CN113112595B

CN113112595B - Automatic construction method of point-line-surface mixed three-dimensional road network of three-dimensional model of real scene of complex

Info

Publication number: CN113112595B
Application number: CN202110469328.8A
Authority: CN
Inventors: 谢潇; 沈婕; 赵露菲; 薛冰; 伍庭晨; 侯盈旭
Original assignee: Nanjing Normal University; Institute of Applied Ecology of CAS
Current assignee: Nanjing Normal University; Institute of Applied Ecology of CAS
Priority date: 2021-04-28
Filing date: 2021-04-28
Publication date: 2021-12-10
Anticipated expiration: 2041-04-28
Also published as: CN113112595A

Abstract

The invention provides a point-line-surface mixed three-dimensional road network automatic construction method of a three-dimensional model of a real scene of a complex. The method comprises the steps of extracting a road network base course, constructing a mapping relation from a three-dimensional model of a target building to the road network base course, and obtaining information data of indoor elements; the information data of the three-dimensional model of the target building are subjected to completion processing to obtain a base scene after the completion processing; constructing a normalized function from the three-dimensional model to the road network, and regularly extracting point, line and surface elements; and combining the point, line and surface elements to form a point, line and surface mixed three-dimensional road network model, and constructing a road network navigation demand data storage structure based on the point, line and surface structures. In this way, the generation efficiency of the navigation network in the complex comprehensive body is improved, the generation cost of the navigation network is reduced, the three-dimensional navigation network in a large-scale indoor space is rapidly and automatically extracted, data support is provided for realizing real-time navigation of each main body in the complex comprehensive body, and further development and application of indoor position service are promoted.

Description

Automatic construction method of point-line-surface mixed three-dimensional road network of three-dimensional model of real scene of complex

Technical Field

The embodiment of the invention relates to the field of indoor navigation, in particular to a point-line-surface mixed three-dimensional road network automatic construction method of a three-dimensional model of a complex real scene.

Background

Aiming at the road network model of a building complex including a large-scale business body and a large-scale transportation hub (such as high-speed rail stations, airports, subway stations and the like), the large-scale business body generally has the volume of tens of thousands to hundreds of thousands, the large-scale transportation hub generally has the volume of hundreds of thousands to hundreds of thousands, the building has the characteristics of multiple covering elements, complex space, multiple functional areas and the like, and in the process of realizing the construction of the irregular road network model in the building, the research objects need to be defined, so the patent introduces the concept of the complex.

Complex sysplex refers to a collection of functional regions with spatial complexity, topological complexity, and semantic complexity. Spatial complexity refers to irregularities in spatial morphology, staggering, etc.; the topology complexity refers to the connectivity, randomness and the like of the traffic space; semantic complexity refers to the conditionality, functional complexity, etc. of the nexus relationship. The complex characteristics specified in the text have the characteristics of large quantity and many objects, and the complex characteristics refer to the large quantity, the many objects and the complex implementation of the objects, and the complex characteristics refer to the complex types, the complex topological space structures and the complex semantic relations of the objects.

Since there is no clear traffic regulation in the room, effective path planning cannot be performed, and thus a movable path needs to be defined in advance indoors to express a navigation network of indoor trafficable conditions. At present, navigation paths can be extracted from two-dimensional plane data or three-dimensional model data, and research finds that no matter two-dimensional plane data-based road network extraction or three-dimensional model-based road network extraction is performed, only a single method is usually considered in the process of constructing a road network, so that a vector road network or a regular/irregular grid road network model has adaptive limitation on expression of trafficable conditions such as complex space unit structures, forms and functions of a complex, the efficiency of routing from the model to the road network is low, and the time cost for manually constructing the road network is extremely high.

Disclosure of Invention

According to the embodiment of the invention, the automatic construction method of the point-line-surface mixed three-dimensional road network of the three-dimensional model of the real scene of the complex is provided. The method comprises the following steps:

extracting a road network base layer from a target building three-dimensional model, constructing a mapping relation from the target building three-dimensional model to the road network base layer, and acquiring information data of indoor elements from the target building three-dimensional model, wherein the information data comprises geometric information, topological information and semantic information;

perfecting the information data of the three-dimensional model of the target building, completing the perfected information data according to data requirements and structural requirements, and fusing the completed information data and the mapping relation between the three-dimensional model of the target building and a road network base layer to obtain a completed base layer scene;

building a normalized function from the three-dimensional model of the target building to a road network, building a road network midpoint, line and surface element data model function, defining abstract characteristics of points, lines and surfaces of a scene object, setting points, structural requirements of the lines and the surface elements, determining extraction methods of the points, the lines and the surface elements, and performing regularized extraction on the points, the lines and the surface elements respectively;

combining the point, line and surface elements to form a point, line and surface mixed three-dimensional road network model; and fusing the information data after the completion processing with point, line and plane structure information in the point, line and plane mixed three-dimensional road network model to construct a road network navigation demand data storage structure based on the point, line and plane structure.

Further, the constructing a mapping relation of the three-dimensional model of the target building to the road network base layer includes:

mapping the three-dimensional space information of the three-dimensional model of the target building from horizontal and vertical view angles to a road network base layer; the road network base layer is convenient for extracting two-dimensional projection surfaces of all objects of road network elements, which are landed on floor floors, and comprises horizontal projection surfaces of floors and vertical projection surfaces of floors.

Further, the obtaining of the information data of the indoor elements from the three-dimensional model of the target building includes:

extracting geometric information, semantic information and topological information of indoor elements from the three-dimensional model of the target building; screening and recombining the geometric information, semantic information and topological information of the indoor elements according to data requirements to obtain the processed geometric information, semantic information and topological information of the indoor elements;

the geometric information is used for describing the geometric position characteristics of the indoor space elements and comprises shape, size, boundary, coordinates and indoor position direction information of the indoor space elements;

the topological information is used for expressing the topological relation among the indoor space elements by using the abstract association among the indoor space element symbolic entities;

and the semantic information is used for describing object attributes and functional attributes of the indoor space elements.

Further, the fusing the geometric information, the semantic information and the topological information which are subjected to the completion processing with the mapping relation from the three-dimensional model of the target building to the road network base layer includes:

and selecting a common point associated with the basic scene information of the road network base and the base scene set information after the completion processing as a target point to calibrate the three-dimensional model information and the two-dimensional plane.

Further, the normalization function from the three-dimensional model of the target building to the road network is as follows:

N_m＝F{P_dm,L_dm,S_dm|R(P_dm,L_dm,S_dm)}

wherein N is_mIs a point-line-surface mixed three-dimensional road network model, P_dmData model extracted for regularization of point elements, L_dmData model regularly extracted for line elements, S_dmData model, R (P), extracted for regularization of surface elements_dm,L_dm,S_dm) The relative topological relation of the point, line and surface road network elements is shown.

Further, constructing a road network point element data model function, defining the point abstract characteristics of a scene object, setting the structure requirement of a point element, determining an extraction method of the point element, and performing regularized extraction on the point element, wherein the extraction method comprises the following steps:

firstly, constructing a point element data model function in a road network:

P_dm＝f_p(P-Rule₁,P-Rule₂,…,P-Rule_n|Me_p)

wherein, P_dmFor the data model of regular extraction of point elements, P-Rule represents some kind of abstract expression of pointsRule, n is the number of point element classes, Me_pA point element regularized extraction method;

secondly, defining point abstraction features of the scene object, including:

forming a screening rule of point elements in a scene by constructing the characteristics needing to be abstracted into the point elements in a scene space; the screening rule of the point elements comprises the following steps:

a point object abstracted by a solid or virtual object with a passage condition in a space is taken as a solid point; taking a point object abstracted from a closed space which needs to express the relative position in the space and does not influence the path as a busy point; a point object abstracted by an entity connecting an indoor space to an outdoor space is taken as a connected point; taking an intersection point formed by connecting the entity point or the communication point with the structural line as a virtual point;

again, the structural requirements of the setpoint element include:

constructing a data format of point element extraction, wherein a point field and a data type of a point element are set in the data format;

finally, the method for extracting the determined point elements comprises the following steps:

extracting the face object of the corresponding point element needing to be extracted in the road network base layer through semantic interpretation, and extracting the centroid point or the central point of the face object to obtain a point element extraction graph of an indoor scene; if the point elements needing to be extracted are entity points or connected points, extracting a central point; and if the point element to be extracted is the occupied point, extracting the centroid point.

Further, the method for constructing the line element data model function in the road network, defining the line abstract characteristics of the scene object, setting the structure requirements of the line elements, determining the extraction method of the line elements and performing the regularized extraction on the line elements comprises the following steps:

firstly, constructing a road network central line element data model function:

L_dm＝f_l(L-Rule₁,L-Rule₂,…,L-Rule_m|Me_l)

wherein L is_dmFor a data model with regular extraction of line elements, L-Rule represents a certain classRule of abstract expression of line of type, m being number of line element classes, Me_lA line element regularized extraction method;

secondly, defining line abstraction features of the scene object, including:

forming a screening rule of the line elements in the scene by constructing the characteristics which need to be abstracted into the line elements in the scene space; the scene space comprises two dimensions of the same floor and cross-floor; the screening rule of the line elements comprises:

line objects abstracted from the same-floor regular traffic area and the cross-floor traffic space are used as structural lines to represent the main trunk of the road network; line objects abstracted by the distance from the entity points to the structural lines are used as connecting lines to represent the connectivity of the graphic elements and the road network trunk; a line object abstracted by a connecting line of the entity point and the occupied point is used as a relation line to represent the accessibility of the graphic element, and the relation line is not a substantial passing path; taking a line object abstracted when a turning line is optimized at an intersection as a turning line; taking a line object abstracted by entities capable of influencing path finding and abstracting a line in the second-class scene as a barrier line;

the second class of scenes refers to scenes in which road networks are required to be constructed in irregular passing areas;

thirdly, the setting of the line element configuration requirements includes:

constructing a data format extracted by the line element, and setting a line field and a data type of the line element in the data format;

finally, the method for extracting the line elements comprises the following steps:

the structure lines comprise a horizontal space structure line and a vertical space structure line, and the extraction method of the horizontal space structure line adopts a middle axis transformation method to extract the central axis of the path of the passing area; the extraction method of the vertical space structure line is to extract the central axis of the three-dimensional space of an elevator shaft and an elevator room in the three-dimensional model of the target building; the extraction method of the connecting line is to connect the entity point with the structure line to obtain the connecting line; the extraction method of the relation line is to connect the position occupying point with the entity point to obtain the relation line; the method for extracting the corner line comprises the steps of measuring and calculating the distance between two central axes which are connected with each other and a boundary line of a corridor, and constructing a half arc tangent to the two central axes at the corner by taking the minimum distance as a radius to serve as the corner line.

Further, the method for constructing a road network surface element data model function, defining surface abstract characteristics of a scene object, setting structural requirements of surface elements, determining an extraction method of the surface elements and performing regularized extraction on the surface elements comprises the following steps:

firstly, constructing a road network middle surface element data model function:

S_dm＝f_s(S-Rule₁,S-Rule₂,…,S-Rule_z|Me_s)

wherein S is_dmFor the data model of regular extraction of the surface elements, S-Rule represents the Rule of surface abstract expression of a certain type, z is the number of surface element classes, Me_sThe method is a face element regularized extraction method;

secondly, defining the surface abstraction feature of the scene object, comprising:

forming a screening rule of the face elements in the scene by constructing the features which need to be abstracted into the face elements in the scene space; the screening rule of the face element comprises the following steps:

taking a surface object abstracted from entity elements capable of influencing path finding in the second type of scenes as an obstacle surface; a surface object obtained by abstracting a passable range in the second type scene is used as a passage surface; wherein, the barrier surface with the passing surface is used as the barrier surface with an inlet; taking the barrier surface without the passing surface as a pure barrier surface;

the second-class scene is a scene which contains a plurality of irregular passing areas and a plurality of obstacles;

thirdly, the setting of the structural requirements of the surface elements comprises the following steps:

constructing a data format extracted by the face element, and setting a face field and a data type of the face element in the data format;

finally, the method for extracting the determined surface element comprises the following steps:

in the second type of scene, the two-dimensional plane is expressed by the surface elements according to the screening rule of the surface elements, and the traffic surface in the surface elements is extracted.

Further, the combining the point, line and surface elements to form a point-line-surface hybrid three-dimensional road network model includes:

connecting the place-occupying point with the entity point to generate a relationship line; connecting the solid point with the structure line to generate a connecting line; constructing a path model of a combination of point elements and horizontal line elements, and supplementing node information;

and combining the path model formed by combining the point elements and the horizontal line elements with the surface elements and the vertical line elements to form a three-dimensional road network model formed by mixing points, lines and surfaces.

Further, the constructing of the data storage structure of the road network navigation requirement based on the point, line and plane structure includes:

adding a topological relation field between elements in a database, wherein the topological relation field comprises adjacent surfaces, adjacent lines and adjacent points.

It should be understood that the statements herein reciting aspects are not intended to limit the critical or essential features of any embodiment of the invention, nor are they intended to limit the scope of the invention. Other features of the present invention will become apparent from the following description.

The invention realizes the construction of the road network model of the complex by constructing a unified data format standard and fusing a plurality of road network extraction methods, thereby not only making up the adaptability limitation of the traditional single vector road network or regular/irregular grid road network model on the expression of the passable conditions such as the complex space unit structure, the form, the function and the like of the complex, but also improving the routing efficiency by fusing a plurality of methods, meeting the requirements better, improving the road network generation efficiency by automatically constructing, and reducing the time cost for manually constructing the road network.

Drawings

The above and other features, advantages and aspects of various embodiments of the present invention will become more apparent by referring to the following detailed description when taken in conjunction with the accompanying drawings. In the drawings, like or similar reference characters designate like or similar elements, and wherein:

FIG. 1 is a flow chart of the automatic construction method of a point-line-surface mixed three-dimensional road network of a three-dimensional model of a real scene of a complex;

FIG. 2 is a schematic diagram illustrating a method for automatically constructing a point-plane hybrid three-dimensional road network of a three-dimensional model of a real scene of a complex according to an embodiment of the present invention;

FIG. 3 shows a schematic diagram of a three-dimensional model of a target building, according to an embodiment of the invention;

FIG. 4 illustrates a point element extraction diagram for a class of scenes in accordance with an embodiment of the present invention;

FIG. 5 illustrates a central axis extraction diagram for a class of scenes in accordance with an embodiment of the present invention;

FIG. 6 illustrates a corridor intersection smooth path construction diagram according to an embodiment of the invention;

FIG. 7 shows a navigation path diagram in a vertical direction according to an embodiment of the invention;

FIG. 8 illustrates a face element extraction diagram for a class II scene, according to an embodiment of the invention;

FIG. 9 illustrates a schematic diagram of a dot-line-plane memory structure according to an embodiment of the invention;

FIG. 10 shows a road network extraction schematic diagram of point, line and surface blending according to an embodiment of the invention;

wherein 1 is a occupational site, 2 is a connecting line, 3 is a corner line, 4 is a structural line, 5 is a virtual point, 6 is a physical point, and 7 is a relationship line.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.

The research object of the invention is a road network model of a building complex including a large commercial body, a large transportation hub (such as high-speed rail stations, airports, subway stations and other places), and the like, wherein the large commercial body generally has the volume of tens of thousands to hundreds of thousands, the large transportation hub generally has the volume of hundreds of thousands to hundreds of thousands, the building has the characteristics of multiple covering elements, complex space, multiple functional areas and the like, and the research object needs to be defined in the process of realizing the construction of an irregular road network model in the building, so the invention introduces the concept of the complex.

The invention refers to the concept of a complex, which is a general term for functional areas with spatial complexity, topological complexity and semantic complexity. Spatial complexity refers to irregularities in spatial morphology, staggering, etc.; the topology complexity refers to the connectivity, randomness and the like of the traffic space; semantic complexity refers to the conditionality, functional complexity, etc. of the nexus relationship. The complex characteristics specified in the text have the characteristics of large quantity and many objects, and the complex characteristics refer to the large quantity, the many objects and the complex implementation of the objects, and the complex characteristics refer to the complex types, the complex topological space structures and the complex semantic relations of the objects.

Since there is no clear traffic regulation in the room, effective path planning cannot be performed, and thus a movable path needs to be defined in advance indoors to express a navigation network of indoor trafficable conditions. The road network formats extracted by the existing method mainly comprise 3 types: grid model, point-line-grid hybrid model. The road network model has different data expression forms and certain similarity, and has no factor construction criterion for extracting the road network of the complex to form a set of complete and standard modeling flow.

At present, the research for extracting the navigation path from the two-dimensional plane data is relatively mature, and there are 3 most common methods: a grid method, a visual graph method, and an abstract method. The building of an indoor navigation network by using three-dimensional model data requires the creation of a model for accurately expressing indoor spatial features and semantic information, and two models are widely applied in a plurality of three-dimensional models at present: the method comprises the following steps of firstly, providing an Industry Foundation Class (IFC) data model standard by the international collaborative union; the other is city geographic markup language (city geographic markup language, city GML) proposed by the geographic information alliance. Both have rich geometric and semantic information, and the latter is closer to Geographic Information System (GIS) application than the former, so the patent uses BIM data in the format of CityGML as a prototype to extract the navigation path.

According to the invention, a road network model of a complex is constructed by integrating multiple road network extraction methods by constructing a unified data format standard, so that the defects caused by a single method are overcome, the routing efficiency is improved by integrating multiple methods, the requirement is better met, and the road network generation efficiency is improved by automatically constructing.

As shown in fig. 1 and 2, the method includes:

s101, extracting a road network base layer from a target building three-dimensional model, constructing a mapping relation from the target building three-dimensional model to the road network base layer, and acquiring information data of indoor elements from the target building three-dimensional model.

The target building is a complex and refers to a general term of functional areas with space complexity, topological complexity and semantic complexity. Wherein, the space complexity refers to the irregularity, the staggering, etc. of the space morphology; the topology complexity refers to the connectivity, randomness and the like of the traffic space; semantic complexity refers to the conditionality, functional complexity, etc. of the nexus relationship. The complex characteristics specified in the text have the characteristics of large quantity and many objects, and the complex characteristics refer to the large quantity, the many objects and the complex implementation of the objects, and the complex characteristics refer to the complex types, the complex topological space structures and the complex semantic relations of the objects. The three-dimensional model of the target Building is a three-dimensional model meeting the CityGML LoD4 or IFC standard, and includes a Building Information Model (BIM), but is not limited to the BIM model. Fig. 3 shows a schematic diagram of a three-dimensional model of the east station in hangzhou.

The road network base layer is a two-dimensional projection plane convenient for extracting all objects of road network elements falling on a floor, wherein inclined objects such as escalators and pillars are extracted from a bottom plate to contact with a bottom surface, and a complex model is divided into a horizontal projection plane for dividing floors and a vertical projection plane for crossing floors from horizontal and vertical visual angles.

As an embodiment of the present invention, extracting a road network base layer from a three-dimensional model of a target building, and constructing a mapping relationship from the three-dimensional model of the target building to the road network base layer, includes: mapping the three-dimensional space information of the three-dimensional model of the target building from horizontal and vertical view angles to a road network base layer; that is, the three-dimensional spatial information and the road network base layer data requirements are mapped one by one, and as shown in table 1, the three-dimensional spatial information and the road network base layer data requirements are data mapping tables:

TABLE 1

As an embodiment of the present invention, acquiring information data of indoor elements from the three-dimensional model of the target building includes: firstly, information data of elements in a target building are extracted from a three-dimensional model of the target building based on Open Database Connectivity (ODBC) in Revit software, wherein the information data comprise geometric information, topological information and semantic information. The geometric information is used for describing the shape, size, boundary, coordinate and direction information among indoor positions of the indoor space elements, and can realize the expression of the characteristics of the indoor space geometric positions; the topological information refers to the fact that the topological relation among the indoor space elements is expressed by using abstract association (adjacency, communication, inclusion, overlapping and the like) among the indoor space element symbolic entities; the semantic information refers to the object attribute, the function attribute and the like of the indoor space element. And then, screening and recombining all the extracted information according to the data requirements to obtain the processed data information. The data requirement is known requirement, namely required data when extracting points, lines and surfaces, such as point coordinates, point types, line lengths, line slopes, point-line-surface attributes and the like.

S102, completing the information data of the three-dimensional model of the target building, completing the completed information data according to data requirements and structural requirements, and fusing the completed information data and the mapping relation from the three-dimensional model of the target building to a road network base layer to obtain a completed base layer scene.

As an embodiment of the present invention, the refining the information data of the three-dimensional model of the target building includes: and unifying the geometric information, the topological information and the semantic information of the three-dimensional model of the target building by fields, such as supplementing the hollow fields.

As an embodiment of the present invention, the data requirement and the structure requirement are known requirements, that is, data and structures required in point, line and plane extraction, such as data items in a data structure table of elements; and ensuring that necessary information such as point coordinates, point types, point belongings, line lengths, line slopes, adjacent line intervals, surface boundary composition, point-line-surface attributes and the like of the elements are complete according to data and structure requirements.

As an embodiment of the present invention, the completion processing includes integrating and unifying the completed information data, that is, supplementing and perfecting the fields required by extracting information from the peer-to-peer plane according to the data requirement and the structure requirement; for example, the length and the slope of the line segment are increased, so that the extraction of elements is facilitated.

As an embodiment of the present invention, the road network base layer does not have related geometric, topological and semantic information, and the extracted data needs to be fused according to the mapping relationship from the constructed complex model to the road network base layer, and the three-dimensional model information and the two-dimensional plane are calibrated through several floor corner points or target points, so that the extracted two-dimensional road network base layer plane has specific geometric, semantic and topological information. For example, the basic scene information of the road network base layer and the final base layer scene are associated and fused through the geometric information of some common points, the fusion purpose is realized through calibration, and the calibration can select the common points of the basic scene information and the final base layer scene, namely the common points; the common point may be any destination point, such as a floor corner point. The searching of the common point needs to consider that the basic scene information and the final basic scene have the point data and the searching is convenient. The number of common points can be arbitrarily selected according to the needs.

Through the process of S102, the data format of the basic scene information is unified in the road network basic scene, so that the extracted two-dimensional road network basic plane has definite geometric, semantic and topological information. The basic scene is the basic scene information corresponding to the road network basic scene, and the information needs to be further improved into the final basic scene.

S103, constructing a normalized function from the three-dimensional model of the target building to the road network, defining abstract characteristics of points, lines and surfaces of the scene object and structural requirements of the set points, lines and surfaces, determining extraction methods of the points, lines and surfaces, and regularly extracting the points, lines and surfaces.

In S103, first, a normalization function from the three-dimensional model of the target building to the road network needs to be constructed, which specifically includes:

N_m＝F{P_dm,L_dm,S_dm|R(P_dm,L_dm,S_dm)}

After the normalized function is constructed, point, line and surface element data model functions of the road network need to be constructed respectively, and point elements, line elements and surface elements need to be subjected to normalized extraction.

As an embodiment of the invention, a point element data model function in a road network is constructed, and point elements are subjected to normalized extraction, wherein the method comprises the following steps: firstly, constructing a point element data model function in a road network:

P_dm＝f_p(P-Rule₁,P-Rule₂,…,P-Rule_n|Me_p)

wherein, P_dmFor the data model extracted by the point element regularization, P-Rule represents a certain type of Rule expressed by a point abstraction, n is the number of point element classes, the number of concrete rules can be adjusted according to the actual scene, for example, if the point element types existing in the road network comprise entity points, occupation points, connection points and virtual points, n is 4, that is, P-Rule₁Rules expressed for entity point abstraction, P-Rule₂For rules expressed abstractly in terms of occupation sites, P-Rule₃Rules expressed for connectivity point abstraction, P-Rule₄Rules for abstract representation of virtual points; me_pThe method is a point element regularization extraction method.

Secondly, defining the point abstraction feature of the scene object explicitly, including: and forming a screening rule of the point elements in the scene by constructing the characteristics which need to be abstracted into the point elements in the scene space.

In this embodiment, the scenes are divided into a first-class scene and a second-class scene. In the present embodiment, the scene of one category refers to a scene with a corridor or a regular traffic space; taking the east Hangzhou station as an example, a commercial interlayer, a ground floor, a commercial area and the like of the 2 th building on the starting floor are taken as a scene. The second kind of scenes refers to rooms in which road networks are required to be constructed in a refined way in irregular passing areas, namely, scenes which are large in number of obstacles and irregular in passing areas and contain a large number of passing areas. Similarly, taking the east Hangzhou station as an example, the departure floor 2, the departure floor 1, the arrival floor 2, the arrival floor 1, the subway floor and the like are two types of scenes, and the specific classification is performed based on the type of the actual complex.

In this embodiment, the point elements mainly include a Room (Room), a Door (Door), an Entrance (Entrance), an Escalator (Escalator) Entrance, an Elevator (Elevator) Entrance, and a stairway (Staircase) Entrance.

In this embodiment, the filtering rule of the point element includes: a point object abstracted by a solid or virtual object with a passage condition in a space is taken as a solid point; the space can pass through under the passing condition; taking a point object abstracted from a closed space which needs to express the relative position in the space and does not influence the path as a busy point; a point object abstracted by an entity connecting an indoor space to an outdoor space is taken as a connected point; taking an intersection point formed by connecting the entity point or the communication point with the structural line as a virtual point;

the type and definition table of the hybrid stereo road network topological relation point in this embodiment is shown in table 2:

TABLE 2

Again, the structural requirements of the setpoint element include: constructing a data format of point element extraction, and explaining the data format; the data format of the dot element is set from the dot field and the data type of the dot element. As shown in table 3, a data structure table of point elements;

TABLE 3

The data format of the point elements mainly comprises a point serial number (P-ID), point coordinates (X, Y, Z), a point Type (Type), a belonging room serial number (R-ID), a belonging floor serial number (F-ID), a belonging floor serial number (B-ID), remarks and the like. The remarks generally set the contents of name, time limit, etc.

Finally, the method for extracting the determined point elements comprises the following steps: and extracting the face object of the corresponding point element to be extracted in the road network base layer through semantic interpretation, and extracting the centroid point or the central point of the point element to be extracted based on the point-face dual principle of a two-dimensional space dual theory to obtain a point element extraction graph of the indoor scene. If the point elements needing to be extracted are entity points or connected points, extracting a central point; and if the point element to be extracted is the occupied point, extracting the centroid point. The centroid is extracted for the room elements, and the centers are extracted for the door, the entrance, the escalator entrance, the elevator entrance and the stair entrance and are replaced by corresponding symbols; the central point coordinate is obtained by taking the key element as a straight line and obtaining the central point; the room elements are surface areas and require area centroids, and for a closed area D, the density formula is F (x, y), and the centroid formula is as follows:

wherein x and y are x and y functions on the closed region D;

respectively as the centroid point of the closed region D; f is a density function of the closed region D; σ is the area of the closed region D;

double integration over the closed area D for the function x;

double integration over the closed area D for the function y;

is the double integral of the function F (x, y) over the closed region D.

As shown in FIG. 4, the point labeled D is an abbreviation of Door, indicating a Door; the point marked C is an abbreviation of Escalator, representing an Escalator; the point marked R is an abbreviation of Room, representing a Room; the point marked S is an abbreviation of Staircase, representing a Staircase; hexagonal dots are used to represent the entrances and exits.

As an embodiment of the invention, a line element data model function in a road network is constructed, and the normalized extraction of line elements is carried out, which comprises the following steps: the method comprises the following steps of constructing a line element data model function in a road network, defining line abstract characteristics of a scene object, setting structural requirements of line elements, determining an extraction method of the line elements, and performing regularized extraction on the line elements, wherein the extraction method comprises the following steps: firstly, constructing a road network central line element data model function:

L_dm＝f_l(L-Rule₁,L-Rule₂,…,L-Rule_m|Me_l)

wherein L is_dmFor the data model extracted for the line element regularization, L-Rule represents a Rule expressed in an abstract way for a certain type of line, m is the number of line element classes, and the number of specific rules can be adjusted according to the actual scene, for example, if the line element types existing in the road network include a structure line, a connecting line, a relation line, a corner line and an obstacle line, m is 5, that is, L-Rule₁Rules expressed for structure line abstraction, L-Rule₂Rules expressed for connecting line abstraction, L-Rule₃For rules of abstract representation of relationship lines, L-Rule₄Rules expressed for corner line abstraction, L-Rule₅Rules that are abstractly expressed for barrier lines; me_lThe method is a line element regularized extraction method.

Secondly, the line abstraction feature that clearly defines the scene object includes: forming a screening rule of the line elements in the scene by constructing the characteristics which need to be abstracted into the line elements in the scene space; the scene space comprises two dimensions of same floor and cross floor. In this embodiment, the same-floor horizontal space passage line mainly includes passage areas such as a Corridor and a Corridor (Corridor); the floor-crossing space traffic line elements mainly comprise traffic elements such as vertical stairs (V-Staircase), vertical elevators (V-Elevator), escalators (NV-escape) and the like.

The screening rule of the line elements comprises: line objects abstracted from the same-floor regular traffic area and the cross-floor traffic space are used as structural lines to represent the main trunk of the road network; line objects abstracted by the distance from the entity points to the structural lines are used as connecting lines to represent the connectivity of the graphic elements and the road network trunk; a line object abstracted by a connecting line of the entity point and the occupied point is used as a relation line to represent the accessibility of the graphic element, and the relation line is not a substantial passing path; taking a line object abstracted when a turning line is optimized at an intersection as a turning line; and line objects which can influence the routing and can be abstracted into entities of lines in the two types of scenes are abstracted into the line objects as barrier lines.

The type and definition table of the hybrid stereo road network topological relation line in this embodiment is shown in table 4:

TABLE 4

The barrier line is not a passing line element, but is an internal element which is convenient for expressing the line element, so that the barrier line is not extracted.

Thirdly, setting the structural requirements of the line elements, including constructing the data format extracted by the line elements, and explaining the data format; the data format of the line element is set from the line field and the data type of the line element.

In the present embodiment, a data structure table of line elements is shown in table 5;

TABLE 5

The data format of the line elements mainly comprises a line serial number (L-ID), end point coordinates (X, Y, Z), a Distance (Distance), a Type (Type), a belonging room serial number (R-ID), a belonging floor serial number (F-ID), a belonging building serial number (B-ID), remarks (name, traffic state and the like) and the like; the passing state comprises personnel passing requirements, time limits and the like. The personnel can be divided into: one class of people (police, management, etc. who have absolute right of way), two class of people (visitors, etc. who have relative rights), and three class of people (other staff who have relative rights).

Finally, the method for extracting the line elements is determined, as shown in fig. 5 to 7, and includes: the structure line 4 comprises a horizontal space structure line and a vertical space structure line, and the extraction method of the horizontal space structure line adopts a Medial Axis Transformation (MAT) to extract the central Axis of the path of the traffic area; the extraction method of the vertical space structure line is to extract the central axis of the three-dimensional space of an elevator shaft and an elevator room in the three-dimensional model of the target building; the extraction method of the connecting line 2 is to connect the entity point 6 with the structure line 4 to obtain the connecting line 2; the extraction method of the relation line 7 is to connect the occupied site 1 with the entity site 6 to obtain the relation line 7; the extraction method of the corner line 3 is to measure and calculate the distance between two central axes which are connected with each other and a boundary line of a corridor, and a half arc tangent to the two central axes is constructed at the corner by taking the minimum distance as a radius to serve as the corner line 3.

As an embodiment of the present invention, it is preferable that the corridor element boundary line be extracted from the floor two-dimensional plane, and a plan view including only the corridor passage space be generated. Changing the two-dimensional plane graph of the floor into a black-and-white binary graph, setting the traffic spaces such as corridors, corridors and the like as a foreground color white, filling the physical spaces such as rooms, stairs, elevators and the like as a background color black, binarizing the generated gray level graph, and extracting the boundary line of the corridor element.

As an embodiment of the present invention, it is preferable that the central axis of the main body corridor, i.e., the structural line, is extracted according to the feature requirement. Firstly, setting a proper length threshold value to filter short line segments at the tail end of the corridor; in the rest long line segments, the corresponding parallel line segments can be matched, for some parallel line pairs with equal or approximate lengths and closer distances, the slopes of the parallel line pairs should be equal, but due to drawing errors and certain errors between floating point numbers and fractions in a computer, an appropriate slope difference threshold value needs to be set; in an indoor environment, the corridor paths may have a parallel planning mode, so that an appropriate distance threshold value needs to be set, each parallel line pair is matched based on the matching conditions, and a central axis of the parallel line pair is generated. In the unmatched line segments, there are collinear line segment groups, in such a case, all the collinear line segment groups need to be screened first, and since the slopes and intercepts of the collinear line segment groups are equal, the slope difference and intercept difference threshold is set to 0.01, and all the collinear line segment groups are screened using this condition. And one side with less collinear line segments is taken as a main part to construct parallel lines which have the same length and have half of the intercept with the long side as the collinear line group, and the central axis of the collinear line group is generated.

As an embodiment of the present invention, it is preferable that corridor intersections (four-way intersections, three-way intersections, T-shaped intersections, staggered intersections, and the like) are individually processed to generate corner lines, so that the route more conforms to the habit of pedestrians. For the treatment of four-branch intersections and three-branch intersections, screening out an intersection end point set corresponding to each intersection end point by using a proper distance threshold, calculating a point set central point, and then connecting each point with the central point of the point set to which the point belongs; one central axis of the T-shaped intersection is extended to the other central axis from the end point; the two end points can be directly connected at the staggered intersection; the path after intersection connection is hard, and the walking habit of pedestrians is not well met at the turning position, the distance between the two connected central axes and the boundary line of the corridor is measured and calculated, and the minimum distance is used as the radius to construct a half arc at the intersection and is tangent to the two central axes.

As an embodiment of the present invention, it is preferable that the elements of the vertical connected space lines are represented by a chain diagram, and the chain diagram is a mixed diagram mode expression having a directed diagram and an undirected diagram. The undirected graph mainly refers to an elevator and a stair, and the directed graph refers to an escalator with upper and lower limits.

As an embodiment of the present invention, a data model function of a surface element in a road network is constructed, and a normalized extraction is performed on the surface element, including: the method comprises the following steps of constructing a road network surface element data model function, defining surface abstract characteristics of a scene object, setting the structural requirements of surface elements, determining an extraction method of the surface elements, and performing regularized extraction on the surface elements, and comprises the following steps: firstly, constructing a road network middle surface element data model function:

S_dm＝f_s(S-Rule₁,S-Rule₂,…,S-Rule_z|Me_s)

wherein S is_dmFor the data model extracted by the surface element regularization, S-Rule represents a Rule of a certain type of surface abstract expression, z is the number of surface element classes, the number of concrete rules can be adjusted according to the actual scene, for example, if the surface element types existing in the road network include an obstacle surface and a traffic surface, z is 2, that is, S-Rule₁Rules expressed for abstract of obstacle surface, S-Rule₂Rules expressed for the traffic surface abstraction; me_sThe method is a face element regularization extraction method.

Secondly, defining the surface abstraction feature of the scene object explicitly, including: forming a screening rule of the face elements in the scene by constructing the features which need to be abstracted into the face elements in the scene space; the surface elements mainly comprise rooms with relatively free activity spaces, such as a ticket selling hall, a rest area, a visiting area and the like.

The screening rule of the face element comprises the following steps: taking a surface object abstracted from entity elements capable of influencing path finding in the second type of scenes as an obstacle surface; a surface object obtained by abstracting a passable range in the second type scene is used as a passage surface; wherein, the barrier surface with the passing surface is used as the barrier surface with an inlet; taking the barrier surface without the passing surface as a pure barrier surface; wherein, the simple obstacle surface refers to a surface area which can not pass through at all; an entry-containing barrier surface is one that includes a traffic surface within the barrier surface and is accessible through an entry, such as a hospital consultation station.

The type and definition table of the topology relationship surface of the hybrid stereo road network in this embodiment are shown in table 6:

TABLE 6

The obstacle surface is a non-traffic surface element, and is convenient for processing internal elements expressed by the surface element, so that the obstacle surface is not extracted.

Thirdly, the setting of the structural requirements of the surface elements comprises the following steps: constructing a data format of the face element extraction, and explaining the data format; the data format of the surface element is set from the surface field and the data type of the surface element.

In the present embodiment, a data structure table of the face elements is shown in table 7;

TABLE 7

The data format of the surface element mainly comprises a surface serial number (S-ID), node coordinates (X, Y, Z), a Type (Type), a belonging room serial number (R-ID), a belonging floor serial number (F-ID), a belonging floor serial number (B-ID), remarks (names and the like) and the like. As shown in FIG. 8, a storage structure facing object S1 may be represented by Table 8;

TABLE 8

Wherein, S1 is any surface area in the second class of scenes; p_(1,2,...,16)Boundary point information extracted for the face region S1; a1 is an obstacle surface in the surface region S1; a1 is a barrier line in the face region S1.

Finally, the method for extracting the face element is determined, as shown in fig. 8, and includes:

in the second type scene, the two-dimensional plane is expressed by the surface elements according to the screening rule of the surface elements and the data structure table of the surface elements, and the traffic surface in the surface elements is extracted.

The surface element is a representation of a passable area in a second type scene; during path navigation, fine processing needs to be performed on the path navigation, a subdivision function set is constructed, and a gridding processing method based on the passing area of the surface is provided. The subdivision function set mainly provides a regular grid method subdivided by a quadrangle and an irregular grid method subdivided by a Delaunay triangulation network.

(1) And (3) regular grid meshing:

1) and extracting data of the indoor passable area through the indoor map data of the second-class scene to determine the area range.

2) And (4) subdividing the space enclosed by the obstacles. According to the side length L of the quadrangle set by the user, carrying out two-dimensional space gridding to generate a grid two-dimensional array, wherein the size calculation formula of the two-dimensional array is as follows:

wherein (x)_min,y_min) As coordinates of origin, (x)_max,y_max) Is a terminal coordinate; l is the length of grid side, and r and c are the number of rows and columns of the two-dimensional grid array.

3) Optimizing the grid and selecting path points. Deleting grids which pass through obstacles or exceed the area; and calculating the coordinates of the corner points of each grid and the adjacent grids in turn from the starting point.

(2) Irregular gridding:

1) and extracting point, line and plane information according to an extraction rule based on the two-dimensional plane graph of the second-class scene.

2) And (4) subdividing the space enclosed by the obstacles, and selecting path points. Creating a TIN irregular triangular network on the extracted trafficable area scatter set; when the subdivision area is a convex polygon, perfect subdivision can be performed to obtain a spatial area subdivision map of an indoor passable area, and when the subdivision area is a concave polygon, redundant subdivision can occur and needs to be deleted to obtain the indoor spatial area subdivision map.

3) And selecting the path points. Quantitatively representing the subdivided Delaunay triangular space region as a path point, and representing by using the gravity center of a feature point of a subdivided small region (triangle); the gravity center of the triangle is the intersection point of the three side middle lines of the triangle, and if the coordinates of the three vertexes of the triangle are respectively A (x1, y1), B (x2, y2) and C (x3, y3), the coordinates of the gravity center are as follows:

4) and (4) establishing and optimizing a path. Connecting adjacent triangles through the generated path points to represent paths which can pass through, and deleting paths of connecting lines passing through obstacles or exceeding a surface area; the link distance between adjacent path points is shortest, but when a path is constructed, the link distance between some non-adjacent path points may be approximately equal to the sum of the link distances between adjacent path points, so as to generate a large number of paths that can be approximately replaced, which wastes storage space, where a triangle judgment method is adopted, and path deletion is performed by setting a parameter δ as a reference value, where A, B, C is assumed to be a path point, and a, b, and c represent paths between path points, respectively, and the judgment formula is as follows:

if the condition in the formula is satisfied, the path c may be replaced, i.e. the path c is deleted, and the selected values of the reserved paths a, b, and δ are determined according to the actual application requirements.

S104, combining the point, line and surface elements to form a point, line and surface mixed three-dimensional road network model; and fusing the information data after the completion processing with point, line and plane structure information in the point, line and plane mixed three-dimensional road network model to construct a road network navigation demand data storage structure based on the point, line and plane structure.

The point-line-surface hybrid three-dimensional road network is a point-line-surface hybrid network with the relationship between geometry (physical connectivity) and semantic topology (including objective traffic conditions and subjective traffic willingness).

As an embodiment of the present invention, the combining the point, line and surface elements to form a point-line-surface hybrid three-dimensional road network model includes:

firstly, combining point elements and horizontal line elements based on the graph theory principle, specifically comprising: connecting the place-occupying point with the entity point to generate a relationship line; connecting the solid point with the structure line to generate a connecting line; and constructing a path model of the combination of the point elements and the horizontal line elements, and supplementing node information. And processing the T-shaped connection position in an intersection processing mode. The intersections also include four-way intersections, three-way intersections, "T" intersections, and the like. The path after the intersection is connected is relatively rigid, and the walking habit of the pedestrian is not well met at the corner; the intersection processing mode is characterized in that the distance between two connected central axes and a boundary line of a corridor is measured, a half arc is constructed at an intersection by taking the minimum distance as a radius and is tangent to the two central axes, and the whole 'point-line' road network is ensured to be more accordant with the walking characteristics of people.

Then, a path model in which the point elements and the horizontal line elements are combined is combined with the surface elements and the vertical line elements to form a three-dimensional road network model in which points, lines and surfaces are mixed.

As an embodiment of the present invention, the geometric information, the topology information and the semantic information after the completion processing are fused with the structural information in the data structure table of the point, line and surface elements in the point, line and surface hybrid stereo road network model.

In this embodiment, as shown in fig. 9, the constructing of the data storage structure of the road network navigation requirement based on the point, line and plane structure includes: and adding a topological relation field among the elements in the database, wherein the topological relation field comprises adjacent surfaces, adjacent lines and adjacent points. The topological relation field, such as the information of the adjacent surface, the adjacent line (left), the adjacent line (right) and the adjacent point, makes the extracted elements which are divided mutually establish the mutual topological relation. As shown in table 9, it is a point, line, and surface object topological relation example table;

TABLE 9

In the present embodiment, as shown in fig. 9, S1, S2 are surface areas of any two rooms; p (1,2,3,4) is an extracted point element; l (1, 2.., 19) is an extracted line element; p_(1,2,...,16)Boundary point information extracted for the face region S1; a1 is an obstacle surface in the face area; a1 is the barrier line in the face area.

The constructed road network information and road network model can be superposed with multi-mode maps such as a two-dimensional indoor map, a three-dimensional model, a live-action map and the like, so as to meet the requirement of indoor position service.

Fig. 10 is a schematic diagram of road network extraction with point, line and plane mixing according to the embodiment of the invention.

According to the embodiment of the invention, an indoor point-line-surface road network model is innovatively extracted, the data production and storage format standard of a point-line-surface road network is provided, a system for automatically constructing a complex point-line-surface mixed three-dimensional road network based on a BIM model is constructed, the adaptability limitation of the traditional single vector road network or regular/irregular grid road network model on the expression of the passable condition of the complex (complex space unit structure, form and function) is made up, the phenomenon that the extraction of the free activity region road network is not fit for actual walking is eliminated, the generation efficiency of the complex synthetic in-vivo navigation network can be effectively improved, the method has the advantages that the generation cost of the navigation network is reduced, the three-dimensional navigation network of a large-scale indoor space is rapidly and automatically extracted, data support is provided for realizing real-time navigation of each main body in a complex comprehensive body, and further development and application of indoor position service are promoted.

Claims

1. A point, line and surface mixed three-dimensional road network automatic construction method of a three-dimensional model of a real scene of a complex is characterized by comprising the following steps:

combining the point, line and surface elements to form a point, line and surface mixed three-dimensional road network model; fusing the information data after the completion processing with point, line and plane structure information in the point, line and plane mixed three-dimensional road network model to construct a data storage structure of road network navigation requirements based on the point, line and plane structures;

the normalization function from the three-dimensional model of the target building to the road network is as follows:

N_m＝F{P_dm,L_dm,S_dm|R(P_dm,L_dm,S_dm)}

wherein N is_mIs a point-line-surface mixed three-dimensional road network model, P_dmData model extracted for regularization of point elements, L_dmData model regularly extracted for line elements, S_dmData model, R (P), extracted for regularization of surface elements_dm,L_dm,S_dm) The relative topological relation of the point, line and surface road network elements is shown;

the method comprises the following steps of constructing a point element data model function in a road network, defining the point abstract characteristics of a scene object and the structural requirements of a set point element, determining an extraction method of the point element, and performing regularized extraction on the point element, wherein the extraction method comprises the following steps:

firstly, constructing a point element data model function in a road network:

P_dm＝f_p(P-Rule₁,P-Rule₂,…,P-Rule_n|Me_p)

wherein, P_dmFor the data model of regular extraction of point elements, P-Rule represents the Rule of some type of point abstract expression, n is the number of point element classes, Me_pA point element regularized extraction method;

secondly, defining point abstraction features of the scene object, including:

again, the structural requirements of the setpoint element include:

extracting the face object of the corresponding point element needing to be extracted in the road network base layer through semantic interpretation, and extracting the centroid point or the central point of the face object to obtain a point element extraction graph of an indoor scene; if the point elements needing to be extracted are entity points or connected points, extracting a central point; if the point element needing to be extracted is the occupied point, extracting the centroid point;

the method comprises the following steps of constructing a line element data model function in a road network, defining line abstract characteristics of a scene object, setting structural requirements of line elements, determining an extraction method of the line elements, and performing regularized extraction on the line elements, wherein the extraction method comprises the following steps:

firstly, constructing a road network central line element data model function:

L_dm＝f_l(L-Rule₁,L-Rule₂,…,L-Rule_m|Me_l)

wherein L is_dmFor the data model of regular extraction of line elements, L-Rule represents the Rule of a certain type of line abstract expression, m is the number of line element classes, Me_lA line element regularized extraction method;

secondly, defining line abstraction features of the scene object, including:

thirdly, the setting of the line element configuration requirements includes:

the structure lines comprise a horizontal space structure line and a vertical space structure line, and the extraction method of the horizontal space structure line adopts a middle axis transformation method to extract the central axis of the path of the passing area; the extraction method of the vertical space structure line is to extract the central axis of the three-dimensional space of an elevator shaft and an elevator room in the three-dimensional model of the target building; the extraction method of the connecting line is to connect the entity point with the structure line to obtain the connecting line; the extraction method of the relation line is to connect the position occupying point with the entity point to obtain the relation line; the method for extracting the corner line comprises the steps of measuring and calculating the distance between two central axes which are connected with each other and a boundary line of a corridor, and constructing a half arc tangent to the two central axes at the corner by taking the minimum distance as a radius to serve as the corner line;

the method comprises the following steps of constructing a road network surface element data model function, defining surface abstract characteristics of a scene object, setting the structural requirements of surface elements, determining an extraction method of the surface elements, and performing regularized extraction on the surface elements, and comprises the following steps:

S_dm＝f_s(S-Rule₁,S-Rule₂,…,S-Rule_z|Me_s)

2. The method of claim 1, wherein said constructing a mapping of said three-dimensional model of said target building to a road network substrate comprises:

3. The method of claim 1, wherein the obtaining information data of indoor elements from the three-dimensional model of the target building comprises:

4. The method according to claim 1, wherein the fusing the complete processed geometric information, semantic information and topological information with the mapping relationship of the three-dimensional model of the target building to the road network base layer comprises:

5. The method according to claim 1, wherein said combining said point, line and surface elements to form a point-line-surface hybrid stereo road network model comprises:

6. The method according to claim 1, wherein said constructing a data storage structure of road network navigation requirements based on point, line and plane structure comprises: