CN112861341A - Multi-scale surface element matching method and system combining shape and environmental characteristics - Google Patents

Abstract

The invention relates to a multi-scale surface element matching method and a system combining shape and environmental characteristics, which comprises the following steps: s100, acquiring a candidate element set of each element, and sequencing the elements in the candidate elements according to the maximum and minimum coordinates of MBR; s200, constructing a solution space tree, and searching potential corresponding MBRs by adopting an MBR combined optimization algorithm; s300, aligning potential corresponding MBRs, and acquiring potential matching pair results by using a bidirectional area overlapping method; s400, establishing a new potential matching correspondence table; s500, establishing an adjacency relation between elements in the matched data set; s600, finding the final result of the surface element matching in the potential matching pair by using a probability relaxation marking method. The method combines an MBR combined algorithm and a probability relaxation iterative algorithm, effectively aggregates potential matching pairs by utilizing an MBR combined optimization algorithm, finds correct matching pairs by utilizing a probability relaxation iterative method, has good identification capability on complex matching, and avoids the defect of manually setting a matching threshold.

Description

Multi-scale surface element matching method and system combining shape and environmental characteristics

Technical Field

The application relates to the field of geographic information science, in particular to a multi-scale surface element matching method and system combining shape and environmental characteristics.

Background

Element matching is a basic technology in the field of spatial data processing and application, and is widely applied to updating, maintaining and fusing spatial data. With the rise of 'spontaneous Geographic Information' (VGI), element matching becomes a key technology for evaluating and improving the data quality of the VGI. However, geospatial data from different sources often are inconsistent in terms of situational, expression scale, geometry, and semantics, making it a challenge to obtain accurate matching results. Especially the matching of multi-scale vector data, because complex matching (one-to-many and many-to-many matching) exists widely in multi-scale matching, and features between matched data are more fuzzy.

To date, there has been a great deal of research in the field of element matching. If the candidate matching is obtained by a buffer area growing method and an area overlapping method, then, the correct matching is found from the candidate matching by utilizing geometric information, attribute information or topological information; and introducing an optimization model, a logistic regression model, a BP neural network and an SVM to improve the matching accuracy and the like. However, the buffer growing method can only obtain 1:1 matching candidates, which cannot determine which elements in the buffer need to be aggregated for matching; the area overlapping method cannot be used when the spatial data has large displacement deviation, even though the matching elements are only roughly aligned through preprocessing such as projection conversion, format conversion, map correction and the like, coordinate deviation still exists, and as a result, a large number of correct matches are missed, and elements with small areas are easily ignored by errors. Therefore, the problem of processing complex matching in multi-scale matching is difficult to effectively solve in the existing element matching result, and the applicability of the matching technology is influenced.

Disclosure of Invention

The invention aims to provide a multi-scale surface element matching method and system combining shape and environmental characteristics, which are used for establishing a corresponding relation between public geographic information and reference geographic data, judging the position precision, the shape precision and the integrity of the public geographic information, realizing map updating and map fusion of surveying and mapping work, establishing a multi-expression spatial database, overcoming the influence of position deviation on surface element matching and improving matching adaptability and accuracy.

The technical scheme adopted by the invention is as follows: a multi-scale surface element matching method combining shape and environmental characteristics comprises the following steps:

s000: acquiring a multi-scale surface element data packet of geographic information, taking vector space data of the multi-scale surface element data packet as a matching data source, and preprocessing the matching data source to obtain a data set to be matched;

s100: two data sets to be matched with different scales are respectively marked as A ═ a₁,a₂,..,a_IB ═ B₁,b₂,..,b_MAny one element in A is marked as a_iEstablishing a_iThe buffer area of (2) records the buffer distance as d_τ Calculating the data set B falling into a by a superposition analysis method_iIs marked as a_iIs marked as Bⁱ＝{b₁,b₂,..,b_NWherein N is BⁱThe number of the middle elements; obtaining BⁱMBR of each surface element, BⁱEach element in (1) is respectively according to x of minimum x coordinate in MBR_minMaximum x coordinate x in MBR_maxMinimum y-coordinate y in MBR_minAnd the maximum y-coordinate y in MBR_maxSorting is carried out, and sorting results are respectively expressed as O_xmin,O_xmax,O_yminAnd O_ymax；

S200: constructing an N-fork complete solution space tree with the depth of 5, and searching a by adopting a backtracking algorithm-based MBR (Membrane biological reactor) combined optimization algorithm according to a depth-first principle_iOf MBR, wherein a_iMBR of (a) is denoted as MBR_i)；

S300: each MBR (a)_i) Of the geometric center point of (a) with respect to the MBR (lambda)Aligning geometric central points, and acquiring potential matching pair result b by using a bidirectional area overlapping method_j1,b_j2,..,b_jm；

S400: a new potential matching correspondence table T1 is established, and the result of the potential matching pair obtained in S300 is represented as (a)_i：b_i1)、(a_i：b_i1,b_i2)，..，(a_i：b_i1,b_i2,..,b_im) Wherein (m is less than or equal to N), respectively calculating b_i1，{b_i1,b_i2}，..，{b_i1,b_i2,..,b_imObtaining new face element c by convex hull₁，c₂，..，c_jThus, a new potential matching pair correspondence (a) is obtained_i：c₁)，(a_i：c₂)，..，(a_i：c_j) And a is_iIs Cⁱ＝{c₁,c₂,..,c_jRecording the corresponding relation in T1; repeating the step S400 to obtain potential matching pairs of all elements in the data set a, calculating their convex hulls to obtain a new data set C ═ C₁,c₂,..,c_jFourthly, a new potential matching corresponding table T1 is finally established;

s500: establishing adjacency relations between elements in the data sets A and C respectively;

s600: and acquiring the final result of the surface element matching in the T1 table by using a probability relaxation marking method.

Further, the specific method of step S000 includes format conversion, topology inspection, and geometric coordinate conversion.

Further, the specific method of step S200 is as follows:

s210: establishing a complete N-way solution space tree with the depth of 5, wherein the root node of the N-way solution space tree is o₀,o₀Deriving N sub-nodes, wherein each sub-node derives N sub-nodes respectively; the N-way spatial tree is a full N-way tree, and comprises (N)⁴+1) nodes, each node being BⁱThe element (1) of (1); wherein the second level sub-nodes are according to O_xminArranged, third layer of sub-nodes pressedAccording to the formula O_xmaxArranged, fourth level of sub-nodes according to O_yminArranged according to the fifth level of sub-nodes O_ymaxArranging;

s220: defining a solution vector lambda, searching the whole solution space tree from a root node according to a depth priority principle, and traversing to o₀When so, then λ ═ null, null, null }; if the second level node is searched, λ ═ o₁Null, null, null }; if the node of the third layer is searched, λ ═ o₁,o₂Null, null }; if the node of the fourth layer is searched, λ ═ o₁,o₂,o₃Null }; when the fifth level node is searched, λ ═ o₁,o₂,o₃,o₄}；o₁,o₂,o₃And o₄Are respectively BⁱA certain element of (1);

s230: defining a constraint function, carrying out condition limitation on the solution space tree when the solution space tree is searched and nodes of each level are traversed, continuing to move the search tree downwards after the condition is met, and returning to the previous level if the condition is not met;

s240: solving lambda, and traversing the search tree in a depth-first order from the root node; and finding all solutions lambda meeting the constraint condition.

Further, the specific expression of the constraint function in step S230 is:

wherein w and h are respectively MBR (a)_i) Is the solution vector of the backtracking algorithm, width (lambda) and height (lambda) are the width and height of MBR (lambda), respectively, epsilon is the search threshold, epsilon belongs to [0.2, 0.5 ]]；

Where o is the junction of the fourth or fifth layer, x_min(lambda) and x_max(λ) is the minimum and maximum abscissa of MBR (λ), respectively, λ ═ o₁，o₂，null，null}；

y_min(o)≤y_min(λ) (3)

Where o is the node of the fourth layer, y_min(o) and y_min(λ) is the minimum ordinate of MBR (o) and MBR (λ), respectively, λ ═ o₁，o₂，null，null}；

Where o is the junction of the fifth layer and y_max(o) and y_max(λ) is the maximum ordinate of MBR (o) and MBR (λ), respectively, λ ═ o₁，o₂，o₃，null}；

When traversing to the third level node, checking whether the node meets the formula (1); when traversing to the node of the fourth layer, checking whether the node meets the formulas (2) and (3); when traversing to the fifth layer node, checking whether the node satisfies the formulas (2) and (4).

Further, the specific method of step S300 is: MBR (a)_i) Is aligned with the center point of MBR (λ), and then using a stack analysis method, a after alignment shift is judged_iAnd b_j1,b_j2,..,b_jmWhether or not the overlapping elements satisfy the following condition:

wherein, M (a)_i) Representing a after alignment translation_i，Area(M(a_i)∩b_jm) Is represented by M (a)_i) And b_jmOverlap Area of (a), min (Area (a)_i),Area(b_jm) Is represented by a)_iAnd b_jmThe smaller value of the areas of the two surface entities, wherein gamma is an area overlapping threshold value; such as b_jmIf the condition is satisfied, then b_jmReserving; otherwise b_jmAre screened out.

Further, the specific method of step S500 is as follows:

s510: establishing an adjacency relation of the data set A, and acquiring geometric center points of all elements in the data set A and recording the geometric center points as V_AAccording to V_AConstruction of a Delaunay triangulation network, denoted G_DT(A)＝(V_A,E_A) In which E_AIs a set of triangulation edges; for any surface element a in A_iAdjacent to the object a_hIs at G_DT(A) And a_iThe elements with connected edges, denoted by N (a)_i)＝{a_h|(a_i,a_h)∈E_AObtaining adjacent objects of all the elements in the A;

s520: establishing a contiguous object of data set C, looking up a according to T1 table_iIs marked as (a)_i：c_j)，a_i∈A，c_jE is C; a is calculated in step S510_iAdjacent object a of_hObtaining a_hAll potential matching pairs of (a)_h：c_k) (ii) a Then, c is established_jBuffer area of (2), recording the buffer distance as d_τ(ii) a By means of a superposition analysis method, c_kWhether or not at a_iIn a buffer zone, e.g. c_kAt a_iIn the buffer area of (1), then c_kIs c_jThe adjacent object of (1); if not, c_kIs other than c_jIs marked as N (c)_j)＝{c_k|(a_h:c_k)&(a_i:c_j)&((a_i,a_h)∈E_A) }; repeating the step S520 to obtain the adjacent objects of all the elements in the C.

Further, the specific method of step S600 is as follows:

s610: construction of an initial matching probability matrix P⁽⁰⁾For data set a ═ a₁,a₂,..,a_iC ═ C₁,c₂,..,c_jI x j matching pairs can be formed; for any matching pair (a) in the candidate matching table T1_i：c_j) Calculating (a)_i：c_j) S (a) of_i,c_j) If c is obtained by looking up from the T1 table_jIs a_iCandidate matching object of (2), i.e. c_i∈CⁱThen a is further obtained by a geometric similarity calculation formula_iAnd c_jGeometric similarity s (a) of_i,c_j) (ii) a If it is

Let s (a)_i,c_j) 0; finally using the formula

Computing initial match probabilities in a probabilistic relaxation iterative framework

S620: calculating any two pairs of candidate matching pairs (a)_i：c_j) And (a)_h：c_k) Coefficient r of compatibility between_ij(h,k)；

S630: for any matching pair (a)_i：c_j) Calculating the support coefficient thereof

Wherein t is the number of iterations;

s640: starting iteration, updating the matching probability matrix P^(t)(ii) a For any matching pair (a)_i：c_j) The iterative calculation formula of the matching probability is as follows:

s650: judging whether the iteration can be ended; computing after each iteration

If min (delta P) is less than or equal to sigma, finishing iteration to obtain a final matching probability, and outputting a matching probability result matrix P based on convergence; for any element in P, if P_ijNot less than 0.5, then (a)_i：c_j) Are matched pairs; due to c_jIs formed by b_j1,b_j2,..,b_jmThe final matching result (a) is obtained_i：b_j1,b_j2,..,b_jm) (ii) a Otherwise a_iAnd b_j1,b_j2,..,b_jmMismatch is not achieved; if min (Δ p)>σ, return to step S620 to continue the iteration.

Further, the specific expression of the geometric similarity calculation formula in step S610 is shown in formula (6):

wherein s is_dis(a_i,c_j)、s_size(a_i,c_j)、s_ori(a_i,c_j) And s_shape(a_i,c_j) Are respectively a_iAnd c_jThe distance similarity, the size similarity, the direction similarity and the shape similarity are calculated according to the following specific calculation formula:

wherein the content of the first and second substances,

and

are respectively a_iAnd c_jThe geometric center point of (a);

wherein, Area (a)_i) Denotes a_iArea of entity, Area (c)_j) Denotes c_jArea of (a), max (Area (a)_i),Area(c_j) Is represented by a)_iAnd c_jThe greater of the two face entity areas;

wherein, theta (a)_i) And theta (c)_j) Are respectively a face element a_iAnd c_jDirection of (a), theta_τIs at [0, π]A direction threshold within a range;

wherein the content of the first and second substances,

and

is the cumulative angle between the tangent to the polygon in the counterclockwise direction and the x-axis, x representing the side length of the polygon,

to represent

And

the greater of the two values; the intersection point of the polygonal outline and the inertia axis is used as the starting point described by the formula (10);

let is described in formula (9)

θ(a_i) And theta (c)_j) By using the direction of the polygonal inertia axis, the formula can be obtained:

wherein the content of the first and second substances,

for the geometrical moments of the polygon, the calculation formula is as follows:

wherein (x)₀,y₀) Is the coordinate of the geometric center point of the polygon, and according to the formula (11), theta can be used to calculate the difference between the two points

And (3) taking a solution corresponding to the long inertia axis.

Further, r in the step 620_ijThe formula for calculating (h, k) is shown in formula (12):

r_ij(h,k)＝rel_dis(ih；jk)*rel_size(ih；jk)*rel_ori(ih；jk)*rel_shape(ih；jk) (12)

wherein rel_dis(ih；jk)、rel_size(ih；jk)、rel_ori(ih; jk) and rel_shape(ih; jk) are each a_iAnd c_jThe relative distance, the relative size, the relative direction and the relative shape of the steel plate are specifically calculated according to the following formula:

the invention adopts another technical scheme that: a multi-scale surface element matching system combining shape and environmental characteristics comprises an MBR combined optimization algorithm identification homonymous MBR module, a bidirectional area overlapping method screening module, a potential matching pair adjacency establishing module, a geometric matching obtaining initial matching probability module, a probability relaxation matrix iteration module and a judgment module;

the MBR combined optimization algorithm identifies the MBR modules with the same name, and identifies the corresponding MBRs of the elements with the same name in the data sets A and B to be matched based on the MBR combined optimization algorithm of the steps 100-240;

the bidirectional area overlap screening module maps MBR (a) based on the method of step 300_i) Is aligned with the center point of MBR (λ), and then using a stack analysis method, a after alignment shift is judged_iAnd b_j1,b_j2,..,b_jmWhether the overlapping elements of (1) satisfy the following condition; such as b_jmIf the condition is satisfied, then b_jmReserving; otherwise b_jmIs screened out;

the module for establishing the adjacency relation of the potential matching pairs respectively establishes the adjacency relation between the elements in the data sets A and C based on the methods of the steps 400 to 520;

the geometric matching initial matching probability module builds an initial matching probability matrix P based on the method of step 610⁽⁰⁾For data set a ═ a₁,a₂,..,a_IC ═ C₁,c₂,..,c_JI multiplied by J matched pairs can be formed; for any matching pair (a) in the candidate matching table T1_i：c_j) Calculating their geometric similarity s (a)_i,c_j) And then further calculate their initial probabilities

The probability relaxation matrix iteration module is based on the methods in the steps 620-640, and iterates to obtain the final probability value of the surface element matching by using a probability relaxation marking method;

the determination module is based on the method of step 650 for (a)_i：c_j) If the final matching probability is more than 0.5, judging that the final matching probability is a matching pair; otherwise, judging the result is not matched.

The invention has the beneficial effects that: the influence of position deviation on the matching of the face elements is overcome, the adaptability is good, and the matching accuracy is high; by setting a buffer query distance, an MBRCO search threshold and a probability relaxation method iteration threshold, most of face element data matching scenes can be adapted, and the matching accuracy can reach more than 93%; the advantages of the MBR combined algorithm and the probability relaxation iterative algorithm are effectively combined, the potential matching pairs are effectively aggregated by the MBR combined optimization algorithm, and the correct matching pairs are found by the probability relaxation iterative method, so that the method not only has good identification capability on complex matching, but also avoids the defect of manually setting a matching threshold.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 is a detailed flow chart of an embodiment of the present invention.

Fig. 2 is a schematic diagram of an example of an MBR combinatorial optimization algorithm according to an embodiment of the present invention.

Fig. 3 is a solution space tree constructed by the example shown in fig. 2.

FIG. 4 is a diagram illustrating neighborhood construction according to an embodiment of the present invention.

Detailed Description

In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those specifically described herein, and thus the present invention is not limited to the specific embodiments disclosed below.

Example 1:

as shown in fig. 1, a multi-scale surface element matching method combining shape and environmental features includes the following steps:

s000: acquiring a multi-scale surface element data packet of geographic information, taking vector space data of the multi-scale surface element data packet as a matching data source, preprocessing the matching data source, eliminating systematic errors of the matching data source, and acquiring a data set to be matched; the multi-scale surface element data packet can be obtained through a surveying and mapping means, crowdsourcing mode sharing, professional geographic information website downloading and the like; the preprocessing comprises format conversion, topology inspection and geometric coordinate conversion; the format conversion, topology inspection and geometry coordinate conversion are all conventional preprocessing means and will not be described in detail here.

S100: two data sets to be matched with different scales are respectively marked as A ═ a₁,a₂,..,a_IB ═ B₁,b₂,..,b_MAny one element in A is marked as a_iEstablishing a_iThe buffer area of (2) records the buffer distance as d_τCalculating the data set B falling into a by a superposition analysis method_iIs marked as a_iIs marked as Bⁱ＝{b₁,b₂,..,b_NWherein N is BⁱThe number of the middle elements; obtaining BⁱMBR of each surface element, BⁱEach element in (1) is respectively according to x of minimum x coordinate in MBR_minMaximum x coordinate x in MBR_maxMinimum y-coordinate y in MBR_minAnd the maximum y-coordinate y in MBR_maxSorting is carried out, and sorting results are respectively expressed as O_xmin,O_xmax,O_yminAnd O_ymax。

S200: constructing an N-fork complete solution space tree with the depth of 5, and searching a by adopting a backtracking algorithm-based MBR (Membrane biological reactor) combined optimization algorithm according to a depth-first principle_iThe MBR of (1); wherein a is_iMBR of (a) is denoted as MBR_i) The specific method comprises the following steps:

s210: as shown in FIG. 3, a complete N-way solution space tree with a depth of 5 is established, and the root node of the N-way solution space tree is o₀,o₀Deriving N sub-nodes, wherein each sub-node derives N sub-nodesPoint; the N-way spatial tree is a full N-way tree, and comprises (N)⁴+1) nodes, each node being BⁱThe element (1) of (1); wherein the second level sub-nodes are according to O_xminArranged, third level sub-nodes according to O_xmaxArranged, fourth level of sub-nodes according to O_yminArranged according to the fifth level of sub-nodes O_ymaxArranging;

s220: defining a solution vector lambda, searching the whole solution space tree from a root node according to a depth priority principle, and traversing to o₀When so, then λ ═ null, null, null }; if the second level node is searched, λ ═ o₁Null, null, null }; if the node of the third layer is searched, λ ═ o₁,o₂Null, null }; if the node of the fourth layer is searched, λ ═ o₁,o₂,o₃Null }; when the fifth level node is searched, λ ═ o₁,o₂,o₃,o₄}；o₁,o₂,o₃And o₄Are respectively BⁱA certain element of (1); if o₁＝o₂＝o₃＝o₄Then a is_iOnly one element of the potential matching pair.

S230: defining a constraint function, as shown in fig. 2, performing conditional restriction on the solution space tree when the solution space tree is searched and nodes of each level are traversed, wherein the search tree continues to move downwards after the condition is met, and returns to the previous level if the condition is not met; the specific expression of the constraint function is as follows:

wherein w and h are respectively MBR (a)_i) Is the solution vector of the backtracking algorithm, width (lambda) and height (lambda) are the width and height of MBR (lambda), respectively, epsilon is the search threshold, epsilon belongs to [0.2, 0.5 ]](ii) a The larger epsilon, the lower the recall rate of the search algorithm, but the more time-consuming;

where o is the junction of the fourth or fifth layer, x_min(lambda) and x_max(λ) is the minimum and maximum abscissa of MBR (λ), respectively, λ ═ o₁，o₂，null，null}；

y_min(o)≤y_min(λ) (3)

Where o is the node of the fourth layer, y_min(o) and y_min(λ) is the minimum ordinate of MBR (o) and MBR (λ), respectively, λ ═ o₁，o₂，null，null}；

Where o is the junction of the fifth layer and y_max(o) and y_max(λ) is the maximum ordinate of MBR (o) and MBR (λ), respectively, λ ═ o₁，o₂，o₃，null}；

When traversing to the node of the third level, checking whether the node meets the formula (1), if λ meets the constraint condition, obtaining a solution vector λ ═ o₁,o₂Null, null, otherwise, λ ═ null, null, null, left out.

When traversing to the node of the fourth layer, checking whether the node satisfies the formulas (2) and (3), wherein λ ═ o₁,o₂Null, null }, and detecting λ ═ o₁,o₂Null, null } whether to satisfy constraint formula (2) to achieve width constraint; detecting λ ═ o₁,o₂Null, null } satisfies formula (4) to achieve height constraint; when λ satisfies the constraint condition, the solution vector λ ═ o can be obtained₁,o₂,o₃Null, null, otherwise, get λ ═ null, null, null }, null, and discard.

When traversing to the fifth node, check whether the node satisfies the formulas (2) and (4), where λ ═ o₁,o₂，o₃Null, there are 3 sub-elements o₁，o₂，o₃Detecting a solution vector lambda ═ { o ] composed of the first two subelements₁,o₂Null, null } whether or not it satisfies a constraintEquation (2), achieving width constraint; and detecting the integral lambda of the solution vector as { o ═ o₁,o₂，o₃Null } if it satisfies formula (4), achieving a height constraint; when λ satisfies the constraint condition, the solution vector λ ═ o can be obtained₁,o₂,o₃,o₄Else, get λ ═ { null, null, null, null }, null, and discard.

S240: solving lambda, and traversing the search tree in a depth-first order from the root node; finding all solutions lambda meeting constraint conditions;

s300: each MBR (a)_i) Is aligned with the geometric center point of MBR (lambda), and a bidirectional area overlapping method is used for obtaining a result b of potential matching pairs_j1,b_j2,..,b_jm(ii) a The specific method is to mix MBR (a)_i) Is aligned with the center point of MBR (λ), and then using a stack analysis method, a after alignment shift is judged_iAnd b_j1,b_j2,..,b_jmWhether or not the overlapping elements satisfy the following condition:

wherein, M (a)_i) Representing a after alignment translation_i，Area(M(a_i)∩b_jm) Is represented by M (a)_i) And b_jmOverlap Area of (a), min (Area (a)_i),Area(b_jm) Is represented by a)_iAnd b_jmThe smaller of the two surface solid areas, γ is the area overlap threshold, and γ is 0.3 in example 1. Such as b_jmIf the condition is satisfied, then b_jmReserving; otherwise b_jmAre screened out.

S400: a new potential matching correspondence table T1 is established, and the result of the potential matching pair obtained in S300 is represented as (a)_i：b_i1)、(a_i：b_i1,b_i2)，..，(a_i：b_i1,b_i2,..,b_im) Wherein (m.ltoreq.N), (a)_i：b_i1) For one-to-one matching correspondence results, (a)_i：b_i1,b_i2)，..，(a_i：b_i1,b_i2,..,b_im) The corresponding result is a one-to-many match. Respectively calculate b_i1，{b_i1,b_i2}，..，{b_i1,b_i2,..,b_imObtaining new face element c by convex hull₁，c₂，..，c_jThus, a new potential matching pair correspondence (a) is obtained_i：c₁)，(a_i：c₂)，..，(a_i：c_j) And a is_iIs Cⁱ＝{c₁,c₂,..,c_jRecording the corresponding relation in T1; repeating the step S400 to obtain potential matching pairs of all elements in the data set a, calculating their convex hulls to obtain a new data set C ═ C₁,c₂,..,c_jAnd finally, establishing a new potential matching correspondence table T1.

S500: as shown in fig. 4, the adjacency relations between the elements in the data sets a and C are respectively established, and the specific steps are as follows:

s510: establishing an adjacency relation of the data set A, and acquiring geometric center points of all elements in the data set A and recording the geometric center points as V_AAccording to V_AConstruction of a Delaunay triangulation network, denoted G_DT(A)＝(V_A,E_A) In which E_AIs a set of triangulation edges; for any surface element a in A_iAdjacent to the object a_hIs at G_DT(A) And a_iThe elements with connected edges, denoted by N (a)_i)＝{a_h|(a_i,a_h)∈E_AAnd obtaining adjacent objects of all the elements in the A.

S520: establishing a contiguous object of data set C, looking up a according to T1 table_iIs marked as (a)_i：c_j)，a_i∈A，c_jE is C; a is calculated in step S510_iAdjacent object a of_hObtaining a_hAll potential matching pairs of (a)_h：c_k) (ii) a Then, c is established_jBuffer area of (2), recording the buffer distance as d_τ(ii) a By means of a superposition analysis method, c_kWhether or not at a_iIn a buffer zone, e.g. c_kAt a_iIn the buffer area of (1), then c_kIs c_jThe adjacent object of (1); if not, c_kIs other than c_jIs marked as N (c)_j)＝{c_k|(a_h:c_k)&(a_i:c_j)&((a_i,a_h)∈E_A) }; repeating the step S520 to obtain the adjacent objects of all the elements in the C.

S600: and finding the final result of the surface element matching in the T1 table by using a probability relaxation labeling method, wherein the specific method is as follows:

s610: construction of an initial matching probability matrix P⁽⁰⁾For data set a ═ a₁,a₂,..,a_iC ═ C₁,c₂,..,c_jI x j matching pairs can be formed; for any matching pair (a) in the candidate matching table T1_i：c_j) Calculating (a)_i：c_j) S (a) of_i,c_j) If c is obtained by looking up from the T1 table_jIs a_iCandidate matching object of (2), i.e. c_i∈CⁱThen a is further obtained by a geometric similarity calculation formula_iAnd c_jGeometric similarity s (a) of_i,c_j) (ii) a If it is

Let s (a)_i,c_j) 0; finally using the formula

Computing initial match probabilities in a probabilistic relaxation iterative framework

The specific expression of the geometric similarity calculation formula is shown in formula (6):

wherein s is_dis(a_i,c_j)、s_size(a_i,c_j)、s_ori(a_i,c_j) And s_shape(a_i,c_j) Are respectively a_iAnd c_jThe distance similarity, the size similarity, the direction similarity and the shape similarity are calculated according to the following specific calculation formula:

wherein the content of the first and second substances,

and

are respectively a_iAnd c_jThe geometric center point of (a);

wherein, Area (a)_i) Denotes a_iArea of entity, Area (c)_j) Denotes c_jArea of (a), max (Area (a)_i),Area(c_j) Is represented by a)_iAnd c_jThe greater of the two face entity areas;

wherein, theta (a)_i) And theta (c)_j) Are respectively a face element a_iAnd c_jDirection of (a), theta_τIs at [0, π]A direction threshold within a range;

wherein the content of the first and second substances,

and

is the cumulative angle between the tangent to the polygon in the counterclockwise direction and the x-axis, x representing the side length of the polygon,

to represent

And

the larger of the two values. Since the equation (10) is susceptible to the starting point when calculating the similarity of the polygon shape, in order to solve this problem, embodiment 1 adopts the intersection point of the polygon outline and the inertia axis as the starting point described by the rotation angle function.

Let in

θ(a_i) And theta (c)_j) By using the direction of the polygonal inertia axis, the formula can be obtained:

wherein the content of the first and second substances,

for the geometrical moments of the polygon, the calculation formula is as follows:

wherein (x)₀,y₀) Is the coordinate of the geometric center point of the polygon, and according to the formula (11), theta can be used to calculate the difference between the two points

The solution of (1) is a solution corresponding to the long inertia axis. Few surface elements, e.g. squares and circles, without obvious orientation, let the orientation threshold

The basis for judging whether the section elements have obvious directions is as follows: of surface elements

The face element is considered to have no apparent orientation, where l_shortIs the length of the short inertia axis,/_longIs the length of the long inertia axis.

S620: calculating any two pairs of candidate matching pairs (a)_i：c_j) And (a)_h：c_k) Coefficient r of compatibility between_ij(h, k); said r_ijThe formula for calculating (h, k) is shown in formula (12):

r_ij(h,k)＝rel_dis(ih；jk)*rel_size(ih；jk)*rel_ori(ih；jk)*rel_shape(ih；jk) (12)

wherein rel_dis(ih；jk)、rel_size(ih；jk)、rel_ori(ih; jk) and rel_shape(ih; jk) are each a_iAnd c_jThe relative distance, the relative size, the relative direction and the relative shape of the steel plate are specifically calculated according to the following formula:

s630: for any matching pair (a)_i：c_j) Calculating the support coefficient thereof

Wherein t is the number of iterations; t is 0 in the first iteration;

s640: starting iteration, updating the matching probability matrix P^(t)(ii) a For any matching pair (a)_i：c_j) The iterative calculation formula of the matching probability is as follows:

s650: judging whether the iteration can be ended; computing after each iteration

If min (delta P) is less than or equal to sigma, finishing iteration to obtain a final matching probability, and outputting a matching probability result matrix P based on convergence; for any element in P, if P_ijNot less than 0.5, then (a)_i：c_j) Are matched pairs; due to c_jIs formed by b_j1,b_j2,..,b_jmThe final matching result (a) is obtained_i：b_j1,b_j2,..,b_jm) (ii) a Otherwise a_iAnd b_j1,b_j2,..,b_jmMismatch is not achieved; if min (Δ p)>σ, return to step S620 to continue the iteration.

By the multi-scale surface element matching method described in embodiment 1, the correspondence between the crowd-sourced geographic information and the reference geographic data can be established, and the position accuracy, the shape accuracy, and the integrity of the crowd-sourced geographic information can be determined. The processed data can be used to: base maps of various thematic maps of vehicle navigation, city planning, city management and the like are drawn; constructing a navigation software map and a sky map; and establishing the relationship among map elements of different ages, and analyzing the urban expansion condition and the urban transition condition.

Example 2:

embodiment 2 is a multi-scale surface element matching system combining shape and environmental characteristics based on embodiment 1, and includes an MBR combinatorial optimization algorithm identifying homonymous MBR module, a bidirectional area overlap method screening module, a module for establishing a potential matching pair adjacency relation, a module for obtaining an initial matching probability by geometric matching, a probability relaxation matrix iteration module, and a judgment module;

the MBR combined optimization algorithm identifies the MBR modules with the same name, and identifies the corresponding MBRs of the elements with the same name in the data sets A and B to be matched based on the MBR combined optimization algorithm of the steps 100-240;

the bidirectional area overlap screening module maps MBR (a) based on the method of step 300_i) Is aligned with the center point of MBR (λ), and then using a stack analysis method, a after alignment shift is judged_iAnd b_j1,b_j2,..,b_jmWhether the overlapping elements of (1) satisfy the following condition; such as b_jmIf the condition is satisfied, then b_jmReserving; otherwise b_jmIs screened out;

the module for establishing the adjacency relation of the potential matching pairs respectively establishes the adjacency relation between the elements in the data sets A and C based on the methods of the steps 400 to 520;

the geometric matching initial matching probability module builds an initial matching probability matrix P based on the method of step 610⁽⁰⁾For data set a ═ a₁,a₂,..,a_IC ═ C₁,c₂,..,c_JI multiplied by J matched pairs can be formed; for any matching pair (a) in the candidate matching table T1_i：c_j) Calculating their geometric similarity s (a)_i,c_j) And then further calculate their initial probabilities

The probability relaxation matrix iteration module is based on the methods in the steps 620-640, and iterates to obtain the final probability value of the surface element matching by using a probability relaxation marking method;

the determination module is based on the method of step 650 for (a)_i：c_j) If the final matching probability is more than 0.5, judging that the final matching probability is a matching pair; otherwise, judging the result is not matched.

Embodiment 2 can adapt to most of the face element data matching scene settings by setting parameters. The set parameters comprise a buffer query distance, an MBRCO (minimum area outsourcing rectangle combination optimization algorithm) search threshold and a probability relaxation iteration threshold. The buffer query distance represents the position deviation distance of the homonymous elements, the MBRCO search threshold represents the change condition of the area size of the homonymous entities, and the iteration threshold of the probability relaxation method represents the number of control iterations.

The embodiment 1 and the embodiment 2 of the invention overcome the influence of position deviation on the matching of the face elements, and have good adaptability and high matching accuracy; by setting a buffer query distance, an MBRCO search threshold and a probability relaxation method iteration threshold, most of face element data matching scenes can be adapted, and the matching accuracy can reach more than 93%; the method effectively combines the advantages of the MBR combined algorithm and the probability relaxation iterative algorithm, effectively aggregates potential matching pairs by utilizing the MBR combined optimization algorithm, finds correct matching pairs by utilizing the probability relaxation iterative method, has good identification capability on complex matching, and avoids the defect of manually setting a matching threshold.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A multi-scale surface element matching method combining shape and environmental characteristics is characterized by comprising the following steps:

s000: acquiring a multi-scale surface element data packet of geographic information, taking vector space data of the multi-scale surface element data packet as a matching data source, and preprocessing the matching data source to obtain a data set to be matched;

s100: two data sets to be matched with different scales are respectively marked as A ═ a₁，a₂，..，a_IB ═ B₁，b₂，..，b_MAny one element in A is marked as a_iEstablishing a_iThe buffer area of (2) records the buffer distance as d_τCalculating the data set B falling into a by a superposition analysis method_iIs marked as a_iIs marked as Bⁱ＝{b₁，b₂，..，b_NWherein N is BⁱThe number of the middle elements; obtaining BⁱMBR of each surface element, BⁱEach element in (1) is respectively according to x of minimum x coordinate in MBR_minMaximum x coordinate x in MBR_maxMinimum y-coordinate y in MBR_minAnd the maximum y-coordinate y in MBR_maxSorting is carried out, and sorting results are respectively expressed as O_xmin，O_xmax，O_yminAnd O_ymax；

S200: constructing an N-fork complete solution space tree with the depth of 5, and searching a by adopting a backtracking algorithm-based MBR (Membrane biological reactor) combined optimization algorithm according to a depth-first principle_iOf MBR, wherein a_iMBR of (a) is denoted as MBR_i)；

S300: each MBR (a)_i) Is aligned with the geometric center point of MBR (lambda), and a bidirectional area overlapping method is used for obtaining a result b of potential matching pairs_j1，b_j2，..，b_jm；

S400: a new potential matching correspondence table T1 is established, and the result of the potential matching pair obtained in S300 is represented as (a)_i：b_j1)、(a_i：b_j1，b_j2)，..，(a_i：b_j1，b_j2，..，b_jm) Wherein (m is less than or equal to N), respectively calculating b_j1，{b_j1，b_j2}，..，{b_j1，b_j2，，..，b_jmObtaining new face element c by convex hull₁，c₂，..，c_jThus, a new potential matching pair correspondence (a) is obtained_i：c₁)，(a_i：c₂)，..，(a_i：c_j) And a is_iIs a candidate ofSet of counterparts Cⁱ＝{c₁，c₂，..，c_jRecording the corresponding relation in T1; repeating the step S400 to obtain potential matching pairs of all elements in the data set a, calculating their convex hulls to obtain a new data set C ═ C₁，c₂，..，c_jFourthly, a new potential matching corresponding table T1 is finally established;

s500: establishing adjacency relations between elements in the data sets A and C respectively;

s600: and acquiring the final result of the surface element matching in the T1 table by using a probability relaxation marking method.

2. The method for matching elements of multi-scale surfaces with combination of shape and environment features as claimed in claim 1, wherein the specific methods of step S000 include format conversion, topology inspection and geometric coordinate conversion.

3. The method for matching elements of multi-scale surface with combination of shape and environmental features as claimed in claim 1, wherein the specific method of step S200 is as follows:

s210: establishing a complete N-way solution space tree with the depth of 5, wherein the root node of the N-way solution space tree is o₀，o₀Deriving N sub-nodes, wherein each sub-node derives N sub-nodes respectively; the N-way spatial tree is a full N-way tree, and comprises (N)⁴+1) nodes, each node being BⁱThe element (1) of (1); wherein the second level sub-nodes are according to O_xminArranged, third level sub-nodes according to O_xmaxArranged, fourth level of sub-nodes according to O_yminArranged according to the fifth level of sub-nodes O_ymaxArranging;

s220: defining a solution vector lambda, searching the whole solution space tree from a root node according to a depth priority principle, and traversing to o₀When so, then λ ═ null, null, null }; if the second level node is searched, λ ═ o₁Null, null, null }; if the node of the third layer is searched, λ ═ o₁，o₂Null, null }; if searchingTo the node of the fourth layer, λ ═ o₁，o₂，o₃Null }; when the fifth level node is searched, λ ═ o₁，o₂，o₃，o₄}；o₁，o₂，o₃And o₄Are respectively BⁱA certain element of (1);

s230: defining a constraint function, carrying out condition limitation on the solution space tree when the solution space tree is searched and nodes of each level are traversed, continuing to move the search tree downwards after the condition is met, and returning to the previous level if the condition is not met;

s240: solving lambda, and traversing the search tree in a depth-first order from the root node; and finding all solutions lambda meeting the constraint condition.

4. The method for matching elements of a multi-scale surface combining shape and environmental features as claimed in claim 3, wherein said constraint function in step S230 is expressed as:

wherein w and h are respectively MBR (a)_i) Is the solution vector of the backtracking algorithm, width (lambda) and height (lambda) are the width and height of MBR (lambda), respectively, epsilon is the search threshold, epsilon belongs to [0.2, 0.5 ]]；

Where o is the junction of the fourth or fifth layer, x_min(lambda) and x_max(λ) is the minimum and maximum abscissa of MBR (λ), respectively, λ ═ o₁，o₂，null，null}；

y_min(o)≤y_min(λ) (3)

Where o is the node of the fourth layer, y_min(o) and y_min(λ) is the minimum ordinate of MBR (o) and MBR (λ), respectively, λ ═ o₁，o₂，null，null}；

Where o is the junction of the fifth layer and y_max(o) and y_max(λ) is the maximum ordinate of MBR (o) and MBR (λ), respectively, λ ═ o₁，o₂，o₃，null}；

When traversing to the third level node, checking whether the node meets the formula (1); when traversing to the node of the fourth layer, checking whether the node meets the formulas (2) and (3); when traversing to the fifth layer node, checking whether the node satisfies the formulas (2) and (4).

5. The method for matching elements of multi-scale surfaces by combining shape and environmental features as claimed in claim 1, wherein the specific method of step S300 is: MBR (a)_i) Is aligned with the center point of MBR (λ), and then using a stack analysis method, a after alignment shift is judged_iAnd b_j1，b_j2，..，b_jmWhether or not the overlapping elements satisfy the following condition:

wherein, M (a)_i) Representing a after alignment translation_i，Area(M(a_i)∩b_jm) Is represented by M (a)_i) And b_jmOverlap Area of (a), min (Area (a)_i)，Area(b_jm) Is represented by a)_iAnd b_jmThe smaller value of the areas of the two surface entities, wherein gamma is an area overlapping threshold value; such as b_jmIf the condition is satisfied, then b_jmReserving; otherwise b_jmAre screened out.

6. The method for matching elements of multi-scale surface with combination of shape and environmental features as claimed in claim 1, wherein the specific method of step S500 is as follows:

s510: establishing an adjacency relation of the data set A, and acquiring geometric center points of all elements in the data set A and recording the geometric center points as V_AAccording to V_AConstruction of a Delaunay triangulation network, denoted G_DT(A)＝(V_A，E_A) In which E_AIs a set of triangulation edges; for any surface element a in A_iAdjacent to the object a_hIs at G_DT(A) And a_iThe elements with connected edges, denoted by N (a)_i)＝{a_h|(a_i，a_h)∈E_AObtaining adjacent objects of all the elements in the A;

s520: establishing a contiguous object of data set C, looking up a according to T1 table_iIs marked as (a)_i：c_j)，a_i∈A，c_jE is C; a is calculated in step S510_iAdjacent object a of_hObtaining a_hAll potential matching pairs of (a)_h：c_k) (ii) a Then, c is established_jBuffer area of (2), recording the buffer distance as d_τ(ii) a By means of a superposition analysis method, c_kWhether or not at a_iIn a buffer zone, e.g. c_kAt a_iIn the buffer area of (1), then c_kIs c_jThe adjacent object of (1); if not, c_kIs other than c_jIs marked as N (c)_j)＝{c_k|(a_h：c_k)&(a_i：c_j)&((a_i，a_h)∈E_A) }; repeating the step S520 to obtain the adjacent objects of all the elements in the C.

7. The method for matching elements of multi-scale surface with combination of shape and environmental features as claimed in claim 1, wherein the specific method of step S600 is as follows:

s610: construction of an initial matching probability matrix P⁽⁰⁾For data set a ═ a₁，a₂，..，a_iC ═ C₁，c₂，..，c_jI x j matching pairs can be formed; for any matching pair (a) in the candidate matching table T1_i：c_j) Calculating (a)_i：c_j) S (a) of_i，c_j) If c is obtained by looking up from the T1 table_jIs a_iCandidate matching object of (2), i.e. c_i∈CⁱThen a is further obtained by a geometric similarity calculation formula_iAnd c_jGeometric similarity s (a) of_i，c_j) (ii) a If it is

Let s (a)_i，c_j) 0; finally using the formula

Computing initial match probabilities in a probabilistic relaxation iterative framework

S620: calculating any two pairs of candidate matching pairs (a)_i：c_j) And (a)_h：c_k) Coefficient r of compatibility between_ij(h，k)；

S630: for any matching pair (a)_i：c_j) Calculating the support coefficient thereof

Wherein t is the number of iterations;

s640: starting iteration, updating the matching probability matrix p^(t)(ii) a For any matching pair (a)_i：c_j) The iterative calculation formula of the matching probability is as follows:

s650: judging whether the iteration can be ended; computing after each iteration

If min (delta p) is less than or equal to sigma, iteration is ended to obtain final matching probability, and the matching probability based on convergence is outputA result matrix P; for any element in P, if P_ijNot less than 0.5, then (a)_i：c_j) Are matched pairs; due to c_jIs formed by b_j1，b_j2，..，b_jmThe final matching result (a) is obtained_i：b_j1，b_j2，..，b_jm) (ii) a Otherwise a_iAnd b_j1，b_j2，..，b_jmMismatch is not achieved; if min (Δ p) > σ, return to step S620 to continue the iteration.

8. The method for matching elements of multi-scale surface with combination of shape and environmental features as claimed in claim 7, wherein the geometric similarity calculation formula in step S610 is expressed as formula (6):

wherein s is_dis(a_i，c_j)、s_size(a_i，c_j)、s_ori(a_i，c_j) And s_shape(a_i，c_j) Are respectively a_iAnd c_jThe distance similarity, the size similarity, the direction similarity and the shape similarity are calculated according to the following specific calculation formula:

wherein the content of the first and second substances,

and

are respectively a_iAnd c_jThe geometric center point of (a);

wherein, Area (a)_i) Denotes a_iArea of entity, Area (c)_j) Denotes c_jArea of (a), max (Area (a)_i)，Area(c_j) Is represented by a)_iAnd c_jThe greater of the two face entity areas;

wherein, theta (a)_i) And theta (c)_j) Are respectively a face element a_iAnd c_jDirection of (a), theta_τIs at [0, π]A direction threshold within a range;

wherein the content of the first and second substances,

and

is the cumulative angle between the tangent to the polygon in the counterclockwise direction and the x-axis, x representing the side length of the polygon,

to represent

And

the greater of the two values; the intersection point of the polygonal outline and the inertia axis is used as the starting point described by the formula (10);

let is described in formula (9)

θ(a_i) And theta (c)_j) By using the direction of the polygonal inertia axis, the formula can be obtained:

wherein the content of the first and second substances,

for the geometrical moments of the polygon, the calculation formula is as follows:

wherein (x)₀，y₀) Is the coordinate of the geometric center point of the polygon, and according to the formula (11), theta can be used to calculate the difference between the two points

And (3) taking a solution corresponding to the long inertia axis.

9. The method for matching elements of a multi-scale surface combining shape and environmental features as claimed in claim 7, wherein r in step 620 is_ijThe formula for calculating (h, k) is shown in formula (12):

r_ij(h，k)＝rel_dis(ih；jk)*rel_size(ih；jk)*rel_ori(ih；jk)*rel_shape(ih；jk) (12)

wherein rel_dis(ih；jk)、rel_size(ih；jk)、rel_ori(ih; jk) and rel_shape(ih; jk) are each a_iAnd c_jThe relative distance, the relative size, the relative direction and the relative shape of the steel plate are specifically calculated according to the following formula:

10. a multi-scale surface element matching system combining shape and environmental characteristics is characterized by comprising an MBR combined optimization algorithm identification homonymous MBR module, a bidirectional area overlapping method screening module, a potential matching pair adjacency relation establishing module, a geometric matching obtaining initial matching probability module, a probability relaxation matrix iteration module and a judging module;

the MBR combined optimization algorithm identifies the MBR modules with the same name, and identifies the corresponding MBRs of the elements with the same name in the data sets A and B to be matched based on the MBR combined optimization algorithm of the steps 100-240;

the bidirectional area overlap screening module maps MBR (a) based on the method of step 300_i) Is aligned with the center point of MBR (λ), and then using a stack analysis method, a after alignment shift is judged_iAnd b_j1，b_j2，..，b_jmWhether the overlapping elements of (1) satisfy the following condition; such as b_jmIf the condition is satisfied, then b_jmReserving; otherwise b_jmIs screened out;

the module for establishing the adjacency relation of the potential matching pairs respectively establishes the adjacency relation between the elements in the data sets A and C based on the methods of the steps 400 to 520;

the geometric matching initial matching probability module builds initial matching based on the method of step 610Probability matrix P⁽⁰⁾For data set a ═ a₁，a₂，..，a_IC ═ C₁，c₂，..，c_JI multiplied by J matched pairs can be formed; for any matching pair (a) in the candidate matching table T1_i：c_j) Calculating their geometric similarity s (a)_i，c_j) And then further calculate their initial probabilities

The probability relaxation matrix iteration module is based on the methods in the steps 620-640, and iterates to obtain the final probability value of the surface element matching by using a probability relaxation marking method;

the determination module is based on the method of step 650 for (a)_i：c_j) If the final matching probability is more than 0.5, judging that the final matching probability is a matching pair; otherwise, judging the result is not matched.

Images