28 2002-05-14 Electronics And Telecommunications Research Institute Insertion method in a high-dimensional index structure for content-based image retrieval
CN101430693B (en) * 2008-11-12 2010-08-11 山东理工大学 Spacing query method for triangular gridding curve model
CN101510315B (en) * 2009-03-26 2012-07-18 山东理工大学 Method for establishing space index structure of product STL model

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5781906A (en) * 1996-06-06 1998-07-14 International Business Machines Corporation System and method for construction of a data structure for indexing multidimensional objects
US6389424B1 (en) * 1998-10-28 2002-05-14 Electronics And Telecommunications Research Institute Insertion method in a high-dimensional index structure for content-based image retrieval
CN101430693B (en) * 2008-11-12 2010-08-11 山东理工大学 Spacing query method for triangular gridding curve model
CN101510315B (en) * 2009-03-26 2012-07-18 山东理工大学 Method for establishing space index structure of product STL model

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
孙永伟: "基于最小包围盒及自适应聚类的三维R*-树索引结构", 《山东理工大学硕士学位论文》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105868355A (en) * 2016-03-29 2016-08-17 贵州大学 Large-scale multimedia data spatial index method
CN107294656A (en) * 2017-06-06 2017-10-24 长安大学 A kind of distributed arithmetic code coding/decoding method based on depth-first
CN107294656B (en) * 2017-06-06 2021-02-02 长安大学 Depth-first-based distributed arithmetic code decoding method
CN107391601A (en) * 2017-06-30 2017-11-24 安徽四创电子股份有限公司 A kind of construction method of the high dimensional indexing of face feature vector
CN110048945A (en) * 2019-04-24 2019-07-23 湖南城市学院 A kind of node mobility cluster-dividing method and system

Similar Documents

Publication Publication Date Title
CN105760503B (en) A kind of method of quick calculating node of graph similarity
KR20120095869A (en) Method and apparatus for encoding a mesh model, encoded mesh model, and method and apparatus for decoding a mesh model
CN102073700A (en) Discovery method of complex network community
CN101430693B (en) Spacing query method for triangular gridding curve model
CN102831241A (en) Dynamic index multi-target self-adaptive construction method for product reverse engineering data
Liu et al. Configuration space evolutionary algorithm for multi-objective unequal-area facility layout problems with flexible bays
CN101751694A (en) Method for rapidly simplifying and drawing complex leaf
CN104731984B (en) Automobile wheel hub surface sampling point R tree overflow node incremental clustering optimization method
CN104778281A (en) Image index parallel construction method based on community analysis
CN101692230A (en) Three-dimensional R tree spacial index method considering levels of detail
CN106503092A (en) A kind of method using multidimensional technique construction Spatial Multi-Dimensional degree search tree
CN110147377A (en) General polling algorithm based on secondary index under extensive spatial data environment
CN101510315B (en) Method for establishing space index structure of product STL model
CN103092992A (en) Vector data preorder quadtree coding and indexing method based on Key / Value type NoSQL (Not only SQL)
CN113094463A (en) Unstructured point cloud storage method, device, equipment and medium
CN104933143A (en) Method and device for acquiring recommended object
CN103198103A (en) Microblog pushing method and device based on dense word clustering
CN103500165B (en) A kind of combination cluster and the high-dimensional vector quantity search method of double key value
Xiao A clustering algorithm based on artificial fish school
CN105138607B (en) A kind of KNN querying methods based on combination grain distributed memory grid index
CN112214488A (en) European style spatial data index tree and construction and retrieval method
Tian et al. Research on group animation design technology based on artificial fish swarm algorithm
CN103425787B (en) The triangle gridding of a kind of asymptotic optimization repeats the quick minimizing technology in summit
Wang et al. A hybrid spatial indexing structure of massive point cloud based on octree and 3D R*-tree
Liu et al. Research on hybrid index based on 3D multi-level adaptive grid and R+ Tree

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20121219