JP2006323608A - Apparatus and method for creating model of group of three-dimensional structure and system for creating three-dimensional model - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To create a three-dimensional model that closely resembles the real image in a short period of time. <P>SOLUTION: A point-group data mesh classification part 12 classifies point-group data 1 into the form of meshes within a predetermined range. A flat analysis part 13 extracts flat shape information by analyzing the distribution of the point-group data 1 for each of the meshes into which the data are classified. A height-direction distribution grouping part 14 extracts the meshes having the shape information and groups the meshes for every height direction. A mesh pattern straight-line extracting part 17 analyzes the patterns of the meshes for each of the groups and extracts information on straight lines connecting the meshes. An attribute information adding/creating part 18 adds structure attribute information to the information on the straight line for each of the groups extracted and creates the boundary line of a structure. Then a three-dimensional model creating part 20 creates a three-dimensional model of a group of structures by means of the boundary line of each of the groups created, the point-group data 1, and the pushing function of CAD. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a three-dimensional structure group model creation device, a three-dimensional structure group model creation method, and a three-dimensional model creation that generate a three-dimensional structure group in an urban area or the like using three-dimensional point cloud data obtained by aviation laser measurement. About the system.

  In recent years, aviation laser measurement has been put into practical use, where a laser scanner is installed in an aircraft and altitude data is acquired over a wide range, and a three-dimensional space can be easily obtained. Aviation laser measurement is equipped with a scanning laser range finder, GPS (Global Positioning System) and IMU (Inertial Measurement System) on an aircraft, and the irradiated laser beam is controlled by controlling the position and orientation of laser irradiation. This is a measurement method in which the time until the light is reflected and returned is measured and position information (latitude, longitude, altitude) is acquired as digital data.

  The data obtained by this aviation laser measurement is expected to be used in various fields such as disaster prevention, GIS (Geographical Information System), and city planning. In particular, research on the effective use of laser measurement data is considered to contribute to the development of software support, as major disasters such as typhoons, torrential rains, and earthquakes occur frequently and awareness of disaster prevention is increasing.

  As data using such data obtained by aviation laser measurement, in Patent Document 1, in determining an adjacent vertex adjacent to a determined triangle, a side of interest of the triangle is defined on an extension plane of the determined triangle. Assuming that the target edge is a vertex candidate adjacent to the target edge as a vertex candidate, and applying the three-dimensional KL expansion method to the point cloud data in the vicinity of the vertex candidate, the first, second, and First, second and third eigenvectors corresponding to the third main direction are obtained, and upper and lower limits of the inner product of the first and second eigenvectors corresponding to the first and second main directions and the point group are used. A three-dimensional polygon data creation method has been proposed in which polygon data on the surface of an object is acquired in a good and stable manner by determining adjacent vertices.

  Further, in Patent Document 2, each stereo aerial photograph image is divided into regions, thereby obtaining an image region with uniform colors, and stereo matching using an elevation value of a three-dimensional point cloud of a laser scanner overlapping each color region as an initial value. When the calculation speed of the altitude value obtained as a result is almost the same as the initial value, the altitude value of the calculation result is given to the area, a DSM is generated, and then the DSM obtained above is We have proposed a method for generating a high-precision city model that recognizes horizontal regions by region segmentation, performs straight line extraction in the peripheral region on the stereo image corresponding to each region, and polygonizes the result.

In Patent Document 3, the first stage of reading the three-dimensional point cloud data obtained by the three-dimensional measurement into a computer and separating it into the point cloud data of the terrain portion and the point cloud data of the feature portion; The second stage of creating contour polygons by computer loading and image processing, and the point cloud data of the terrain part and the point part data of feature parts separated in the first stage, and the cubic of the urban area from the contour polygons created in the second stage A third step of creating an original model, extending the contour polygon to the three-dimensional geographic coordinate space, automatically determining the top surface form and feature type of the feature, and determining the three-dimensional geographic coordinate A method to create a former urban space model is proposed.
JP 2003-346182 A JP 2003-323640 A JP 2002-074323 A

  However, in the above-described Patent Document 1, it is possible to automatically create a triangular surface and a solid from X, Y, and Z point cloud data, but the point, surface, and solid are handled in mathematics. Since it is simple geometric information and does not have structure attribute information indicating what the point / surface represents, it is difficult to obtain a three-dimensional model that is close to a real image of a three-dimensional structure or the like in an urban area. There was a problem that there was.

  On the other hand, as in Patent Documents 2 and 3, the structure of the three-dimensional structure from the aerial photograph image is obtained by combining the aerial photograph image and the three-dimensional point cloud data obtained by the three-dimensional measurement by the laser scanner. Object attribute information can be obtained. For example, there is a difference between the size and imaging accuracy of an aerial photo image and the measurement range and measurement accuracy of three-dimensional point cloud data, and processing for accurately correcting these differences Since this is necessary, there has been a problem that the creation time of a three-dimensional structure or the like in an urban area is prolonged.

  The present invention has been made in view of such circumstances, and provides a three-dimensional structure group model creation device, a three-dimensional structure group model creation method, and a three-dimensional model creation system that can solve the above-described problems. With the goal.

The three-dimensional structure group model creation device of the present invention is a three-dimensional structure group model creation device that creates a three-dimensional model of a structure group using three-dimensional point cloud data obtained by aviation laser measurement, Point cloud data classifying means for classifying the point cloud data into a mesh within a predetermined range in the horizontal direction, and analyzing the distribution of the point cloud data for each classified mesh and determining whether or not it is flat Flat analysis means for extracting information, and a height for grouping the meshes based on the point cloud data in the extracted mesh for each height direction, extracting a mesh having the shape information made flat. The direction distribution grouping means and the mesh pattern pattern for analyzing the grouped mesh pattern for each group and extracting straight line information connecting each mesh for each group. An attribute information addition creating means for adding structure attribute information to the extracted straight line information for each group and creating a boundary line of the structure; and the boundary line for each created group; 3D model creation means for creating a 3D model of a structure group by the point group data and a CAD push function.
The structure attribute information may be data in a height direction in the point cloud data.
Further, when grouping by the height direction distribution grouping means is performed, a height direction group determination for determining that the meshes having the point cloud data that fall within a predetermined range in the height direction as one group In order to obtain only the meshes of the group to be analyzed from the meshes grouped by the height direction distribution grouping means when the straight line information is extracted by the means and the mesh pattern straight line extracting means, Group removal means for removing the mesh of the group may be provided.
Further, the flat analysis means can determine that a mesh in which the occupancy rate of the point cloud data is 100% within a reference value including an error due to the aviation laser measurement is flat.
Further, the mesh pattern straight line extracting means may perform a search based on a basic pattern divided according to the presence of data around the mesh to be analyzed when analyzing the mesh pattern. it can.
In addition, it is possible to provide group attribute determination means for determining whether or not the boundary line for each group created by the attribute information addition creation means is within the same region in the height direction.
Further, the three-dimensional model creation means can synthesize the point cloud data that is a low altitude portion with the created three-dimensional model of the structure group.
The three-dimensional structure group model creation method of the present invention is a three-dimensional structure group model creation method for creating a three-dimensional model of a structure group using three-dimensional point cloud data obtained by aviation laser measurement, The point cloud data is classified into a mesh shape within a predetermined range in the horizontal direction, the distribution of the point cloud data is analyzed for each classified mesh, and shape information as to whether or not it is flat is extracted, and it is made flat. The mesh having the shape information is extracted, the meshes are grouped based on the point cloud data in the extracted mesh for each height direction, and the grouped mesh pattern is analyzed for each group. The straight line information connecting each mesh for each group is extracted, and the structure attribute information is added to the extracted straight line information for each group, and the boundary line of the structure is created. And the boundary line of each loop, and the point group data, by the extrusion features CAD, characterized by creating a three-dimensional model of the structure group.
The structure attribute information may be data in a height direction in the point cloud data.
Further, when the grouping is performed, the meshes having the point cloud data that falls within a predetermined range in the height direction are determined to be one group, and when the straight line information is extracted, the grouping is performed. In order to obtain only the meshes of the group to be analyzed from the meshes obtained, the meshes of other groups can be removed.
Further, it is possible to determine that a mesh in which the occupation rate of the point cloud data is 100% within a reference value including an error due to the aviation laser measurement is flat.
Further, when analyzing the mesh pattern, it is possible to perform a search based on a basic pattern divided into patterns according to the presence of data around the mesh to be analyzed.
Further, it can be determined whether or not the boundary line created for each group is in the same region in the height direction.
In addition, the point cloud data to be a low altitude portion can be synthesized with the created three-dimensional model of the structure group.
The three-dimensional model creation system of the present invention includes the three-dimensional structure group model creation device according to any one of claims 8 to 14.
In the present invention, the three-dimensional point cloud data obtained by the aviation laser measurement is classified into a mesh within a predetermined range in the horizontal direction by the point cloud data mesh classification means, and the mesh classified by the flat analysis means Analyzing the distribution of point cloud data every time, extracting the shape information as to whether or not it is flat, extracting the mesh having the flat shape information by the height direction distribution grouping means, and for each height direction Next, the meshes are grouped based on the point cloud data in the extracted mesh, and the mesh pattern straight line extraction means analyzes the grouped mesh patterns for each group, and the straight line information connecting each mesh for each group is obtained. Extract and add structure attribute information to the extracted straight line information for each group by the attribute information addition creation means to create the boundary line of the structure Then, since the 3D model creation means creates the 3D model of the structure group by the created boundary line for each group, the point cloud data, and the CAD push function, an aerial photograph image is used. 3D structure group can be expressed by structure attribute information, and the difference between the size and imaging accuracy of the aerial photo image and the measurement range and measurement accuracy of the three-dimensional point cloud data can be accurately corrected. No processing is required.

  According to the present invention, a three-dimensional structure group can be represented by structure attribute information without using an aerial photograph image, and the size and imaging accuracy of the aerial photograph image and the measurement range of three-dimensional point cloud data In addition, since a process for accurately correcting a difference in measurement accuracy is not required, a three-dimensional model close to a real image can be created in a short time.

  In the present embodiment, the point cloud data mesh classification means classifies the three-dimensional point cloud data obtained by the aviation laser measurement into a mesh shape within a predetermined range in the horizontal direction, and the flat analysis means classifies the data. Analyzes the distribution of point cloud data for each mesh, extracts the shape information whether or not it is flat, and extracts the mesh with the flat shape information by the height direction distribution grouping means, the height direction Each time, the meshes are grouped based on the point cloud data in the extracted mesh, and the mesh pattern straight line extraction means analyzes the grouped mesh pattern for each group, and the straight line information connecting each mesh for each group The attribute attribute information is added to the extracted straight line information for each group by the attribute information addition creation means, and the boundary line of the structure is added. Then, the 3D model creation means creates a 3D model of the structure group by the created boundary line for each group, the point cloud data, and the push function of CAD. 3D structure group can be expressed by structure attribute information without using it, and the difference between the size and imaging accuracy of aerial photo images and the measurement range and measurement accuracy of 3D point cloud data can be corrected with high accuracy. By eliminating the need for processing, it is possible to create a three-dimensional model close to live action in a short time.

Embodiments of the present invention will be described below.
FIG. 1 is a block diagram showing an embodiment of a three-dimensional structure group model creating apparatus of the present invention, and FIGS. 2 to 19 illustrate a three-dimensional structure group model creating method by the three-dimensional structure group model creating apparatus of FIG. It is a figure for doing.

  As shown in FIG. 1, the three-dimensional structure group model creation device 10 reads point cloud data 1 that is three-dimensional laser measurement data obtained by aviation laser measurement, and performs a three-dimensional structure by analyzing the point cloud data 1. It generates a model of a group of objects.

  The three-dimensional structure group model creation device 10 is mounted on an information processing device such as a personal computer, and can be configured by either hardware or software.

  The three-dimensional structure group model creation device 10 includes a CPU 10a, a point group data reading unit 11, a point group data mesh classification unit 12, a flat analysis unit 13, a height direction distribution grouping unit 14, a height direction group determination unit 15, A group removal unit 16, a mesh pattern straight line extraction unit 17, an attribute information addition creation unit 18, a group attribute determination unit 19, and a three-dimensional model creation unit 20 are provided. These are connected via a data bus.

  The CPU 10a controls the operation of each unit based on a predetermined control program.

  The point cloud data reading unit 11 reads point cloud data 1 which is three-dimensional laser measurement data obtained by aviation laser measurement. The outline of the laser measurement data will be described later.

  The point cloud data mesh classification unit 12 converts the point cloud data 1 read by the point cloud data reading unit 11 into a mesh in a range of, for example, 5.0 m in the horizontal x and y directions, as will be described later. Classify. This is because the point cloud data 1 is random data, so that data analysis is facilitated.

  The flat analysis unit 13 checks how the point cloud data 1 classified into the mesh shape by the point cloud data mesh classification unit 12 is distributed in the height direction (z direction). The entire data is scanned, and flat (F) shape information is extracted for each mesh as described later.

  The height direction distribution grouping unit 14 extracts a mesh in which the occupancy rate within the reference value of the data in the mesh is 100% from the flat (F) shape information for each mesh extracted by the analysis by the flat analysis unit 13. Then, the extracted meshes are grouped for each height direction of the data. The occupation ratio in the reference value will be described later.

  The height direction group determination unit 15 includes data having data in which an interval in a predetermined height direction (z direction) is within a predetermined range when the meshes by the height direction distribution grouping unit 14 are grouped. Are determined as one group.

  The group removal unit 16 removes the meshes of other groups in order to obtain only the meshes of the group to be analyzed from the meshes grouped by the height direction distribution grouping unit 14.

  The mesh pattern straight line extraction unit 17 analyzes the mesh pattern left after being removed by the group removal unit 16 for each group, and extracts straight line information connecting the meshes for each group. The extraction of the straight line information of the mesh pattern will be described later.

  The attribute information addition creating unit 18 adds the height direction data (z), which is the structure attribute information, to the straight line information for each group extracted by the mesh pattern straight line extracting unit 17, and the boundary line of the structure or the like is added. create. The height direction data (z) is obtained from the point cloud data 1.

  The group attribute determination unit 19 determines whether the boundary line of the structure or the like created by the attribute information addition creation unit 18 is in the same region in the height direction (z direction). Thereby, for example, it can be determined whether or not the building is a building protruding on the roof of the same building.

  The three-dimensional model creation unit 20 uses the determination result by the group attribute determination unit 19, the point cloud data 1 that is x, y, z laser measurement data, and the CAD push-out function to generate a three-dimensional structure or the like. Create a model. In addition, it is possible to faithfully represent the actual object by synthesizing Ground Data for the created three-dimensional model such as a structure, which will be described later.

Next, a method for creating a three-dimensional structure group model will be described.
First, as shown in FIG. 2, the point cloud data reading unit 11 reads point cloud data 1 which is three-dimensional laser measurement data obtained by aerial laser measurement (step S1).

  Here, aviation laser measurement, for example, as shown in FIG. 3, is equipped with a scanning laser rangefinder, GPS (Global Positioning System) and IMU (Inertial Measurement System) on the aircraft, and the position and orientation of laser irradiation This is a measurement method in which the time until the irradiated laser light is reflected and returned to the ground surface is measured while controlling and the position information (latitude, longitude, altitude) is acquired as digital data.

  The laser measurement data obtained in this manner includes coordinate data in the x and y directions on the horizontal plane and coordinate data in the z direction in the height direction from the altitude of the aircraft and the return time of the laser light, etc. Random point cloud data 1 is assumed.

  As an outline of the point cloud data 1, for example, as shown in FIG. Fig. 4 is a graph of sample data of laser measurement data. The horizontal axis shows the altitude (m) as the height, and the vertical axis shows the point occupation rate (%) according to the altitude according to the height. ing.

  4 (a) shows an urban area, FIG. 4 (b) shows a residential land area, FIG. 4 (c) shows an urban suburban area, FIG. 4 (d) shows a field area, and FIG. 4 (d) shows a mountain area. ing.

  Here, it can be seen that in the urban area and the residential land area shown in FIGS. 4A and 4B, data obtained by laser measurement is distributed in a certain height direction. From this, it can be seen that there is data of roads represented by low elevations and the like and buildings such as middle elevations in the data. In addition, as a difference between an urban area and a residential land part, it can compare by the difference of an altitude range. That is, there are relatively high buildings in urban areas, and there are relatively low buildings and houses in residential areas.

  Moreover, although the occupation ratio protrudes in some altitude parts, it can be determined that this is a ground surface part such as a road or a park. The degree of protrusion of the ground surface portion indicates the degree of existence of a building or the like in the data. Note that the presence rate is high when the degree of protrusion is small, and the presence rate is low when it is large.

  In the city suburb of FIG. 4C, the situation is the same as that of the residential land, but the altitude range is slightly narrowed. Also in this data, the occupancy rate is prominent at some altitudes, and it can be determined that it is a road or a field portion. 4D, the altitude range is narrower, and it can be seen that the altitude difference is small in the flat area.

  In the mountain area of FIG. 4 (e), data is distributed in a wide altitude range from the low altitude part to the high altitude part, and an altitude part with a prominent occupation rate as seen in other data is not seen. In this way, it is considered that a rough judgment can be made as to what kind of terrain the area is by comparing the occupancy rate by area and by altitude.

  In this way, the fact that the distribution situation differs depending on the region means that a three-dimensional structure group model is generated for each region by using a data processing method suitable for the terrain after considering the characteristics. Can be said to be possible.

  Next, the point cloud data 1 read by the point cloud data reading unit 11 is, for example, within a range of 5.0 m in the horizontal x and y directions by the point cloud data mesh classification unit 12 as shown in FIG. (Step S2).

  This is because the point cloud data 1 is random data, so that data analysis is facilitated. Here, the mesh range in the x and y directions may be determined according to the number of points by laser measurement, and of course may be outside the range of 5.0 m.

  Here, when the mesh classification for all the point cloud data 1 is completed (step S3), the point cloud data 1 classified in the mesh shape is subjected to the height direction (z direction) by the flat analysis unit 13. As shown in FIG. 6, the entire data is scanned while checking how the distribution is performed, and flat (F) shape information is extracted for each mesh as shown in FIG. (Step S4).

  Here, when extracting flat (F) shape information, as shown in FIGS. 8 and 9, the average value of the data in the mesh is set to ± 0.20 m as a reference value, and the occupation of the data in the reference value is as follows. The mesh when the rate is 100% is determined as a flat (F) shape. Here, the average value is a value obtained by averaging the data in the mesh in the height direction. Further, the gradient of each data is determined to be flat (F) up to 2%.

  The reason for flat (F) up to 2% slope is that the average single slope used in road crossing design is 2%, and flat (F) if the slope is 2% even in the longitudinal direction. This is because it was judged that it would be acceptable. In setting the reference value ± 0.20 m, 0.15 m as an error at the time of laser measurement is taken into account.

  When the extraction of flat (F) shape information for all the meshes is completed (step S5), the flat (F) shape extracted by the analysis by the flat analysis unit 13 by the height direction distribution grouping unit 14 As shown in FIG. 10, meshes in which the occupation ratio in the reference value of the data in the mesh is 100% are extracted from the information, and the extracted meshes are grouped for each height direction (step S <b> 6).

  In the grouping, the height direction group determination unit 15 determines to make one group when the distance from each data aggregate is 0.5 m or more, for example. Here, 0.5 m, which is the interval between the data aggregates, is merely an example, and is not limited to this value.

  An example of the result of grouping is shown in FIG. FIG. 11 shows a case where G01 to G12 are grouped. Here, the group G01 having the largest number of meshes corresponds to an aerial photograph image as shown in FIG. 12, for example. As can be seen from the results of these comparisons, the group G01 can be determined to be data other than the ground surface portion such as a road, so-called buildings.

  This means that the portion where the height direction is lower than the group G01 is ground data, and conversely, the portion where there is no data is a building. Further, by using a blank portion that is a portion without the data, it is possible to extract a contour of a building or the like as will be described later.

  The maximum altitude value of group G01 is 9.903m, and this value is divided into the low altitude part (Ground Data) and the high altitude part (Top Data) as the boundary line that divides group G01 and group G02. Is shown in FIG. As can be seen from FIG. 13, the outline of the building can be extracted as described later by using the blank portion of Ground Data.

  When the grouping is completed (step S7), the group removal unit 16 obtains only the meshes of the group to be analyzed from the meshes grouped by the height direction distribution grouping unit 14, so that another group is obtained. Are removed (step S8).

  Then, the mesh pattern straight line extraction unit 17 extracts the mesh pattern straight line information from the meshes of each group removed by the group removal unit 16 (step S9).

  Here, when extracting a straight line of the mesh pattern, straight line information is extracted based on the attribute value given to each mesh grouped by the height direction distribution grouping unit 14.

  That is, as shown in FIG. 14A, it is assumed that patterns 1 to 9 that are basic patterns are set for the pattern in the initial state. These patterns 1 to 9 are divided into patterns according to the presence of surrounding data, with the hatched square portion to be analyzed as the center.

  Here, first, as shown in FIG. 14B, the pattern search is performed in order from the upper left of the pattern in the initial state, for example. As shown in FIG. 14C, the pattern search is performed by searching for data of surrounding meshes centering on the mesh a. That is, data of 8 meshes are searched centering on the mesh a.

  In this way, as shown in FIG. 14B, information indicating the pattern obtained by the pattern search of each mesh is added as an attribute value. FIG. 14B shows a case where the pattern 1 is searched and the information of the pattern 1 is added as an attribute value.

  FIG. 15A shows a case where the pattern 9 is searched and the information of the pattern 9 is added as an attribute value. FIG. 15B shows a case where the pattern 2 is searched and the information of the pattern 2 is added as an attribute value.

  When all the pattern searches are completed in this way, information on all patterns is added as attribute values as shown in FIG.

  Then, in the case of pattern 1 → search for the right and connect pattern 2 or 5, in the case of pattern 2 → search for the bottom and connect pattern 3 or 6, in the case of pattern 3 → search for the left and pattern 4 or 7 In the case of the pattern 4 → search for the top and connect the pattern 1 or 8, in the case of the pattern 5 → search for the top and connect the pattern 1 or 8, in the case of the pattern 6 → the right search, the pattern 2 or In the case of pattern 7 linking 5, the search is performed below, and in the case of pattern 8 linking the pattern 3 or 6, the straight line information of the mesh pattern in which pattern 4 or 7 is coupled is retrieved.

  The extraction of the mesh pattern straight line information is performed for each mesh grouped by the height direction distribution grouping unit 14. Here, the extracted straight line information is expressed as shown in FIG. 16, for example. The extraction of the straight line information of the mesh pattern may be set based on a predetermined rule.

  Next, when straight line information is extracted for each grouped mesh, the attribute information addition creating unit 18 adds height direction data (z) as structure attribute information to the straight line information for each group. Then, a boundary line such as a structure is created (step S10).

  Next, the group attribute determining unit 19 determines whether the boundary line of the structure or the like created by the attribute information addition creating unit 18 is in the same region in the height direction (z direction) (step) S11). Thereby, for example, it can be determined whether or not the building is a building protruding on the roof of the same building.

  When the determination of the boundary line of the structure or the like is completed as described above (step S12), the determination result by the group attribute determination unit 19 and the x, y, z laser measurement data are obtained by the three-dimensional model creation unit 20. A three-dimensional model such as a structure as shown in FIG. 17 is created by extruding, which is a CAD function, using the point cloud data 1 (step S13).

  The above-described ground data may be further synthesized by the three-dimensional model creation unit 20 with respect to the created three-dimensional model such as a structure. In this case, as shown in FIG. A three-dimensional model close to a three-dimensional live action can be obtained.

  Further, when texture mapping with an aerial photograph image is performed by the three-dimensional model creation unit 20, a three-dimensional model that is closer to a real image as shown in FIG. 19 is obtained.

  Thus, in this embodiment, the point cloud data mesh classification unit 12 classifies the three-dimensional point cloud data 1 obtained by the aviation laser measurement into a mesh within a predetermined range in the horizontal direction, and performs flat analysis. The unit 13 analyzes the distribution of the point cloud data 1 for each classified mesh, extracts shape information as to whether or not it is flat, and the height direction distribution grouping unit 14 converts the flat shape information. The meshes are extracted and grouped for each height direction based on the point cloud data 1 in the extracted mesh, and the mesh pattern straight line extraction unit 17 analyzes the grouped mesh patterns for each group. The straight line information connecting the meshes for each group is extracted, and the attribute information addition creating unit 18 extracts the structure attribute information from the extracted straight line information for each group. Is added and the boundary line of the structure is created, the 3D model creation unit 20 creates the 3D structure group by the created boundary line for each group, the point cloud data 1, and the CAD push function. Since a model is created, a three-dimensional structure group can be expressed by structure attribute information without using an aerial photograph image, and the size and imaging accuracy of the aerial photograph image and a three-dimensional point cloud can be expressed. Since it is possible to eliminate the need to accurately correct the difference in the measurement range and measurement accuracy of the data, it is possible to create a three-dimensional model close to a real image in a short time.

  The present invention is also applicable to a three-dimensional model creation system that creates a three-dimensional model other than a group of structures.

It is a block diagram showing one embodiment of a solid structure group model creation device of the present invention. It is a flowchart for demonstrating the three-dimensional structure group model creation method by the three-dimensional structure group model production apparatus of FIG. It is a figure for demonstrating the outline | summary of the aviation laser measurement for obtaining the point cloud data by the three-dimensional structure group model creation apparatus of FIG. It is a figure for demonstrating the outline | summary of the point cloud data by the three-dimensional structure group model production apparatus of FIG. It is a figure for demonstrating the case where a point cloud data mesh-shaped classification | category part of FIG. 1 classifies point cloud data into a mesh shape. It is a figure for demonstrating the case where flat (F) shape information is extracted by the flat analysis part 13 of FIG. It is a figure for demonstrating the case where flat (F) shape information is extracted by the flat analysis part 13 of FIG. It is a figure for demonstrating the case where flat (F) shape information is extracted by the flat analysis part 13 of FIG. It is a figure for demonstrating the case where flat (F) shape information is extracted by the flat analysis part 13 of FIG. It is a figure for demonstrating the case where a mesh is grouped for every height direction by the height direction distribution grouping part of FIG. It is a figure for demonstrating an example of a result when a mesh is grouped for every height direction by the height direction distribution grouping part of FIG. FIG. 12 is a diagram for explaining a comparison between a group G01 having the largest number of meshes and an aerial photograph image in the example of the result of FIG. It is a figure for demonstrating the result divided | segmented into the low altitude part (Ground Data) and the high altitude part (Top Data) from an example of the result of FIG. It is a figure for demonstrating the case where the straight line information of a mesh pattern is extracted by the mesh pattern straight line extraction part of FIG. It is a figure for demonstrating the case where the straight line information of a mesh pattern is extracted by the mesh pattern straight line extraction part of FIG. It is a figure for demonstrating the case where the straight line information of the mesh pattern extracted by the mesh pattern straight line extraction part of FIG. 1 is expressed. It is a figure for demonstrating the case where 3D models, such as a structure, are created by the 3D model creation part of FIG. It is a figure for demonstrating the case where Ground Data is synthesize | combined with the three-dimensional model of FIG. It is a figure for demonstrating the case where the texture mapping with an aerial photograph image is performed with respect to the three-dimensional model of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Point cloud data 10 3D structure group model creation apparatus 11 Point cloud data reading part 12 Point cloud data mesh classification | category part (Point cloud data mesh classification means)
13 Flat analysis unit (Flat analysis means)
14 Height direction distribution grouping section (height direction distribution grouping means)
15 Height direction group determination unit (height direction group determination means)
16 Group removal section (group removal means)
17 Mesh pattern straight line extraction unit (mesh pattern straight line extraction means)
18 Attribute information addition creation part (attribute information addition creation means)
19 Group attribute determination unit (group attribute determination means)
20 3D model creation unit (3D model creation means)

Claims (15)

A three-dimensional structure group model creation device for creating a three-dimensional model of a structure group using three-dimensional point cloud data obtained by aviation laser measurement,
Point cloud data mesh classification means for classifying the point cloud data into a mesh within a predetermined range in the horizontal direction;
Flat analysis means for analyzing the distribution of the point cloud data for each classified mesh and extracting shape information as to whether or not it is flat;
Height direction distribution grouping means for extracting the mesh having the shape information made flat and grouping the mesh based on the point cloud data in the extracted mesh for each height direction;
Analyzing the grouped mesh patterns for each group, mesh pattern straight line extracting means for extracting straight line information connecting each mesh for each group,
Attribute information addition creating means for adding structure attribute information to the extracted straight line information for each group and creating a boundary line of the structure;
A three-dimensional structure characterized by comprising a three-dimensional model creating means for creating a three-dimensional model of a structure group by the boundary line, the point cloud data, and a CAD push-out function for each created group. Group model generator.   The three-dimensional structure group model creation device according to claim 1, wherein the structure attribute information is data in a height direction in the point group data. A height direction group determination unit that determines that the meshes having the point cloud data that fall within a predetermined range in the height direction are grouped when the height direction distribution grouping unit performs grouping; ,
When the straight line information is extracted by the mesh pattern straight line extraction unit, in order to obtain only the meshes of the group to be analyzed from the meshes grouped by the height direction distribution grouping unit, the meshes of other groups The three-dimensional structure group model creating device according to claim 1, further comprising: a group removing unit that removes the structure.   The flat analysis means determines that a mesh having an occupancy ratio of the point cloud data within a reference value including an error due to the aviation laser measurement as 100% is flat. The three-dimensional structure group model creation device according to any one of the above.   The mesh pattern straight line extraction means, when analyzing the mesh pattern, performs a search based on a basic pattern divided according to the presence of data around the mesh to be analyzed. The three-dimensional structure group model creation apparatus in any one of 1-4.   6. A group attribute determining means for determining whether or not the boundary line for each group created by the attribute information addition creating means is within the same region in the height direction. The three-dimensional structure group model creation device according to any one of the above.   The three-dimensional model according to any one of claims 1 to 6, wherein the three-dimensional model creating means synthesizes the point cloud data to be a low altitude portion with the created three-dimensional model of the structure group. Structure group model creation device. A three-dimensional structure group model creation method for creating a three-dimensional model of a structure group using three-dimensional point cloud data obtained by aeronautical laser measurement,
The point cloud data is classified into a mesh shape in a predetermined range in the horizontal direction,
Analyzing the distribution of the point cloud data for each classified mesh, extracting shape information as to whether it is flat,
Extract the mesh having the shape information made flat, group the mesh based on the point cloud data in the extracted mesh for each height direction,
Analyzing the grouped mesh pattern for each group, extracting straight line information connecting each mesh for each group,
Add structure attribute information to the extracted straight line information for each group, create a boundary line of the structure,
A three-dimensional structure group model generation method characterized in that a three-dimensional model of a structure group is created by the boundary line, the point group data, and the CAD push-out function for each created group.   The three-dimensional structure group model creation method according to claim 8, wherein the structure attribute information is data in a height direction in the point group data. When the grouping is performed, the mesh having the point cloud data that falls within a predetermined range in the height direction is determined to be one group,
The mesh of another group is removed to obtain only the mesh of the group to be analyzed from the grouped mesh when the straight line information is extracted. The three-dimensional structure group model creation method described.   The mesh with which the occupation rate of the point cloud data within a reference value including an error due to the aviation laser measurement is 100% is determined to be flat. 3D structure group model creation method.   12. The analysis according to claim 8, wherein when the mesh pattern is analyzed, the search is performed based on a basic pattern divided according to the presence of data around the mesh to be analyzed. 3D structure group model creation method.   The method for creating a three-dimensional structure group model according to any one of claims 8 to 12, wherein it is determined whether or not the boundary line created for each group is in the same region in the height direction. .   The method for creating a three-dimensional structure group model according to any one of claims 8 to 13, wherein the point group data to be a low altitude portion is synthesized with the created three-dimensional model of the structure group.   A three-dimensional model creation system comprising the three-dimensional structure group model creation device according to claim 8.

Images