CN116361665B - Building object plane element matching method and system based on environment information - Google Patents

Abstract

The invention discloses a building surface element matching method and system based on environmental information, comprising the following steps: acquiring potential matching pairs from two face element sets to be matched; calculating initial probability between each pair of potential matching pairs, and constructing an initial probability matrix based on the initial probability; defining a neighborhood object of each pair of potential matching pairs, and calculating a compatibility coefficient between each pair of potential matching pairs and the neighborhood matching pairs; and updating the initial probability matrix in an iteration mode according to the compatibility coefficient until the probability difference between the two iterations is smaller than a preset threshold value, stopping the iteration, and obtaining a final matching result. The matching method does not need to set a similarity threshold value and weight and does not need to train samples; and the defect that the MBR combination optimization algorithm cannot distinguish adjacent buildings with similar shapes is overcome, and the matching precision is high.

Description

Building object plane element matching method and system based on environment information

Technical Field

The invention belongs to the field of geographic information science, and particularly relates to a building surface element matching method and system based on environmental information.

Background

Element matching is a fundamental technology in the fields of spatial data processing and application, and is widely applied to updating, maintaining and fusing of geospatial data. In recent years, "spontaneous geographic information" (VGI) has become possible, which is a new way of spontaneously (voluntarily) participating in creating and sharing geographic information by non-professional mappers based on web2.0, mobile intelligent terminals, positioning technology (GPS or beidou), location services (Location Based Service, LBS) technology and open high-resolution remote sensing images, as well as on personal spatial-aware geographic knowledge. Compared with traditional professional mapping data, VGI has the advantages of rich semantic information, strong action, low acquisition cost and the like. Element matching becomes a key technology for VGI quality evaluation and VGI-based data updating.

The traditional element matching method is mainly used for solving the problems of position offset and different levels of detail (LOD) existing between data to be matched. However, VGI-based matching methods face new challenges. Firstly, the LOD of VGI data is not uniform, which varies from object to object; secondly, the position accuracy of VGI data is unstable; finally, volunteers created VGI independent of the update cycle. In contrast, authoritative geospatial data has well-defined production and update specifications, which ensure that the data has strict metrology accuracy. Therefore, the conventional element matching method is difficult to be directly used for identifying VGI matching objects, and the improvement of the conventional matching method is necessary.

Therefore, in 2018, an MBR combination optimization algorithm is provided, and the problems of position offset and different LODs in traditional matching are solved well. However, typical MBR combinatorial optimization algorithms also suffer from the following drawbacks: (1) It requires setting a similarity threshold, but the data quality of VGI is not balanced in spatial distribution, and it is impractical to find all correct matching pairs by a fixed threshold; (2) It cannot distinguish between adjacent similarly shaped buildings, but there are many adjacent similarly shaped buildings in a city, such as residential buildings of one cell or dormitory buildings of a school, which have highly similar shapes. Therefore, further improvements to MBR combination optimization algorithms are needed to meet the need for less accurate spatial data matching.

Disclosure of Invention

The invention provides a building surface element matching method and system based on environment information, which are used for solving the problem that matching pairs cannot be identified correctly in the prior art.

In order to achieve the above object, the present invention provides a method for matching elements of a building surface based on environmental information, including:

acquiring potential matching pairs from the two building surface elements to be matched in a centralized manner;

calculating initial probability between each pair of potential matching pairs, and constructing an initial probability matrix based on the initial probability;

defining a neighborhood object of each pair of potential matching pairs, and calculating a compatibility coefficient between each pair of potential matching pairs and the neighborhood matching pairs; and updating the initial probability matrix P according to the compatible coefficient iteration until the probability difference between the two iterations is smaller than a preset threshold value, stopping the iteration, and obtaining a final matching result.

Optionally, calculating an initial probability between each pair of potential matches, the process of constructing an initial probability matrix P based on the initial probabilities includes,

for each pair of potential matching pairs (a ₁ ，a ₂ ，..，a _i ：b ₁ ，b ₂ ，..，b _j ) Will { a } ₁ ，a ₂ ，..，a _i Sum { b } ₁ ，b ₂ ，..，b _j Viewed as an aggregated polygon, denoted as a respectively ⁱ And B ^j Then B is ^j Is A ⁱ Is defined as B ^j ∈C(A ⁱ )，C(A ⁱ ) Represented by A ⁱ A set of potentially matching objects; representing all potential matching relationships as M _i：j ＝{(A ⁱ ：B ^j ) I=1, 2, p. j=1, 2,..q } after q } an initial probability matrix P is established.

Optionally, the calculation formula of the initial probability matrix P is:

wherein M (A) ⁱ ) Is vector (Deltax, deltay) translating polygon A ⁱ Is ((. DELTA.x,. DELTA.y) is the function of MBR (A ⁱ ) And MBR (B) ^j ) Deviation between centroid coordinates;

the MBR centroid coordinate calculation formula is as follows:

x ₀ ＝(x _max -x _min )/2，

y ₀ ＝(y _max -y _min )/2，

(x _min ，y _min ) And (x) _max ，y _max ) The minimum and maximum coordinates of the MBR, respectively.

Optionally, the calculation formula of the adjustment coefficient α is:

wherein,and->Respectively represent A ⁱ And B ^j Is a sum of all the face element contour points.

Optionally, the defining the neighborhood object for each pair of potential matching pairs includes: finding the AND a in the potential matching pair correspondence table by recursion _i All potential matching pairs of interest;

suppose that detection:

(A ⁱ ：B ^j )＝(a _i ，a _i+1 ：b _j )，(A ⁱ⁺¹ ：B ^j+1 )＝(a _i+1 ，a _i+2 ：b _j+1 ) And (A) ⁱ⁺² ：B ^j+2 )＝(a _i+2 ，a _i+3 ：b _j+2 )，

Then it will: gA ⁱ ＝A ⁱ ∪A ⁱ⁺¹ ∪A ⁱ⁺² ＝{a _i ，a _i+1 ，a _i+2 ，a _i+3 }；

MBR extraction (gA) ⁱ ) And so on, obtain the point set V _A The method comprises the steps of carrying out a first treatment on the surface of the According to the point set V _A Establishing Delaunay triangle network with expression G _DT (A)＝(V _A ，E _A )，E _A For the collection of triangle net edges, MBR (A ⁱ ) Is defined by the geometric center point of (A) and MBR (A) ^h ) Is defined by E _A Either one of the edges is connected, A ⁱ And A ^h Are neighbors of each other, A ^h Is A ⁱ Is a neighbor of the element in the neighborhood set;

A ⁱ is defined as: n (A) ⁱ )＝{A ^h |(A ⁱ ，A ^h )∈E _A }

Wherein N (A) ⁱ ) Represented by A ⁱ Is defined by:

N(A ⁱ )＝N(A ⁱ⁺¹ )＝N(A ⁱ⁺² )＝N(gA ⁱ )；

B ^j is defined as:

wherein N (B) ^j ) Denoted as B ^j Neighborhood set of B ^j ∈C(A ⁱ ) Denoted as B ^j And A ⁱ Is a potential matching pair and,representation B ^j And B ^k Not containing the same element, B ^k ∈C(A ^h ) Denoted as B ^k And A ^h Is a potential matching pair, (A) ⁱ ，A ^h )∈E _A Representation A ⁱ And A ^h Is a neighbor.

Optionally, the process of calculating the compatibility coefficient between each pair of potential matching pairs and its neighborhood matching pairs includes: defining a compatibility coefficient through the distance similarity and the area overlapping rate, and eliminating ambiguity between correct matching pairs in each iteration based on the compatibility coefficient;

design (A) ⁱ ：B ^j ) And (A) ^h ：B ^k ) Is an adjacent potential matching pair, then (A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The compatibility coefficient between the two is calculated as follows:

c(A ⁱ ，B ^j ；A ^h ，B ^k )＝r _dis (A ⁱ ，B ^j ；A ^h ，B ^k )×r _overlapArea (A ⁱ ，B ^j ；A ^h ，B ^k )

wherein the relative distance r _dis (A ⁱ ，B ^j ；A ^h ，B ^k ) The calculation formula is as follows:

wherein Dis (A) ⁱ ，B ^j ) Representation A ⁱ And B ^j The distance between the two is calculated as follows:

wherein,and->Respectively represent MBR (A) ⁱ ) And MBR (B) ^j ) Centroid point coordinates of (c);

(A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The relative area overlap ratio r of (2) _overlapArea (A ⁱ ，B ^j ；A ^h ，B ^k ) The calculation formula of (2) is as follows:

wherein,M(A ⁱ ) Is vector (Deltax, deltay) translating polygon A ⁱ α is the adjustment coefficient.

Optionally, the initial probability matrix P is iteratively updated according to the compatibility coefficient until the probability difference between the two iterations is smaller than a preset threshold, the iteration is stopped, and the process of obtaining the final matching result includes,

in updating the probability matrix for each iteration, (A) ⁱ ：B ^j ) All the support is obtained from the neighborhood of the support function, and the calculation formula of the support function is as follows:

finally, updating the probability matrix according to the calculated support, wherein the calculation formula is as follows,

hold after each iterative update

When min (p) ^(t+1) (A ⁱ ：B ^j )-p ^(t) (A ⁱ ：B ^j ) And) 0.001, the iteration is terminated; for A ⁱ ，B ^j ∈C(A ⁱ ) If p (A) ⁱ ：B ^j ) =max, then (a ⁱ ：B ^j ) And the final matching result.

In order to achieve the above object, the present invention further provides a building surface element matching system based on environmental information, including:

the data acquisition module is used for acquiring potential matching pairs from the two face element sets to be matched;

the calculation module is used for calculating the initial probability between each pair of potential matching pairs, and constructing an initial probability matrix based on the initial probability, wherein the error of the area overlapping rate between homonymous surface elements is required to be balanced through adjusting the coefficient alpha;

the matching module is used for defining a neighborhood object of each pair of potential matching pairs and calculating a compatibility coefficient between each pair of potential matching pairs and the neighborhood matching pairs; and the method is also used for iteratively updating the initial probability matrix according to the compatible coefficient until the probability difference between the two iterations is smaller than a preset threshold value, and stopping the iteration to obtain a final matching result.

The invention discloses the following technical effects:

according to the building surface element matching method and system based on the environmental information, the problems of position difference, LOD difference, threshold setting and automatic matching are solved by improving a minimum bounding rectangle combination optimization (MBR combination optimization) algorithm, and a good result is obtained in VGI data matching with lower data quality. In particular, the method has the following advantages: (1) The setting of similarity threshold and weight is avoided, training samples are not needed, and the process is realized based on space context information and an optimization algorithm; (2) The defect that the MBR combination optimization algorithm cannot distinguish adjacent buildings with similar shapes is overcome; (3) Under the condition that a large number of invalid potential matching pairs are obtained for the MBR combination optimization algorithm, the definition of the neighborhood and the compatibility coefficient in the probability relaxation algorithm is optimized to adapt to the faced new situation.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed 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 invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic diagram of identifying potential matching pairs based on an MBR combination optimization algorithm according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a probability relaxation iteration process for expressing a matching evaluation problem according to an embodiment of the present invention; wherein (a) a schematic diagram of a polygon is constructed in the real world; (b) a matching relationship diagram; (c) constructing a schematic representation of an initial probability matrix;

FIG. 3 is a schematic diagram of a neighborhood structure according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.

The invention provides a building surface element matching method based on environmental information, which comprises the following two parts:

(1) Identifying potential matching pairs based on a traditional MBR combination optimization algorithm;

(2) The correct matching pair is found from the potential matching pairs based on an improved probability relaxation algorithm.

In the present embodiment, there are the following advantages: (1) The similarity threshold and the weight are not required to be set, and the sample is not required to be trained; (2) Under the condition that a large number of invalid candidate matching pairs are obtained aiming at the MBR combination optimization algorithm, a compatible coefficient and a neighborhood calculation model are improved; (3) The method overcomes the defect that the MBR combination optimization algorithm cannot distinguish adjacent buildings with similar shapes. Similar buildings are numerous in cities, such as residential or school dormitory buildings of a cell. By using the building geospatial data set disclosed by the OSM and the Goldmap to carry out a large number of experiments, the method is verified, and the experimental results show that the automatic matching method provided by the method is superior to all MBR combination optimization methods based on the threshold value, the accuracy rate reaches 97.8%, and the recall rate reaches 99.2%.

In particular, for the identification of matching relationships, the objective is to find potential matching pairs. In general, there are four cases of matching relations: one-to-zero (1:0), one-to-one (1:1), one-to-many (1:N), and many-to-many (M:N).

Wherein, 1:N and M:N match is a complex correspondence. For recognition of the polygon matching relationship, a bidirectional area overlapping method (bidirectional area overlapping method) is most commonly used. The bidirectional area overlapping method realizes the aggregation of matching pairs of 1:N and M:N, but has the limitation of higher position accuracy of data to be matched. To solve this problem, an MBR combination optimization algorithm is proposed to minimize the position difference between homonymous face elements, and then a two-way area overlap method is used to identify the matching relationship.

Further optimizing scheme, the embodiment discloses a face element matching method based on an MBR combination optimizing algorithm and a probability relaxation marking algorithm.

Specifically, identifying potential matching pairs based on the MBR combination optimization algorithm includes:

if the two face elements are homonymous elements, they generally have similar shapes. In this case, their Minimum Bounding Rectangles (MBRs) are also close, as shown in fig. 1. Thus, having a similar MBR is a prerequisite when two elements (or sets of elements) correspond to each other. The MBR combination optimization algorithm is a backtracking-based algorithm, and can quickly find the same-name MBR. Thus, the present embodiment utilizes an MBR combination optimization algorithm to identify potential matching pairs. The MBR combination optimization algorithm is described below.

Let a= { a _i I=1, 2,..m } and b= { B _j I j=1, 2,..n } is two sets of elements to be matched, a _i From data sets a and b _j From data set B. Suppose B ⁱ ＝{b ₁ ，b ₂ ，..，b _j Is } is a _i Obtained by buffer analysis (buffer analysis radius d _τ =30 meters).

As shown in table 1, first, a complete j-ary tree with a depth of 5 is created, the root node of the tree being the virtual node. The root node splits out 5 child nodes, each representing a B ⁱ The face elements of these nodes are x according to their MBR _min Arranged (MBR of one element is defined by x _min ，y _min ，x _max And y _max Defined by the specification). Each node of the second layer derives j child nodes of the third layer, which are based on x of each face element _min The values are ordered from large to small. Similarly, the parent node derives the fourth level (from y for each facet element _max Value order from small to large) and a fifth stage (x according to each face element _max Value ordered from big to small) j sub-itemsAnd (5) a node.

Second, search for all the homonymous MBRs. The search tree is traversed recursively in depth-first order starting from the root.

TABLE 1

Algorithm 1.MBR combination optimization Algorithm

However, some of the homonymous elements have significantly different shapes. To solve this problem, the present embodiment relaxes the setting of the threshold epsilon. An initialization stage, setting epsilon=0.2; then, the search is stopped when ε=ε+0.1 and max (ε) =0.5 are relaxed.

After all the MBR pairs with the same name are obtained, calculating a _i Is defined, i.e. the deviation between the centroid coordinates of the homonymous MBR. The centroid coordinates are described as follows: x is x ₀ ＝(x _max -x _min )/2，y ₀ ＝(y _max -y _min )/2. By the translational correction, the positional deviation between the homonymous elements can be greatly improved.

Further, the face elements are aggregated using a two-way area overlapping method. In an aligned scenario, irrelevant surface elements are excluded from the aggregation. The potential matching pairs determined therefrom using equation (1) are as follows:

wherein M (a) _i ) Is vector (Deltax, deltay) translation plane element a _i Is a function of (2); γ is a threshold value, and γ=0.3 is generally set. If polygon a _i And a plurality of polygons b ₁ ，b ₂ ，..，b _j Overlap b ₁ ，b ₂ ，..，b _j Aggregation into a polygon b _j '. Next oneStep, if b _j ' and a ₁ ，a ₂ ，..，a _p Overlapping ofWill a ₁ ，a ₂ ，..，a _p ，a _i And also polymerized. Potential pairs of homonym elements are identified by recursion of the above process until the aggregated polygon overlaps all of the polygons in the corresponding dataset.

Further, for matching evaluation, it is to find the correct matching pair from the potential matching pair. The method specifically comprises the following steps: calculating the initial probability between each pair of potential matching pairs, so as to construct an initial probability matrix P; constructing a neighborhood and compatibility coefficients of a probability relaxation marking algorithm; the initial probability matrix P is iteratively updated until a final matching result is obtained.

Further, evaluating potential matching pairs based on a probability relaxation algorithm includes:

identifying the correct match pair from all potential match pairs based on the relaxation labeling algorithm includes three parts: (1) Expressing the matching evaluation problem as a relaxation marking process, and calculating an initial probability matrix P; (2) Constructing a neighborhood structure in the event that many potentially matching pairs conflict with each other, which is an important component of relaxed token matching; (3) The compatibility coefficients are calculated and then in each iteration the initial probability matrix P is updated with the compatibility coefficients of the two adjacent matching pairs until global agreement is reached.

Further, the element matching method based on the probability relaxation marking algorithm comprises the following steps:

the basic idea of the probability relaxation marking algorithm is to iteratively update the probability matrix using spatial context information until a globally consistent result is reached. In this embodiment, the probabilistic relaxation labeling algorithm finds the final matching pair taking into account spatial context and geometric consistency, provided that potential matching pairs have been obtained. For each pair of potential matches (a ₁ ，a ₂ ，..，a _i ：b ₁ ，b ₂ ，..，b _j ) The present embodiment will { a } ₁ ，a ₂ ，..，a _i Sum { b } ₁ ，b ₂ ，..，b _j Seen as an aggregated polygon, representing a respectively ⁱ And B ^j ，B ^j Is A ⁱ Is defined as B ^j ∈C(A ⁱ ) As shown in fig. 2 (a) and (b), all potential matching relationships are expressed as: m is M _i：j ＝{(A ⁱ ：B ^j ) I=1, 2, p. j=1, 2, q). Finally, an initial probability matrix P is established, as shown in fig. 2 (c). Here, an initial matching probability p (A ⁱ ：B ^j ) The calculation formula of (2) is as follows:

wherein M (A) ⁱ ) Is vector (Deltax, deltay) translating polygon A ⁱ (Δx, Δy) is the MBR (A ⁱ ) And MBR (B) ^j ) Deviation between centroid coordinates;

the MBR centroid coordinate calculation formula is as follows:

x ₀ ＝(x _max -x _min )/2，

y ₀ ＝(y _max -y _min )/2，

(x _min ，y _min ) And (x) _max ，y _max ) The minimum and maximum coordinates of the MBR, respectively.

Since the elements LOD in VGI vary from object to object, there may be a large imbalance in the area overlap ratio between the homonymous elements, as shown in fig. 2 (a ₁ ：b ₁ ，b ₂ ) And (a) ₂ ，a ₃ ：b ₃ ，b ₄ ). For this reason, the present embodiment proposes an adjustment coefficient α to balance the error, and the calculation formula of the adjustment coefficient is as follows:

wherein,and->Respectively represent A ⁱ And B ^j Is a sum of all the face element contour points. The larger the difference in the number of contour points between two homonymous elements, the larger the difference in LOD between them, which will pull down the area overlap ratio between them, the larger alpha is obtained in the formula, so that the area overlap ratio is effectively compensated.

Further, after the potential matching pair is obtained through the MBR combination optimization algorithm, the final matching result is obtained through iteration of the probability relaxation algorithm, which will be 1: n and M: the matching problem of N also translates to 1: 1. Because the probability relaxation algorithm is mainly used for identifying the one-to-one correspondence relationship between point sets, the embodiment converts all the 1:1, 1:N and M:N matching types into the 1:1 mode to calculate the matching confidence coefficient, and the accuracy of matching evaluation is greatly improved.

Further, constructing the neighborhood includes: the neighborhood is defined using Delaunay triangle subdivision. Delaunay triangulation has proven to be a powerful tool to capture spatial proximity when in spatial clustering. Since the MBR combination optimization algorithm is used to identify potential matching pairs, there are a large number of incorrect matching pairs in the resulting potential matching pairs, i.e., the number of incorrect matching pairs is greater than the number of correct matching pairs in all potential matching pairs. Thus, in this case, a number of adjacent potential matching pairs are in conflict with each other, as shown in FIG. 2 (b), (a) ₁ ：b ₃ ) And (a) ₂ ，a ₃ ：b ₃ ：b ₄ ) Are adjacent potential matching pairs, but they all contain b ₃ They cannot be simultaneously established (the same element can only belong to one matching pair). In this case, if the Delaunay triangulation network is directly used to construct the neighborhood structure, a large number of neighbors of the correct matching pair are incorrect matching pairs, and thus the correct matching pair is not effectively supported by the neighborhood matching pair, so that the probability relaxation iterative algorithm fails in matching. Thus, the present embodiment improves the neighborhood definition of the probability relaxation algorithm:

assume that there are three pairs of potential matching pairs:

(A ⁱ ：B ^j )＝(a _i ，a _i+1 ：b _j )；

(A ⁱ⁺¹ ：B ^j+1 )＝(a _i+1 ，a _i+2 ：b _j+1 )；

(A ⁱ⁺² ：B ^j+2 )＝(a _i+2 ，a _i+3 ：b _j+2 )；

in this case, A ⁱ ∩A ⁱ⁺¹ ＝a _i+1 ，A ⁱ⁺¹ ∩A ⁱ⁺² ＝a _i+2 Thus, adjacent potential matching pairs are conflicting. Here, the invention will be gA ⁱ ＝A ⁱ ∪A ⁱ⁺¹ ∪A ⁱ⁺² ＝{a _i ，a _i+1 ，a _i+2 ，a _i+3 Viewed as a union, gA ⁱ The neighborhood of (A) is A ⁱ 、A ⁱ⁺¹ And A ⁱ⁺² Is a neighborhood of (c). In this case, (A) ⁱ ：B ^j )、(A ⁱ⁺¹ ：B ^j+1 ) And (A) ⁱ⁺² ：B ^j+2 ) The evaluations are performed in the same neighborhood and they will all be effectively supported from the neighborhood matching pairs. Similarly, two adjacent matched pairs cannot contain the same element from data set B.

Further, the neighborhood acquisition steps are as follows:

first, find the sum a in the potential match correspondence table using recursion _i All potential matching pairs involved, assume (a ⁱ ：B ^j )＝(a _i ，a _i+1 ：b _j )，(A ⁱ⁺¹ ：B ^j+1 )＝(a _i+1 ，a _i+2 ：b _j+1 ) And (A) ⁱ⁺² ：B ^j+2 )＝(a _i+2 ，a _i+3 ：b _j+2 ) It is detected that the detection of the presence of a certain amount of a,

gA is to be gA ⁱ ＝A ⁱ ∪A ⁱ⁺¹ ∪A ⁱ⁺² ＝{a _i ，a _i+1 ，a _i+2 ，a _i+3 }。

Then, MBR (gA) ⁱ ) By this, a set of points V is obtained _A 。

Finally, according to the point set V _A A Delaunay triangle network is established, and the expression is as follows:

G _DT (A)＝(V _A ，E _A )，

E _A for the collection of triangle net edges, MBR (A ⁱ ) Is defined by the geometric center point of (A) and MBR (A) ^h ) Is defined by E _A Either one of the edges is connected, A ⁱ And A ^h Are neighbors of each other, A ^h Is A ⁱ As shown in fig. 3.

A ⁱ The neighborhood definition formula of (c) is as follows:

N(A ⁱ )＝{A ^h |(A ⁱ ，A ^h )∈E _A } (4)

here, N (A) ⁱ ) Represented by A ⁱ And defines:

N(A ⁱ )＝N(A ⁱ⁺¹ )＝N(A ⁱ⁺² )＝N(gA ⁱ )。

B ^j is defined as:

wherein N (B) ^j ) Denoted as B ^j Neighborhood set of B ^j ∈C(A ⁱ ) Denoted as B ^j And A ⁱ Is a potential matching pair and,representation B ^j And B ^k Not containing the same element, B ^k ∈C(A ^h ) Denoted as B ^k And A ^h Is a potential matching pair, (A) ⁱ ，A ^h )∈E _A Representation A ⁱ And A ^h Is a neighbor.

Further, a compatibility coefficient and an iterative update probability matrix are calculated:

the compatibility coefficients are responsible for disambiguating between correctly matched pairs in each iteration. The present embodiment is based on distance similarity and area overlapping rateTo define compatibility coefficients. For potential matches (A ⁱ ：B ^j ) And (A) ^h ：B ^k )，(A ⁱ ，A ^h )∈E _A The relative distance calculation formula is as follows:

wherein Dis (A) ⁱ ，A ^h ) The calculation formula is as follows:

wherein,and->Respectively represent MBR (A) ⁱ ) And MBR (B) ^j ) Center point coordinates.

(A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The relative area overlap ratio of (2) is calculated as follows:

wherein,M(A ⁱ ) Is vector (Deltax, deltay) translating polygon A ⁱ Is a function of (2).

(A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The compatibility coefficient between the two is calculated as follows:

c(A ⁱ ，B ^j ；A ^h ，B ^k )＝r _dis (A ⁱ ，B ^j ；A ^h ，B ^k )×r _overlapArea (A ⁱ ，B ^j ；A ^h ，B ^k ) (9)

in updating the probability matrix for each iteration, (A) ⁱ ：B ^j ) All support is obtained from its neighborhood, and the support function calculation formula is as follows:

finally, updating the probability matrix according to the calculated support, wherein the calculation formula is as follows,

hold after each iterative updateBetween two iterations

min(p ^(t+1) (A ⁱ ：B ^j )-p ^(t) (A ⁱ ：B ^j ) And) 0.001, the iteration terminates. For A ⁱ ，B ^j ∈C(A ⁱ ) If p (A) ⁱ ：B ^j ) =max, then (a ⁱ ：B ^j ) And the final matching result.

In an alternative embodiment, the invention also discloses a building face element matching system based on the environment information, which comprises:

the data acquisition module is used for acquiring potential matching pairs from the two face element sets to be matched;

the calculation module is used for calculating the initial probability between each pair of potential matching pairs, and constructing an initial probability matrix based on the initial probability, wherein the error of the area overlapping rate between homonymous surface elements is required to be balanced through adjusting the coefficient alpha;

the matching module is used for defining a neighborhood object of each pair of potential matching pairs and calculating a compatibility coefficient between each pair of potential matching pairs and the neighborhood matching pairs; and the method is also used for iteratively updating the initial probability matrix according to the compatible coefficient until the probability difference between the two iterations is smaller than a preset threshold value, and stopping the iteration to obtain a final matching result.

The above embodiments are only illustrative of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention, and various modifications and improvements made by those skilled in the art to the technical solutions of the present invention should fall within the protection scope defined by the claims of the present invention without departing from the design spirit of the present invention.

Claims (4)

1. The building face element matching method based on the environment information is characterized by comprising the following steps of:

acquiring potential matching pairs from the two building surface elements to be matched in a centralized manner;

calculating initial probability between each pair of potential matching pairs, and constructing an initial probability matrix P based on the initial probability, wherein errors of area overlapping rates between homonymous surface elements are required to be balanced through adjusting coefficients;

defining a neighborhood object of each pair of potential matching pairs, and calculating a compatibility coefficient between each pair of potential matching pairs and the neighborhood matching pairs; updating the initial probability matrix P according to the compatibility coefficient iteration until the probability difference between the two iterations is smaller than a preset threshold value, stopping the iteration, and obtaining a final matching result;

wherein, the calculation formula of the adjustment coefficient is:

wherein,and->Respectively represent A ⁱ And B ^j The sum of all the surface element contour points;

the process of calculating compatibility coefficients between each pair of potential matching pairs and its neighborhood matching pairs includes: defining a compatibility coefficient through the distance similarity and the area overlapping rate, and eliminating ambiguity between correct matching pairs in each iteration based on the compatibility coefficient;

design (A) ⁱ ：B ^j ) And (A) ^h ：B ^k ) Is an adjacent potential matching pair, then (A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The compatibility coefficient between the two is calculated as follows:

c(A ⁱ ,B ^j ；A ^h ,B ^k )＝r _dis (A ⁱ ，B ^j ；A ^h ，B ^k )×r _overlapArea (A ⁱ ,B ^j ；A ^h ,B ^k )

wherein the relative distance r _dis (A ⁱ ，B ^j ；A ^h ，B ^k ) The calculation formula is as follows:

wherein Dis (A) ⁱ ，B ^j ) Representation A ⁱ And B ^j The distance between the two is calculated as follows:

wherein,and->Respectively represent MBR (A) ⁱ ) And MBR (B) ^j ) Centroid point coordinates of (c);

(A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The relative area overlap ratio r of (2) _overlapArea (A ⁱ ，B ^j ；A ^h ，B ^k ) The calculation formula of (2) is as follows:

wherein,M(A ⁱ ) Is vector (Deltax, deltay) translating polygon A ⁱ α is an adjustment coefficient;

the process of defining neighborhood objects for each pair of potential matching pairs includes: finding the AND a in the potential matching pair correspondence table by recursion _i All potential matching pairs of interest; suppose that detection:

(A ⁱ ：B ^j )＝(a _i ,a _i+1 ：b _j )，(A ⁱ⁺¹ ：B ^j+1 )＝(a _i+1 ,a _i+2 ：b _j+1 ) And (A) ⁱ⁺² ：B ^j+2 )＝(a _i+2 ,a _i+3 ：b _j+2 )，

Then it will: gA ⁱ ＝A ⁱ ∪A ⁱ⁺¹ ∪A ⁱ⁺² ＝{a _i ，a _i+1 ，a _i+2 ，a _i+3 }；

MBR extraction (gA) ⁱ ) And so on, obtain the point set V _A The method comprises the steps of carrying out a first treatment on the surface of the According to the point set V _A Establishing Delaunay triangle network with expression G _DT (A)＝(V _A ,E _A )，E _A For the collection of triangle net edges, MBR (A ⁱ ) Is defined by the geometric center point of (A) and MBR (A) ^h ) Is defined by E _A Either one of the edges is connected, A ⁱ And A ^h Are neighbors of each other, A ^h Is A ⁱ Is a neighbor of the element in the neighborhood set;

A ⁱ is defined as:

N(A ⁱ )＝{A ^h |(A ⁱ ,A ^h )∈E _A }

wherein N (A) ⁱ ) Represented by A ⁱ Is defined by:

N(A ⁱ )＝N(A ⁱ⁺¹ )＝N(A ⁱ⁺² )＝N(gA ⁱ )；

B ^j is a neighborhood of (a)The definition is as follows:

wherein N (B) ^j ) Denoted as B ^j Neighborhood set of B ^j ∈C(A ⁱ ) Denoted as B ^j And A ⁱ Is a potential matching pair and,representation B ^j And B ^k Not containing the same element, B ^k ∈C(A ^h ) Denoted as B ^k And A ^h Is a potential matching pair, (A) ⁱ ，A ^h )∈E _A Representation A ⁱ And A ^h Is a neighbor;

updating the initial probability matrix P according to the compatible coefficient iteration until the probability difference between the two iterations is smaller than a preset threshold value, stopping the iteration, and obtaining a final matching result,

in updating the probability matrix for each iteration, (A) ⁱ ：B ^j ) All the support is obtained from the neighborhood of the support function, and the calculation formula of the support function is as follows:

finally, updating the probability matrix according to the calculated support, wherein the calculation formula is as follows,

hold after each iterative update

When min (p) ^(t+1) (A ⁱ ：B ^j )-p ^(t) (A ⁱ ：B ^j ) And) 0.001, the iteration is terminated; for a pair ofIn A ⁱ ，B ^j ∈C(A ⁱ ) If p (A) ⁱ ：B ^j ) =max, then (a ⁱ ：B ^j ) And the final matching result.

2. The method of matching building surface elements based on environmental information of claim 1, wherein calculating an initial probability between each pair of potential matching pairs, constructing an initial probability matrix P based on the initial probabilities comprises,

for each pair of potential matching pairs (a ₁ ,a ₂ ,..,a _i ：b ₁ ,b ₂ ,..,b _j ) Will { a } ₁ ,a ₂ ,..,a _i Sum { b } ₁ ,b ₂ ,..,b _j Viewed as an aggregated polygon, denoted as a respectively ⁱ And B ^j Then B is ^j Is A ⁱ Is defined as B ^j ∈C(A ⁱ )，C(A ⁱ ) Represented by A ⁱ A set of potentially matching objects; representing all potential matching relationships as M _i：j ＝{(A ⁱ ：B ^j ) I=1, 2, p. j=1, 2,..q } after q } an initial probability matrix P is established.

3. The method for matching building surface elements based on environmental information according to claim 2, wherein,

the calculation formula of the initial probability matrix P is as follows:

wherein M (A) ⁱ ) Is vector (Deltax, deltay) translating polygon A ⁱ (Δx, Δy) is the MBR (A ⁱ ) And MBR (B) ^j ) Deviation between centroid coordinates;

the MBR centroid coordinate calculation formula is as follows:

x ₀ ＝(x _max -x _min )/2,

y ₀ ＝(y _max -y _min )/2，

(x _min ,y _min ) And (x) _max ,y _max ) The minimum and maximum coordinates of the MBR, respectively.

4. A building face element matching system based on environmental information, comprising:

the data acquisition module is used for acquiring potential matching pairs from the two face element sets to be matched;

the calculation module is used for calculating the initial probability between each pair of potential matching pairs, and constructing an initial probability matrix based on the initial probability, wherein the error of the area overlapping rate between the homonymous surface elements is required to be balanced through the adjustment coefficient;

the matching module is used for defining a neighborhood object of each pair of potential matching pairs and calculating a compatibility coefficient between each pair of potential matching pairs and the neighborhood matching pairs; the method is also used for iteratively updating the initial probability matrix according to the compatibility coefficient until the probability difference between two iterations is smaller than a preset threshold value, and the iteration is stopped to obtain a final matching result;

wherein, the calculation formula of the adjustment coefficient is:

wherein,and->Respectively represent A ⁱ And B ^j The sum of all the surface element contour points;

the process of calculating compatibility coefficients between each pair of potential matching pairs and its neighborhood matching pairs includes: defining a compatibility coefficient through the distance similarity and the area overlapping rate, and eliminating ambiguity between correct matching pairs in each iteration based on the compatibility coefficient;

design (A) ⁱ ：B ^j ) And (A) ^h ：B ^k ) Is an adjacent potential matching pair, then (A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The compatibility coefficient between the two is calculated as follows:

c(A ⁱ ,B ^j ；A ^h ,B ^k )＝r _dis (A ⁱ ,B ^j ；A ^h ,B ^k )×roverlapArea(A ⁱ ,B ^j ；A ^h ,B ^k )

wherein the relative distance r _dis (A ⁱ ，B ^j ；A ^h ，B ^k ) The calculation formula is as follows:

wherein Dis (A) ⁱ ，B ^j ) Representation A ⁱ And B ^j The distance between the two is calculated as follows:

wherein,and->Respectively represent MBR (A) ⁱ ) And MBR (B) ^j ) Centroid point coordinates of (c);

(A ⁱ ：B ^j ) And (A) ^h ：B ^k ) The relative area overlap ratio r of (2) _overlapArea (A ⁱ ,B ^j ；A ^h ,B ^k ) The calculation formula of (2) is as follows:

wherein,M(A ⁱ ) Is vector (Deltax, deltay) translating polygon A ⁱ α is an adjustment coefficient;

the process of defining neighborhood objects for each pair of potential matching pairs includes: finding the AND a in the potential matching pair correspondence table by recursion _i All potential matching pairs of interest; suppose that detection:

(A ⁱ ：B ^j )＝(a _i ,a _i+1 ：b _j )，(A ⁱ⁺¹ B ^j+1 )＝(a _i+1 ,a _i+2 ：b _j+1 ) And (A) ⁱ⁺² ：B ^j+2 )＝(a _i+2 ,a _i+3 ：b _j+2 )，

Then it will: gA ⁱ ＝A ⁱ ∪A ⁱ⁺¹ ∪A ⁱ⁺² ＝{a _i ,a _i+1 ，a _i+2 ,a _i+3 }；

MBR extraction (gA) ⁱ ) And so on, obtain the point set V _A The method comprises the steps of carrying out a first treatment on the surface of the According to the point set V _A Establishing Delaunay triangle network with expression G _DT (A)＝(V _A ,E _A )，E _A For the collection of triangle net edges, MBR (A ⁱ ) Is defined by the geometric center point of (A) and MBR (A) ^h ) Is defined by E _A Either one of the edges is connected, A ⁱ And A ^h Are neighbors of each other, A ^h Is A ⁱ Is a neighbor of the element in the neighborhood set;

A ⁱ is defined as:

N(A ⁱ )＝{A ^h |(A ⁱ ，A ^h )∈E _A }

wherein N (A) ⁱ ) Represented by A ⁱ Is defined by:

N(A ⁱ )＝N(A ⁱ⁺¹ )＝N(A ⁱ⁺² )＝N(gA ⁱ )；

B ^j is defined as:

wherein N (B) ^j ) Represented asB ^j Neighborhood set of B ^j ∈C(A ⁱ ) Denoted as B ^j And A ⁱ Is a potential matching pair and,representation B ^j And B ^k Not containing the same element, B ^k ∈C(A ^h ) Denoted as B ^k And A ^h Is a potential matching pair, (A) ⁱ ，A ^h )∈E _A Representation A ⁱ And A ^h Is a neighbor;

updating the initial probability matrix P according to the compatible coefficient iteration until the probability difference between the two iterations is smaller than a preset threshold value, stopping the iteration, and obtaining a final matching result,

in updating the probability matrix for each iteration, (A) ⁱ ：B ^j ) All the support is obtained from the neighborhood of the support function, and the calculation formula of the support function is as follows:

finally, updating the probability matrix according to the calculated support, wherein the calculation formula is as follows,

hold after each iterative update

When min (p) ^(t+1) (A ⁱ ：B ^j )-p ^(t) (A ⁱ ：B ^j ) And) 0.001, the iteration is terminated; for A ⁱ ，B ^j ∈C(A ⁱ ) If p (A) ⁱ ：B ^j ) =max, then (a ⁱ ：B ^j ) And the final matching result.