JP4206449B2 - Method for generating 3D electronic map data - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a three-dimensional modeling technique for generating three-dimensional electronic data of a building, and a technique for generating three-dimensional electronic map data using the modeling method.
[0002]
[Prior art]
In recent years, the use of map data digitized so as to be usable by computers (hereinafter referred to as “electronic map data”) has become widespread. The electronic map data is used for displaying a map on a so-called personal computer, a vehicle-mounted navigation system, providing a map via the Internet, and creating a map as a printed matter. In navigation systems, three-dimensional display is being used so that the driver can intuitively determine the course. Displaying a building or the like three-dimensionally has an advantage that it is easy to grasp the current position and the route.
[0003]
In order to perform three-dimensional display, a three-dimensional model of a building is required. It requires a great amount of labor to individually identify three-dimensional data, particularly height-related data, for a large number of buildings to be displayed on a map. Conventionally, various techniques for reducing such labor have been proposed.
[0004]
For example, Patent Document 1 discloses a technique for generating a three-dimensional model using three-dimensional coordinates of points on the surface of the ground surface obtained from a laser emitted from an aircraft and photographic images / video images taken from the aircraft. Yes. The three-dimensional model is performed by selecting one of contour polygons prepared in advance and deforming it so as to match the measurement result.
[0005]
[Patent Document 1]
JP-A-2002-74323
[0006]
[Problems to be solved by the invention]
In the prior art, since a three-dimensional model is generated in a range of contour polygons prepared in advance, a three-dimensional model cannot be generated for a building having a complicated shape not included in the contour polygon. Further, since the measured height data is used only for the purpose of determining the height of the contour polygon, the height data cannot be fully utilized.
[0007]
In recent years, the measurement of height data using a laser has made the interval between measurement points as small as several meters and the measurement accuracy has been improved to several tens of centimeters. There remains room for further improvement in the generation accuracy of the three-dimensional model. In general, measurement data contains a lot of noise, and therefore, appropriate processing of noise becomes one of the issues when using measurement data.
[0008]
The present invention has been made in view of such a problem, and an object thereof is to enable accurate three-dimensional modeling with a light load by utilizing measured height data.
[0009]
[Means for solving the problems and their functions and effects]
  In order to solve at least a part of the above problems, in the present invention, three-dimensional electronic data of a building is generated by the following three-dimensional modeling method. Two-dimensional electronic map data including a building to be generated is input. Moreover, the height data measured about many points in the area | region containing the building used as a production | generation object are input. Then, the building area of the building is identified from the two-dimensional data, and the height data existing in the building area is extracted as the building height data. A three-dimensional model of the building is generated by specifying the height of the building based on the building height data thus extracted. Moreover, the process which smoothes each wall surface of the produced | generated three-dimensional model by a plane is performed.A building is a building having a hierarchical structure. When generating the three-dimensional model, the height of each layer of the building is specified based on the frequency distribution of the building height data. This height data is data whose height is measured in a predetermined mesh unit defined on the ground surface. The smoothing process is performed by a vertical plane that specifies a corner mesh positioned at a corner of the building and passes through representative points of the two corner meshes.
[0010]
  According to the three-dimensional modeling method of the present invention, by using two-dimensional electronic map data, it is possible to accurately extract data useful for building generation from the measurement result of height data. That is, useless data can be removed from the measurement result of height data by extracting a specific area based on the two-dimensional electronic map data and incorporating it into the three-dimensional model. For example, it is possible to easily and accurately remove noise caused by trees planted around a building, cars and people existing around the building, and the like. It is also possible to remove noise easily and accurately by specifying areas other than buildings to be modeled, such as general roads, alleys between buildings, parks, etc., from 2D electronic map data and attribute information thereof. It becomes possible. By extracting the height data related to the building in this way, the noise processing itself included in the extracted building height data can be facilitated and the modeling accuracy can be improved.Moreover, since the process which smooth | blunts each wall surface of the produced | generated three-dimensional model with a plane is performed, the reality of a three-dimensional model can be improved more.
[0011]
In the present invention, the building height data can be extracted by expressing the measurement points of the height data in the same coordinate system as the two-dimensional electronic map data, for example. When the height data is measured in units of a measurement surface having a predetermined area, only the measurement surface data included in the building area may be extracted, or at least a part of the building area may be extracted. You may extract the measurement surface which overlaps.
[0012]
The building height data thus extracted may be further subjected to processing for removing noise by comparing the magnitude relationship between the measured value of the height data and a predetermined reference value. For example, a method can be used in which a low value that cannot be taken by a building is set as a reference value, and data below this reference value is regarded as noise. Conversely, a method of removing noise by setting a high value that can be regarded as building measurement data as a reference value and extracting data that is equal to or higher than the reference value as normal data may be adopted. . In addition, the distribution of height data is analyzed at multiple measurement points, and scattered data that does not reach the standard area that can be regarded as a part of the building is regarded as noise. A method of removing noise by extracting data distributed in a batch as normal data may be adopted.
[0013]
Building modeling can be implemented in various ways. As a first aspect, in the present invention, when a building having a hierarchical structure is to be modeled, the height of each layer can be specified based on the frequency distribution of the building height data. In the frequency distribution, a height range (hereinafter referred to as “hierarchical data group”) in which the frequency increases for each hierarchy appears according to the shape of the building. For example, the maximum frequency is obtained for each hierarchical data group. The height of each layer can be specified using various statistical values such as height, average value, and median value. When three or more hierarchical data groups appear, the height may be determined independently for each group, or the constraint that the hierarchical height changes linearly may be considered.
[0014]
When three-dimensional modeling of a building having a hierarchical structure is performed, it is preferably performed in hierarchical order. As an example, the target hierarchy to be modeled is set in order from the lower hierarchy, and the part having the building height data equal to or higher than the height of the target hierarchy is specified as the region belonging to the target hierarchy, and the target hierarchy The method of modeling can be taken. In general, since the flat seats of a building are larger in the lower level, it is possible to model smoothly by stacking models of each level in the virtual space. Further, by performing three-dimensional modeling of a building having a hierarchical structure by such a method, for example, water tanks and elevator facilities existing on the roof can be modeled relatively faithfully. The modeling result generated in this way can be used for radio interference simulation, atmospheric simulation, sunshine simulation, and the like.
[0016]
When the height is measured in units of a predetermined mesh defined on the ground surface, the wall of the 3D model is likely to have irregularities according to the mesh shape, so in such cases, smoothing processing is particularly effective. It is. Smoothing can be performed, for example, by specifying a corner mesh located at a corner of a building and defining a vertical plane passing through a representative point of the two corner meshes as a wall surface.
[0017]
Various methods can be used to identify the corner mesh. For example, among meshes corresponding to the building height data, the outer periphery mesh that defines the outer periphery of the building is specified, and the line segments sequentially connecting the outer periphery meshes are identified. You may specify based on the angle made. For example, a mesh in which the angle between line segments is 90 ° or less can be identified as a square mesh. For the lower hierarchy, the corner of the building may be found from the building area of the two-dimensional map data, and the mesh corresponding to this corner may be specified as the corner mesh.
[0018]
For the lower layer, the wall surface may be smoothed by a Boolean operation that cuts the three-dimensional model along the vertical plane corresponding to the building area.
[0019]
In the present invention, the three-dimensional modeling method may be performed in the second mode described below. In the second aspect, a three-dimensional basic model used for generating a three-dimensional model is prepared in advance in association with attribute information representing the attribute of the building. Moreover, the attribute information of the building used as a production | generation object is input, and a basic model is selected according to this attribute information. Modeling is performed by deforming and arranging the basic model selected in this way based on the shape of the building area and the building height data. By using the shape of the building area and the building height data, the basic model can be deformed and arranged with higher accuracy, and the reality of the three-dimensional model can be improved.
[0020]
The attribute information can include the use of a building such as a condominium, office building, commercial building, or highway. This attribute information may be individually input by an operator, or may be read from this electronic map data when recorded in two-dimensional electronic map data.
[0021]
The present invention may be configured as a 3D map data generation method for generating 3D electronic map data including a plurality of buildings. In the present invention, three-dimensional electronic map data is generated by the following procedure. First, for a target area for generating 3D electronic map data, 3D electronic map data representing each building with a 3D model whose appearance has not been generated is input. It does not matter whether this three-dimensional model is generated by the three-dimensional model generation method of the present invention. Along with the three-dimensional model, wide-area color image data obtained by photographing the target area is also input. From the color image data thus input, the surface color of each building is specified, and appearance data representing the appearance of the three-dimensional model is generated using the specified color.
[0022]
By doing so, it is possible to arrange the appearance of a plurality of buildings without having to prepare surface pasted images or the like individually. Moreover, reality can be improved about each building by employ | adopting the color near reality. For the generation of appearance data, methods such as image pasting and computer graphics drawing can be applied. As the appearance data, not only the color of the model surface but also data such as windows and doors may be appropriately arranged according to the type and shape of the building. The method of the present invention is particularly useful for generating wide-area three-dimensional map data in which a large number of buildings are arranged.
[0023]
If the building is a building having a hierarchical structure, appearance data is generated by the above-described method for two or more floors of the building, and appearance data is separately generated for the first floor of the building. May be. For example, for the first floor, individually captured image data may be pasted. In buildings, since the appearance often differs between the first floor and the second floor or more, the reality of the model can be further improved by generating the appearance data by dividing the two.
[0024]
The appearance data of the first floor portion may be generated using, for example, basic appearance data used for generation of appearance data in advance in association with attribute information representing the attribute of a building. That is, building attribute information may be input, and basic appearance data may be selected in accordance with the attribute information to generate first floor appearance data. For example, an entrance that seems to be a condominium, an entrance that seems to be an office building, and the like are prepared in advance as basic appearance data, and this basic appearance data is selected and used in accordance with attribute information such as an apartment or office building. By carrying out like this, the reality of the 1st floor part can be improved more.
[0025]
In the 3D map data generation method of the present invention, an accessory to be attached to the surface of the building such as a signboard, a veranda, or an antenna may be further generated. For example, prepare basic accessory data to be used for generating accessories in association with the attribute information representing the attributes of the building, and select and use this basic accessory data based on the attribute information. A method of generating object data can be taken. By doing so, the reality of the three-dimensional model can be further improved.
[0026]
As the accessory data, the basic accessory data may be applied as it is, or the attribute information may be reflected in the basic accessory data. For example, the shape may be changed, and the name of a building or a store in the building may be attached to a signboard or the like.
[0027]
In the present invention, as described above, it is possible to select and use a three-dimensional basic model, appearance data, and accessory data by using building attribute information. In this selection, other conditions may be considered in addition to the building attribute information. For example, in an area where the external appearance of a building is unified, external data common to each building may be applied.
[0028]
The present invention may be configured as a three-dimensional electronic map data method in which the above-described three-dimensional modeling method and three-dimensional electronic map data generation method are combined. Moreover, you may comprise with aspects, such as a three-dimensional modeling apparatus and a three-dimensional map data generation apparatus. Moreover, you may comprise as a recording medium which recorded the computer program for implement | achieving the three-dimensional modeling of this invention and the production | generation of three-dimensional electronic map data with a computer, and this program. Recording media include flexible disks, CD-ROMs, magneto-optical disks, IC cards, ROM cartridges, punch cards, printed matter printed with codes such as bar codes, computer internal storage devices (memory such as RAM and ROM), and Various computer-readable media such as an external storage device can be used.
[0029]
DETAILED DESCRIPTION OF THE INVENTION
The embodiment of the present invention will be described by dividing it into the following items.
A. Configuration of map data generator:
B. Database structure:
C. 3D modeling:
D. Virtual space preprocessing:
E. Building height data setting processing:
F. Building modeling process:
G. Housing and road modeling process:
H. Appearance data generation processing:
I. Accessory data generation processing:
[0030]
A. Configuration of map data generator:
FIG. 1 is an explanatory diagram showing a configuration of an electronic map data generation system 100 (hereinafter also simply referred to as a system 100) as an embodiment. The system 100 performs three-dimensional modeling of a building and generates three-dimensional electronic map data. In this embodiment, the system 100 is configured by constructing the functional blocks shown in the figure in software. In the following example, a case where the system is configured by a single computer will be described as an example, but a configuration in which a host computer and a terminal are connected via a network may be used.
[0031]
The data input unit 101 inputs modeling data from the outside. The modeling data is data required for three-dimensional modeling, and includes height measurement data, aerial photographs, and shooting positions for a wide area for which an electronic map is generated. The height measurement data is the result of dividing the area into a mesh of several meters and measuring the height in mesh units. A technology that can measure the height of each point with a mesh of about 2m and a measurement system of less than about 20cm by scanning a laser from the aircraft toward the ground is known. Data obtained from Kogyo Co., Ltd. (registered trademark) can be used. The modeling data may be input through a medium such as the magneto-optical disk MO, or may be input via a network or other communication.
[0032]
The preprocessing unit 110 performs processing such as setting of a virtual space used in three-dimensional modeling and correction of height data as preprocessing of three-dimensional modeling. The preprocessing unit 110 refers to the two-dimensional map database 103 in order to perform this function. The two-dimensional map database 103 is an existing map database on a plane, and records elevations in mesh units that divide the ground surface, planar shapes of buildings, and the like. As the two-dimensional map database 103, for example, data obtained by superimposing an aerial photograph, a satellite photograph, a house map, etc. on the elevation data of the Geographical Survey Institute prepared in 50m mesh can be used.
[0033]
In this embodiment, the building is modeled by two methods. The first method is modeling by raising each part of the building area in the height direction in units of meshes based on the height data. The second method is modeling by modifying a basic three-dimensional model (hereinafter referred to as a basic model) prepared in advance. The building modeling unit 120 performs first modeling. In this embodiment, a relatively large building such as a condominium, an office building, or a commercial building such as a department store is the target of the first modeling.
[0034]
The house / road modeling unit 130 performs second modeling. In the present embodiment, general houses and elevated roads such as expressways were modeled. The basic model referred to for performing the second modeling is stored in the basic model database 102.
[0035]
The appearance data generation unit 140 has a function of adjusting the appearance of the modeled building. A surface is affixed to a wireframe model, the surface is colored, and windows, entrances, etc. are generated. In this embodiment, in order to improve the reality of appearance, parts such as windows and entrances are prepared in advance in the appearance part database 104, and the appearance data generation unit 140 appropriately generates appearance data by referring to this database. It was supposed to be.
[0036]
The accessory data generation unit 150 generates data (hereinafter referred to as accessory data) representing an accessory attached to a building such as a signboard or a veranda. In this embodiment, in order to improve the appearance reality, parts such as a signboard are prepared in the accessory parts database 105 in advance, and the accessory data generation unit 150 generates accessory data with reference to this database as appropriate. To do. The accessory data may include, for example, data necessary for expressing the cityscape other than buildings, such as trees, signals, guardrails, and the like.
[0037]
The integration unit 106 has a function of associating data generated in each of the above-described functional blocks and adjusting the format of the electronic map data. The setting of characters to indicate buildings and place names and various symbols to be displayed on the map are also performed. The integration unit 106 outputs the electronic map data thus integrated to the map database 10. The electronic map data may be recorded on a DVD-ROM or other recording medium ME.
[0038]
B. Database structure:
FIG. 2 is an explanatory diagram showing the structure of the two-dimensional map database 103. The two-dimensional map database 103 is an existing map database on a plane, and records elevations in mesh units that divide the ground surface, planar shapes of buildings, and the like. In the center of the figure, a map represented by the two-dimensional map database 103 is illustrated. Each object such as a building is represented by a closed figure (hereinafter referred to as a polygon), and is associated with attribute information that gives the type of each building. An example of a residential building is shown above the figure. For this building, the latitude and longitude of each vertex are stored as polygon shape data together with a polygon-specific ID “HOUSE001”. Further, as the attribute information, a building type “house” and a floor number “2nd floor” are stored. The attribute information can include various types of information other than those exemplified, such as the floor area and the direction of the entrance.
[0039]
Below the figure is an example of a building structure. As in the case of a house, a polygon-specific ID and shape are stored, and the building type “office building” and “floor” are stored as attribute information. Furthermore, the name of the building, the type and name of the tenant contained in the building, and the like may be included in the attribute information. The building type can be set arbitrarily, but in this embodiment, for convenience of explanation, it is assumed that three types of office building, condominium, and commercial building are set as the building type.
[0040]
Here, a house and a building are illustrated, but roads and land are similarly defined by polygons, and attribute information is also stored in each. The road attribute information may include, for example, a road type such as an expressway and a national road, and in the case of an elevated road, its height, the number of lanes of the road, and the like.
[0041]
FIG. 3 is an explanatory diagram showing the structure of the basic model database 102. The basic model is a three-dimensional model that the house / road modeling unit 130 refers to when modeling. The basic model database 102 stores a large number of basic models in accordance with the type of building, the arrangement direction, the height, and the like. In the figure, two types of two-story buildings facing west and two-story facing south are shown as examples of basic housing models. Two or more types of models having different planar shapes may be prepared for sections such as “west facing two stories” and “south facing two stories”.
[0042]
In the modeling of the house / road modeling unit 130, these basic models are appropriately modified to generate a three-dimensional model of each building. At this time, in order to avoid an unrealistic three-dimensional model due to extreme deformation, it is preferable to predefine an allowable deformation range for each basic model. In this embodiment, the deformation rate is defined. For example, for a house on the second floor facing west, expansion and contraction in the range of 0.95 to 1.05 is permitted in the east-west direction, and expansion and contraction of 0.9 to 1.1 is permitted in the north-south direction. The deformation rate may be set for each part of the house. For example, for a south-facing second-floor house, the deformation rate can be set separately in a north-south direction into a part without a balcony (section a) and a part with a balcony (section b). Also in the east-west direction, the deformation rate can be set by dividing into three sections c to e in consideration of the floor plan. Here, the setting of the two-dimensional deformation rate is exemplified, but an allowable range of the deformation rate may be set in the height direction.
[0043]
Basic models are available not only for houses but also for elevated roads. In the case of an elevated road, for example, a model of a road around a unit length including struts, a model such as a toll booth, and a junction can be prepared.
[0044]
A basic model may be prepared for an apartment or office building. In this embodiment, since these buildings are three-dimensionally modeled by the building modeling unit 120, the basic models for these buildings are not necessary in principle. However, if these models are prepared, for example, a building that cannot be modeled by the building modeling unit 120 due to noise of height data or unknown plan shape of the building, etc. Modeling using the model becomes possible.
[0045]
FIG. 4 is an explanatory view showing the structure of the appearance part database 104. The appearance part data is pasted image data that is pasted on the outer surface of the model in order to improve the reality of the building. Appearance part data is prepared according to attribute information such as the type of building. Since the appearance of the building may differ between the second floor and above and the first floor portion, in this embodiment, the appearance part data is prepared by further dividing by the application site. As shown in the figure, for commercial buildings, the exterior part data for the second and higher floors includes various types of windows, and the exterior part data for the first floor includes show windows and entrance data. It is. Data on windows, entrances, etc. can be prepared for office buildings and apartments. For houses, windows, roofs, shutters, etc. can be prepared.
[0046]
FIG. 5 is an explanatory view showing the structure of the accessory parts database 105. The accessory part data is three-dimensional model data for representing an accessory attached to a building in order to improve the reality of the building. The accessory part data is prepared according to attribute information such as the type of building. As shown in the figure, the office building includes a signboard. Since the signboard has a different shape depending on the mounting position, in this embodiment, the signboard is prepared separately from those attached to the side of the building and those attached to the rooftop. For the items attached to the side, information on the attachment position is also stored. As an appendage of the elevated road, for example, a sound insulation wall, a road guide plate, a median strip, etc. can be prepared. For houses, verandas, antennas, etc. can be prepared.
[0047]
C. 3D modeling:
FIG. 6 is a flowchart of the three-dimensional modeling process. This is a process executed by the system 100 of the present embodiment using the various functional blocks and database described above. Here, an overview of the entire process will be described, and detailed contents of each process will be described later in order.
[0048]
When the process is started, the system 100 first inputs the two-dimensional map data and the height measurement data for the target region for generating the three-dimensional map data (step S100).
[0049]
Next, as preprocessing for modeling, virtual space preprocessing (step S200) and building height data setting processing (step S250) are executed. The virtual space preprocessing is processing for defining a virtual space for performing three-dimensional modeling and defining a smoothed ground surface without a step. The building height data setting process is a process of comparing height measurement data with two-dimensional map data and extracting only height data corresponding to a building (hereinafter referred to as building height data). is there. The height data input in the present embodiment is the result of scanning the entire target area regardless of whether it is a building or not. Since the measurement accuracy of the height data is very high, these data include data unrelated to the building such as roadside trees, pedestrians, and vehicles. In the building height data setting process, data unrelated to the building is removed in this way.
[0050]
Based on the data thus obtained, the system 100 refers to the attribute information of the two-dimensional map data, sequentially selects objects corresponding to the building, and performs building modeling (step S300). The building modeling is a modeling method in which the height of each part of the building is determined based on the building height data, not the deformation of the basic model.
[0051]
For objects other than buildings, the system 100 performs house and road modeling (step S400). This is modeling by deforming and arranging the basic model.
[0052]
Building modeling, housing, and road modeling may be performed in any order. For example, only building objects may be extracted, and after building modeling is completed for all objects, housing and road modeling may be performed on the remaining objects. Alternatively, unprocessed objects may be sequentially selected. If the object corresponds to a building, building modeling may be executed. If not, housing and road modeling may be performed.
[0053]
With the above processing, a three-dimensional model is obtained with a wire frame for the target region. Next, the system 100 generates appearance data (step S500) and accessory data for each three-dimensional model (step S600). These data are generated by selecting one suitable for the attribute information of each three-dimensional model from the appearance part database 104 and the accessory part database 105 and transforming it so as to fit the three-dimensional model. The generation of the appearance data and the generation of the accessory data may be performed in parallel, or after either one is performed, the other may be performed.
[0054]
When the data for each three-dimensional model is completed by the above processing, the system 100 associates these data in a predetermined format, registers them as map data, and completes the processing (step S700). Hereinafter, the detailed contents of the above-described processes will be sequentially described.
[0055]
D. Virtual space preprocessing:
FIG. 7 is a flowchart of the virtual space preprocessing. The system 100 first smoothes the ground surface mesh (step S202). The smoothing process outline is illustrated in the figure. The system 100 inputs elevation data from the two-dimensional map database 103. This altitude data records the altitude in mesh units obtained by dividing the ground surface by a predetermined distance d. When the altitude data is reproduced purely in the virtual three-dimensional space, the ground surface has a staircase shape like the ground surface Sg1 in the figure. Although it is possible to use a stepped ground surface, in this embodiment, a ground surface Sg without a step is set by defining a triangular polygon that connects the centers of gravity of the mesh.
[0056]
Next, the system 100 arranges the building on the ground surface thus obtained. First, the system places a two-dimensional polygon of a building on a two-dimensional map database (step S204). In the figure, a state in which the building polygon POL is arranged on the two-dimensional ground surface Sg is shown. Based on this arrangement, the system 100 obtains the intersection (◯ in the figure) between the polygon POL and the triangular polygon that defines the ground surface Sg in addition to the vertex (● in the figure) that defines the polygon POL.
[0057]
The system 100 projects the vertices and intersections thus obtained on the three-dimensional ground surface Sg, and places the two-dimensional polygon on the ground surface (step S206). The system 100 executes the above processing for each polygon in the target region of the three-dimensional modeling, and completes the virtual space preprocessing.
[0058]
E. Building height data setting processing:
FIG. 8 is a flowchart of the building height data setting process. Although it is possible to execute this processing based on the triangular polygon defined in the virtual space preprocessing, in this embodiment, processing is performed using a separately defined mesh of 2 m units.
[0059]
The system 100 superimposes and arranges a building polygon (hereinafter referred to as a building area) POL1 and height data on a two-dimensionally defined mesh of 2 m (step S252). The square in the figure represents a mesh, and the circle represents the measurement position of height data.
[0060]
As shown in the figure, the measurement position of the height data does not necessarily coincide with the mesh. In order to facilitate subsequent processing, the system 100 interpolates the height data and sets the height data of each mesh (step S254). In the figure, an enlarged view of region A is shown. For example, for hatched meshes, the height data at each measurement position is interpolated based on the distance between the four measurement positions (● in the figure) and the center of gravity Mc of the mesh. , Height data can be set.
[0061]
When the height data is obtained in units of meshes in this way, the system 100 sets the height data present in the building area as building height data. The hatched mesh in the figure corresponds to the building height data. In this embodiment, a mesh that overlaps 50% or more in the building area is extracted. In the figure, different hatching is given depending on the height. The system 100 executes the above process for each polygon in the target region of the three-dimensional modeling, and completes the building height data setting process. Thus, by setting the building height data using the building area of the two-dimensional map data, noise data unrelated to the building can be removed relatively easily and accurately, and the subsequent processing Accuracy can be improved.
[0062]
The mesh that defines the building height data is not limited to the rules described above, and can be selected according to various rules. For example, a mesh included in a building area may be extracted. You may extract the mesh which at least one part overlaps with a building area.
[0063]
In this embodiment, the building height data is extracted on the basis of the building area, but conversely, an area other than the building may be used as a reference. For example, specify areas other than the building to be modeled, such as general roads, alleys between buildings, and parks, and remove the height data corresponding to these areas as data unrelated to the building. Also good.
[0064]
In specifying the building height data, the measurement value of the height data may be further taken into consideration. For example, a low value that cannot be taken by a building may be set as a reference value, and data below this reference value may be regarded as noise and removed. Conversely, a method of removing noise by setting a high value that can be regarded as building measurement data as a reference value and extracting data that is equal to or higher than the reference value as normal data may be adopted. .
[0065]
F. Building modeling process:
First three-dimensional modeling in this embodiment, that is, building modeling that determines the height of each part using building height data without using a basic model will be described. In the embodiment, it is intended for buildings, but it can also be applied to ordinary houses and elevated roads.
[0066]
FIG. 9 is an explanatory diagram showing an example of building height data. The perspective view of the building to be modeled is shown in the upper part of the figure, and the building data is shown in the lower part. For convenience of explanation, processing contents will be described by taking such a building as an example. As shown in the figure, this building is composed of three rectangular parallelepiped parts including a lower floor defining a building area and a tower low and a tower height formed separately on the lower floor. The building data includes a white mesh corresponding to the lower floor, a hatching mesh corresponding to the tower low, and a cross hatch mesh corresponding to the tower height.
[0067]
FIG. 10 is a flowchart of the building modeling process. First, the system 100 inputs building height data and analyzes the frequency distribution (step S302). An example of analysis is shown in the figure. The horizontal axis indicates the height, and the vertical axis indicates the frequency, that is, the number of meshes having each height data. Since the building height data includes measurement errors, the height data does not match even for meshes corresponding to lower floors, for example. In addition, since the building area does not necessarily exactly match the mesh, the building height data may include data other than the building such as street trees, vehicles, and pedestrians.
[0068]
As shown in FIG. 9, peaks corresponding to respective heights of a low hierarchy, a tower low, and a tower height appear in the frequency distribution for the hierarchized building. In the example in the figure, group A in the vicinity of 16 m in height is a lower floor, and group B in the vicinity of 28 m in height is a peak corresponding to the tower low. The peak corresponding to the tower height is not shown.
[0069]
Based on the frequency analysis result, the height of each part in the hierarchical structure can be specified. First, data in the vicinity of a height of 2 m is excluded as data unrelated to the building. The height HL serving as a reference for exclusion can be arbitrarily set. For example, the height HL can be set to a height corresponding to the first floor in a normal building.
[0070]
Next, a height corresponding to the first peak, group A (hereinafter referred to as a hierarchical height) is specified. In the present embodiment, the height that is the maximum frequency in the group A is set to the hierarchical height.
[0071]
The layer height can be specified by various methods. For example, the average, median, etc. of group A may be used as the layer height. In the present embodiment, the range in which the frequency is NL or more is set as the group A range, and the average value in this range is set as the hierarchy height. The frequency NL that defines the range of the group A can be arbitrarily set. For example, the frequency NL can be set to a lower limit value that can be said to be a significant frequency.
[0072]
The system 100 quantizes the building height data based on the analysis result (step S304). Quantization is a process for unifying the height data of each mesh to the hierarchical height obtained from the analysis result. For example, quantization can be performed by giving a group height of 16 m to all meshes included in group A, and giving a group height of 28 m to meshes included in group B. You may quantize so that a mesh height may be given with respect to the mesh which is less than the frequency NS corresponding to the area which can be regarded as a part of a building.
[0073]
For a mesh that cannot be specified to which layer, for example, a mesh that is closest to the height data of the mesh may be assigned. For this allocation, a constraint such as not exceeding the mesh height data may be imposed.
[0074]
By quantization, as shown in the figure, a frequency distribution clearly divided for each hierarchy is obtained. The system 100 performs modeling in units of meshes based on the quantized height data (step S310). As will be described later, this process is a process for generating a three-dimensional model as an assembly of quadrangular prisms by lifting each mesh by an amount corresponding to the height data.
[0075]
Since modeling in units of meshes is an assembly of quadrangular prisms, the wall surface may be uneven depending on the positional relationship between the building area and the mesh. Therefore, the system 100 performs a wall surface smoothing process to remove such irregularities (step S370). The above process is executed for all the buildings in the target area, and the system 100 ends the building modeling process.
[0076]
FIG. 11 is a flowchart of modeling in units of meshes. This process corresponds to step S310 in FIG. In this embodiment, the processing is sequentially performed from the lower layer. First, the system 100 sets the processing target hierarchy to the lowest hierarchy located at the bottom of the building (step S312). Here, the lowest hierarchy does not mean the so-called first floor of a building, but the lowest hierarchy recognized by the frequency distribution analysis of FIG. For the building shown in FIG. 9, the lower hierarchy part is set as the target hierarchy.
[0077]
The system 100 performs three-dimensional modeling of the target hierarchy by translating the mesh corresponding to the target hierarchy in the height direction to the building height corresponding to the target hierarchy (step S314). The system 100 executes modeling of all layers while sequentially changing the target layer to the next higher layer (steps S316 and S318).
[0078]
Here, in this embodiment, all the meshes having height data higher than the height corresponding to the target hierarchy are handled as meshes belonging to the target hierarchy. In the figure, a three-dimensional modeling example is shown. A three-dimensional model corresponding to the building shown in FIG. 9 is generated. As shown in this example, at the time of low-level processing, all the meshes corresponding to the tower low and tower height are handled as meshes belonging to the low level. By doing so, it is possible to avoid holes corresponding to the tower low and tower height portions in the low-level three-dimensional model A, and it is possible to facilitate later Boolean calculations.
[0079]
When modeling tower low, meshes corresponding to tower height are all handled as meshes belonging to tower low. Therefore, at the time of processing of this hierarchy, a model B1 with a tower low and a model B2 below the tower height are formed. Finally, by modeling the tower height, a model C above the tower height is formed.
[0080]
When the modeling of all hierarchies is completed, the system 100 executes a process of combining the models (step S320). A Boolean operation can be applied to this processing. By this processing, one three-dimensional model obtained by combining the models A, B1, B2, and C in the figure is generated.
[0081]
In the combination process, for example, a model divided for each of the lower hierarchy, the tower low, and the tower height may be generated. This process is realized by combining only models with the same bottom area, such as models B2 and C, for example.
[0082]
Various methods can be used for modeling in units of meshes. For example, modeling may be performed sequentially from the upper floor. Apart from the concept of hierarchy, after giving a quadrangular prism independently for each mesh, the whole may be combined by a Boolean operation.
[0083]
FIG. 12 is a flowchart of the wall surface smoothing process. This process corresponds to step S370 in FIG. FIG. 13 is an explanatory diagram showing an example of wall surface smoothing processing. A broken line in the figure means a mesh, and a range surrounded by a solid line and hatched mesh means a bottom shape POL of a three-dimensional model of a building. Depending on the relationship between the building area and the mesh, the side surface may be uneven in this way. The purpose of the smoothing process is to form a smoothed wall surface as shown by the thick solid line in the figure. Hereinafter, the content of the smoothing process will be described with reference to FIG. 12 with reference to the example of FIG. 13 as appropriate.
[0084]
The system 100 identifies an edge mesh that is a boundary surface with the outside world from among the meshes forming the three-dimensional model (step S372). The edge mesh can be specified by, for example, extracting a mesh in which at least one of the four side surfaces of each mesh is in contact with a space outside the model.
[0085]
Next, the system 100 performs processing for connecting the extracted edge meshes (step S374). In this process, one edge mesh is arbitrarily selected, and adjacent edge meshes are sequentially searched and connected. In this embodiment, in order to avoid the situation where the search does not proceed due to the intersection of the connecting lines or between the two adjacent edge meshes, the search is performed as shown in the frame of step S374. Priorities were set. Each square in the figure means a mesh, and the numbers in the square mean search priority. Assuming that the current position is the center mesh, the search is performed with the meshes adjacent to the right side having the highest priority, and the priority is sequentially lowered clockwise. By this connection process, a connection line indicated by a thin arrow in FIG. 13 is obtained.
[0086]
Based on this connecting line, the system 100 identifies a corner mesh corresponding to the corner of the building (step S376). In the present embodiment, the mesh whose inner angle formed by the connecting lines is 90 degrees or less is specified as the angular mesh. In the example of FIG. 13, four places with ◯ are identified as corner meshes. The angle that serves as a criterion for determining whether or not the mesh is a square mesh can be arbitrarily set, but the initial unevenness tends to be maintained as it is larger than 90 degrees.
[0087]
When the corner mesh is specified in this way, the system 100 generates a wall surface so as to connect the representative points of the corner mesh (step S378). In the example of FIG. 13, a wall surface indicated by a thick line in the drawing is obtained. In this embodiment, the wall surface is defined by connecting the center of gravity of the mesh. However, the wall surface can be generated at various positions, and the wall surface may be set so as to include the square mesh.
[0088]
The building modeling process is completed by the above process. Since modeling is performed on the basis of building height measurement data, it is possible to accurately model buildings with complex shapes.
[0089]
G. Housing and road modeling process:
Next, the second modeling method of this embodiment, that is, modeling using a basic model will be described. This modeling is executed in a housing / road modeling process.
[0090]
FIG. 14 is a flowchart of the house / road modeling process. This process corresponds to step S400 in FIG. When the process is started, the system 100 selects an unprocessed structure (step S402). In this process, polygons with attributes of houses and elevated roads are selected from the two-dimensional map data. Polygons that have failed modeling in the building modeling described above may be included in the target.
[0091]
Next, the system 100 inputs attribute information, a building area, and a building height for the polygon to be processed (step S404). Also, a basic model is selected from the basic model database 102 based on the attribute information (step S406). At this time, conditions other than the attribute information may be considered. For example, in the case of housing processing, the direction of the entrance may be determined from the positional relationship of the road around the polygon to be processed, or the building area may be compared with the bottom shape of the basic model. You may make it judge roof shape, such as a gable and a dormitory, based on building height data.
[0092]
The system 100 deforms and arranges the selected basic model based on the building height data and the building area (step S408). Building height data can be used to identify the height of a representative portion of the building. For example, as shown in the figure, for the roof of the gable, the height H1 of the central portion, the height H2 of the eaves, and the like can be specified by the building height data. If the building area cannot be obtained with 2D map data, the basic model may be placed without deformation, or the maximum area rectangle inscribed in the land frame to be processed is obtained and built. Processing may be performed assuming a physical region. In the latter process, a shape capable of building a house may be obtained by previously removing a parking space, an interval with a neighboring house, and the like from the land frame, and a rectangle may be obtained based on this shape.
[0093]
In the above-described example, the processing for a house has been described, but the same applies to an elevated road. An elevated road can be modeled by transforming a basic model of an elevated road based on information such as building height data and lanes and arranging it along road polygons.
[0094]
H. Appearance data generation processing:
FIG. 15 is a flowchart of appearance data generation processing. This process corresponds to step S500 in FIG. This is a process for adjusting the appearance by coloring the surface of a three-dimensional polygon and pasting an image such as a window.
[0095]
In this process, the system 100 first inputs color image data of an aerial photograph as data for determining a color for coloring a polygon (step S602). The shooting conditions such as the shooting position, shooting direction, and angle of view are also entered.
[0096]
Next, as processing for taking the correspondence between the three-dimensional polygon and the aerial photograph, an image corresponding to the aerial photograph (hereinafter referred to as a corresponding view) is generated from the three-dimensional model (step S604). The viewpoint PV is set at the latitude, longitude, and altitude corresponding to the shooting position of the aerial photograph, and an image in which the three-dimensional model is viewed is generated by combining the shooting direction and the angle of view with the shooting conditions. When the aerial photograph is taken almost vertically on the ground surface, the corresponding view is a two-dimensional image. Accordingly, two-dimensional map data may be used instead of the corresponding view.
[0097]
The system 100 determines the basic color of each object by superimposing the aerial photograph and the corresponding view (step S608). The outline of the process is shown schematically in the figure. The lower side shows an aerial photograph and the upper side shows a corresponding view. The basic color of the object OBJ included in the corresponding view is the color of the image IMG corresponding to the object OBJ in the aerial photograph. In this way, the basic color can be determined for each object based on the aerial photograph.
[0098]
Next, the system 100 inputs the attribute information of the object from the two-dimensional map database 103 (step S610), and selects appearance part data from the appearance part database 104 based on the attribute information (step S612). At this time, for the building, different external parts are applied on the second floor and the first floor. When there are many applicable appearance parts, the appearance parts may be selected at random or may be selected in consideration of the positional relationship between the building and the road. For example, by considering the positional relationship with the road, appearance parts of the show window and the entrance can be preferentially applied to the road side.
[0099]
As an example, a method for determining the direction of the entrance of a building will be described. First, with reference to the two-dimensional map database 103, a line segment (hereinafter referred to as a road line segment) along the road in contact with the land where the building is arranged is specified. When the two-dimensional map database 103 has road network data for route search, link data representing roads can be used as road line segments. Next, of the line segments constituting the building frame on the two-dimensional map database, the line segment closest to the road line segment is selected. For example, it can be evaluated that the closer the angle between a line segment orthogonal to the road line segment and each line segment constituting the building frame is closer to 90 °, the more parallel the line segment is. When a line segment close to parallel is specified in this way, the entrance of the building can be set on the surface corresponding to the line segment. When a basic model is prepared in advance like a general house, the arrangement direction of the basic model may be determined so that the entrance comes to the surface set in this way.
[0100]
In addition, when there are two or more line segments that touch the land, such as corner lots, a plane that is a candidate for entrance may be specified for each road by the above-described method. Entrances may be provided on all the surfaces thus identified, or one surface may be selected from these surfaces according to a predetermined rule. As a rule, for example, a plane closest to the road may be used as an entrance; a plane in contact with a thick road may be used as an entrance; a specific direction such as south may be preferentially used as an entrance.
[0101]
As described above, when the basic color and appearance part data of each three-dimensional model are determined, the system 100 displays the image with the appearance part data pasted by coloring the surface of the three-dimensional model with the basic color. Appearance data is generated (step S614). When different appearance parts are applied to the second and higher floors and the first floor, coloring with the basic color may be applied only to the second and higher floors. The system 100 executes the above process for each object, and completes the appearance data generation process.
[0102]
In this embodiment, the aerial photograph is used to generate the appearance data, but a photograph taken from the ground may be used. For example, you may use the still image and video image which were image | photographed with the vehicle-mounted camera. Since photographs taken from the ground often take pictures of windows, signs, etc., for example, an area where the change in RGB value of each pixel constituting a building image is small is regarded as a wall, and the color of this area is changed. It is preferable to set a basic color based on this. Also, areas other than the basic colors may be recognized, and based on this, the presence / absence of a window, the presence / absence of a difference in appearance between the second floor and the first floor portion, and the like may be determined and utilized for selection of appearance part data. . By doing so, the reality of appearance data can be improved.
[0103]
I. Accessory data generation processing:
Finally, the accessory data generation process will be described. This process corresponds to S600 in FIG. The processing contents are the same as steps S610 to S614 of the appearance data generation process (FIG. 15).
[0104]
That is, the system 100 inputs attribute information for an object to be processed, and selects accessory part data from the accessory part database 105 based on the attribute information. The accessory part data obtained in this way is transformed and fitted so as to match the target three-dimensional model.
[0105]
When attaching the accessory, it is preferable to change not only the shape but also the color and the like. For example, when a signboard is attached, it is preferable to set the display contents of the signboard based on the building name and tenant information. Accessories may be colored in the same color as the building.
[0106]
According to the map data generation apparatus of the present embodiment described above, a three-dimensional model of a building can be generated with high accuracy with a relatively light load. Moreover, the data used for model generation is not individual surveying data of buildings, but data that can be identified in a wide area in a short time by laser measurement from an aircraft. Therefore, according to the present embodiment, it is possible to drastically reduce the burden for generating a wide-area three-dimensional model.
[0107]
About the above-mentioned Example, a part of process may be abbreviate | omitted or a further process may be added. In this embodiment, each process is automatically executed, but an operator's operation may be added as appropriate.
[0108]
Although various embodiments of the present invention have been described above, the present invention is not limited to these embodiments, and it goes without saying that various configurations can be adopted without departing from the spirit of the present invention. For example, each process described above may be realized by software or hardware.
[Brief description of the drawings]
FIG. 1 is an explanatory diagram showing a configuration of an electronic map data generation system 100 as an embodiment.
FIG. 2 is an explanatory diagram showing a structure of a two-dimensional map database 103. FIG.
FIG. 3 is an explanatory diagram showing a structure of a basic model database 102;
FIG. 4 is an explanatory diagram showing a structure of an appearance part database 104;
FIG. 5 is an explanatory diagram showing a structure of an accessory parts database 105;
FIG. 6 is a flowchart of a three-dimensional modeling process.
FIG. 7 is a flowchart of virtual space preprocessing.
FIG. 8 is a flowchart of a building height data setting process.
FIG. 9 is an explanatory diagram showing an example of building height data.
FIG. 10 is a flowchart of a building modeling process.
FIG. 11 is a flowchart of modeling in units of meshes.
FIG. 12 is a flowchart of wall surface smoothing processing;
FIG. 13 is an explanatory view showing an example of wall surface smoothing processing;
FIG. 14 is a flowchart of a house / road modeling process.
FIG. 15 is a flowchart of appearance data generation processing.
[Explanation of symbols]
10 ... Map database
100 ... Electronic map data generation system
101: Data input section
102 ... Basic model database
103 ... 2D map database
104 ... Appearance parts database
105 ... Accessory parts database
106. Integration unit
110: Pre-processing unit
120 ... Building Modeling Department
130… Housing, road modeling department
140 ... Appearance data generation unit
150 ... accessory data generation unit

Claims (4)

A 3D modeling method for generating 3D electronic data of a building,
(A) a computer that inputs two-dimensional electronic map data including a building to be generated and stores it in a memory ;
(B) The computer inputs height data measured for a number of points in an area including the building to be generated and stores the height data in the memory ;
(C) the computer identifying a building area of the building from the two-dimensional electronic map data, and extracting height data existing in the building area as building height data;
(D) the computer generating a three-dimensional model of the building by specifying the height of the building based on the building height data;
(E) the computer smoothing each wall surface of the generated three-dimensional model with a plane;
I have a,
The building is a building having a hierarchical structure,
The step (d) includes a step of identifying the height of each level of the building based on the frequency distribution of the building height data,
The height data is data in which the height is measured in a predetermined mesh unit defined on the ground surface,
The step (e) specifies a corner mesh located at a corner of the building, and performs the smoothing by a vertical plane passing through representative points of the two corner meshes . The three-dimensional modeling method according to claim 1 ,
The step (d)
Associating with the attribute information representing the attribute of the building, preparing a three-dimensional basic model used for generating the three-dimensional model in advance;
Inputting attribute information of the building to be generated;
Selecting the basic model according to the attribute information;
Deforming and arranging the basic model based on the shape of the building area and the building height data; and
In order from the lowest level, setting the target level to be modeled,
A three-dimensional modeling method comprising: a step of modeling a target hierarchy by specifying a part having building height data equal to or higher than the height of the target hierarchy as a region belonging to the target hierarchy. A 3D modeling device for generating 3D electronic data of a building,
A two-dimensional map input unit for inputting two-dimensional electronic map data including a building to be generated;
A height data input unit for inputting height data measured for a number of points in the region including the building to be generated;
An extraction unit for identifying a building area of the building from the two-dimensional data, and extracting height data existing in the building area as building height data;
A model generation unit that generates a three-dimensional model of the building by specifying the height of the building based on the building height data;
A smoothing processing unit for smoothing each wall surface of the generated three-dimensional model with a plane;
I have a,
The building is a building having a hierarchical structure,
The model generation unit specifies the height of each level of the building based on the frequency distribution of the building height data,
The height data is data in which the height is measured in a predetermined mesh unit defined on the ground surface,
The three-dimensional modeling apparatus , wherein the smoothing processing unit specifies a corner mesh positioned at a corner of the building and performs the smoothing by a vertical plane passing through representative points of the two corner meshes . A computer program for generating three-dimensional electronic data of a building,
A function for inputting two-dimensional electronic map data including a building to be generated;
A function of inputting height data measured for a large number of points in an area including the building to be generated;
A function of identifying a building area of the building from the two-dimensional data, and extracting height data existing in the building area as building height data;
A model generation function for generating a three-dimensional model of the building by specifying the height of the building based on the building height data;
A smoothing function for smoothing each wall surface of the generated three-dimensional model in a plane;
Is a computer program for realizing by a computer ,
The building is a building having a hierarchical structure,
The model generation function specifies the height of each level of the building based on the frequency distribution of the building height data,
The height data is data in which the height is measured in a predetermined mesh unit defined on the ground surface,
The smoothing processing function is a computer program that identifies a corner mesh positioned at a corner of the building and performs the smoothing by a vertical plane that passes through representative points of the two corner meshes .

Images