JP2002150270A - Three-dimensional model formation method and device, and record medium recording execution program of the method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a three-dimensional model formation method and device calculating a three-dimensional shape from range data obtained using laser distance measuring equipment. SOLUTION: This three-dimensional model formation device receives a processing range by a command of an operator, calculates a height difference between an object point and an adjoining point from distance information obtained from the range data, and calculates a feature quantity related to an edge having such a shape that the height difference is formed into a certain specific pattern. The object point included in a range surrounded therewith is retrieved from the feature quantity related to the edge having the calculated pattern to calculate the feature quantity related to the range of the object point. The shape is extracted from the feature quantity related to the calculated edge and the feature quantity related to the range of the object point. A gradient of the calculated range is calculated from the range data in the range, a closed range is divided based on the feature quantity related to the calculated gradient to perform shaping, and the three-dimensional model is calculated from the shape to be stored in the three-dimensional database.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a geographic information system (GIS).
The present invention relates to a method for creating three-dimensional graphic data used in, for example, on-system, and particularly relates to a method for calculating a three-dimensional shape from range data and generating a three-dimensional model.

[0002]

2. Description of the Related Art In recent years, with the complexity of cities, high-rise buildings,
In the underground shopping malls, overpasses, water and sewage systems, electricity, telephones, gas, and other facilities, urban hierarchies that extend to the ground and underground are being advanced. There has been an increasing demand for using a three-dimensional map having a higher expression ability instead of a conventional two-dimensional map.

With respect to such a three-dimensional map generation technique, a technique for generating a three-dimensional map by adding height information to a figure of a two-dimensional map has been developed. For example, the method described in Japanese Patent Application Laid-Open No. 2000-112345 “Method of creating a three-dimensional map” uses floor information and room number information extracted from address information of attribute information assigned to a figure of a two-dimensional map. , Height information.

On the other hand, a technique of measuring height information by using a laser distance measuring device or the like has been developed so that height information with higher accuracy can be used. For example, JP-A-11-508359 or JP-A-8-2172.
The device described in No. 9 has been developed, and height information can be acquired relatively easily.

[0005]

However, in the prior art for generating a three-dimensional map by adding height information to a figure of a two-dimensional map, the height information is determined by the floor information and the floor number of the room number information. It is difficult to provide highly accurate height information because it is calculated using an appropriate height per floor.

Further, height information measured using a laser distance measuring device or the like generally contains many measurement errors. Due to the influence of the measurement error, the height information to be given to the figure of the two-dimensional map is shifted from information such as a two-dimensional map created by accurate surveying by another method. I will. Therefore, if this height information and the two-dimensional map data are used as they are, three-dimensional map data having height information different from the actual height will be generated, and it is very difficult to use it practically. Has become.

[0007] Above all, in these technologies, the underlying 2
At present, three-dimensional map data cannot be generated without such data because three-dimensional map data is required.

As described above, it is very difficult to generate three-dimensional map data when there is no two-dimensional map data in the related art. Therefore, it is necessary to generate three-dimensional map data manually. This can take a considerable amount of time and experience,
Further, automation is strongly desired.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems and technical backgrounds of the related art, and has been made in consideration of a three-dimensional shape based on range data obtained by using a laser distance measuring device or the like. It is an object to provide a method and apparatus for calculating and generating a three-dimensional model.

[0010]

A three-dimensional model generating method according to the present invention calculates a three-dimensional shape from range data,
A method of generating a three-dimensional model, comprising calculating a height difference between a target point and a point adjacent to the target point from distance information obtained from range data, and calculating the height difference between the target point and a point adjacent to the target point. An edge feature value calculation processing procedure for calculating a feature value related to an edge, and a feature point related to the edge of the calculated shape are searched for a target point included in a region surrounded by the edge. A region feature amount calculation processing procedure for calculating a feature amount, a shape extraction processing procedure for extracting a shape from the calculated feature amount regarding the edge of the shape, and the calculated feature amount regarding the area of the target point; For a closed region having a shape obtained, a gradient feature value calculation processing procedure of calculating a feature value related to a gradient from range data in the closed region, and a feature value related to the calculated gradient Performs division of the serial closed region, characterized in that it comprises a shape shaping processing procedure for generating a three-dimensional model to calculate the three-dimensional shape from the extracted shape.

Alternatively, in the above-described three-dimensional model generation method, the edge feature amount calculation processing procedure may include the steps of calculating a sequence of four consecutive points including a target point from distance information obtained from range data.
The method is characterized in that a height difference between points is calculated, and when the height difference has a certain specific pattern, at least a pattern and a position in that direction are calculated as a feature amount.

Alternatively, in the above-described three-dimensional model generation method, the region feature value calculation processing step includes, based on the feature value relating to the edge of the shape calculated in the edge feature value calculation process, a point having the feature value. It is characterized by searching for a target point included in the enclosed area.

Alternatively, in the above-described three-dimensional model generation method, in the shape extraction processing procedure, a set of points having a feature value having the same height difference pattern in the coordinate axis direction such as the X direction and the Y direction is defined as a unit. And extracting a straight line forming the same shape from a straight line group calculated using the feature amount regarding the area of the target point calculated in the area feature amount calculation processing procedure. Features.

Alternatively, in the above-described three-dimensional model generation method, in the gradient feature value calculation processing procedure, a region constituting the shape calculated in the shape extraction processing procedure is calculated based on distance information obtained from range data. 2 consecutive
The method is characterized in that the height difference between points is calculated, and the gradients in the coordinate axis directions such as the X direction and the Y direction are classified into flat, climbing, and hollow.

Alternatively, in the above-described three-dimensional model generation method, in the shape shaping processing procedure, the gradient calculated in the gradient feature value calculation processing procedure is classified, and the classified gradient has the same gradient. Calculating the straight line that forms the region, and integrating the straight line and the straight line that forms the shape calculated in the shape extraction processing procedure that can satisfy the threshold value and can be regarded as the same straight line, Is performed.

A three-dimensional model generation device according to the present invention is a device for calculating a three-dimensional shape from range data to generate a three-dimensional model, wherein a target point and a target point are calculated from distance information obtained from the range data. Edge feature amount calculation processing means for calculating a height difference from a point adjacent to the edge, and calculating a feature amount relating to an edge of a shape in which the height difference is a specific pattern; and a feature amount relating to the edge of the calculated shape. A region feature amount calculation processing means for searching for a target point included in a region surrounded by the edge, and calculating a feature amount relating to the region of the searched target point; and a feature amount relating to the edge of the calculated shape; A shape extraction processing unit for extracting a shape from the calculated feature amount relating to the region of the target point; and for a closed region of the extracted shape, a range data within the closed region. Gradient feature value calculation processing means for calculating a feature value related to the arrangement; dividing the closed region from the calculated feature value related to the gradient; calculating a three-dimensional shape from the extracted shape to generate a three-dimensional model And a shape shaping processing means.

Further, a program for causing a computer to execute the procedure in the above three-dimensional model generation method is recorded on a computer-readable recording medium.

In the present invention, the feature amount calculated from the distance information obtained from the range data is used to obtain
A three-dimensional model is generated by calculating a three-dimensional shape. From the distance information obtained from the range data, a height difference of, for example, four consecutive points including the target point is calculated, and when the height difference has a specific pattern, a pattern and a position in that direction are calculated as a feature amount. As a result, noise caused by a measurement error of the range data is reduced, and very efficient high-speed processing is realized.

Further, a straight line forming a shape is calculated by using a set of points having feature amounts having the same height difference pattern in the coordinate axis direction such as the X direction and the Y direction as one unit, and a region feature amount calculation process is performed. By extracting straight lines constituting the same shape from the straight line group calculated using the feature amount calculated by the procedure / means, noise in the shape extraction procedure / means is reduced, and the processing is made more precise. , Achieve higher speed.

In addition, for a region constituting the shape calculated by the procedure / means of the shape extraction processing, a height difference between two consecutive points is calculated from distance information obtained from the range data,
Flatten the gradient in the coordinate axis direction such as X direction and Y direction,
By classifying uphill and downhill, a model having a complex gradient that is not a plane is generated.

Further, a straight line forming an area having the same gradient is calculated with respect to the gradients calculated by the gradient feature amount calculating process / method, and the straight line and the shape extracting process are performed. By shaping the shape by integrating the straight lines calculated by the means and those that can be regarded as the same straight line that satisfies the threshold value, a model having a complicated shape is generated, and the efficiency is improved. Realize good very fast processing.

[0022]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows the configuration of an apparatus for realizing a three-dimensional model generation method according to an embodiment of the present invention.

The apparatus according to the present embodiment accepts processing from an operator and stores the processing result in a three-dimensional model database 7, and a target point and a target point based on distance information obtained from a range database 8. An edge feature value calculation processing means for calculating a height difference between a point adjacent to the point and calculating a feature value relating to an edge having a shape in which the height difference is a specific pattern; Region feature calculation processing means 3 for searching for a target point included in a region surrounded by the feature in relation to the edge of the shape calculated in step 2 and calculating the feature in the region of the target point; From the feature amount regarding the edge of the shape calculated by the calculation processing unit 2 and the feature amount regarding the region of the target point calculated by the region feature amount calculation processing unit 3 Shape extraction processing unit 4 for extracting Jo
A gradient feature value calculation processing unit 5 for calculating a feature value relating to a gradient from the range data in the region of the range database 8 for the closed region calculated by the shape extraction processing unit 4; Shape shaping processing means 6 for dividing the closed area from the feature amount relating to the gradient in the area calculated by the calculation processing means 5 and calculating a model of the shape;
It has.

In one aspect of the present invention, the three-dimensional model database 7 stores data including data of three-dimensional coordinate points of each point constituting the model shape and data of constituent surfaces. The range database 8 stores data including distance information Range (i, j) for the point (i, j) on the grid, and grid values, and coordinate values corresponding to (0, 0).

The input / output processing means 1 receives a processing target area in accordance with an instruction from an operator, and stores the processing result in the three-dimensional model database 7. This process is performed as follows in the procedure shown in FIG. [Step 1] Acquire a processing target area. [Step 2] Perform edge feature value calculation processing [Step 3] Perform area feature value calculation processing. [Step 4] A shape extraction process is performed. [Step 5] Set the processing target area as a closed area ShapeRg
The threshold is lowered for each n (o), and [Step 2] to [Step 4] are executed again. [Step 6] A gradient feature amount calculation process is performed. [Step 7] A shape shaping process is performed. [Step 8] The processing result is stored in the three-dimensional model database.

The above threshold value may be determined in advance as a value for each image or each processing target area, or may be dynamically determined according to the distribution of the range data of the target points included in the obtained processing target area. May be set.

The edge feature amount calculation processing means 2 calculates a height difference from a point adjacent to the target point based on distance information obtained from the range data, and calculates a feature relating to an edge having a shape whose height difference becomes a specific pattern. Calculate the amount. This process is performed as follows in the procedure shown in FIG. [Step 1] Range data and a threshold value for a target point included in the processing area are acquired. [Step 2] For the target point (i, j), (i−
(1, j)-(i, j), (i, j)-(i + 1, j)
The height difference between (i + 1, j)-(i + 2, j) is calculated. [Step 3] It is checked whether the calculated height difference satisfies the threshold value, and the height relationship in the X direction between the target points is classified as と (climbing), ↓ (downhill), → (flat). [Step 4] The classified pattern of the height relation is examined, and in the case of the corresponding pattern, the feature amount F (n) = (ty)
pe, sx, sy, i, j) are stored. type is a classification type, and sx is a pattern in the X direction (Up or Dow).
no or Plane or Between), sy
Is the pattern in the Y direction (Up or Downnor Pl)
anne or Between), i and j indicate positions. If → ↑ →, (1, Up, NULL,
i, j), (1, Up, NULL, i + 1, j), and if it becomes → ↓ →, (1, Down, NUL)
L, i, j), (1, Down, NULL, i + 1,
j). [Step 5] For (i, j), (i, j-1)
− (I, j), (i, j) − (i, j + 1),
The height difference between (i, j + 1)-(i, j + 2) is calculated. [Step 6] It is checked whether the calculated height difference satisfies the threshold value, and the height relationship in the Y direction between the target points is represented by ↑, ↓,
Classify as → [Step 7] The classified pattern of the height relation is examined, and if the pattern is a corresponding pattern, F (n) is stored. →
If ↑ →, then (1, NULL, Up, i,
j), (1, NULL, Up, i, j + 1), →
If ↓ →, (1, NULL, Down, i,
j), (1, NULL, Down, i, j + 1). [Step 8] [Step 2] to [Step 7] until the processing is completed for all target (i, j).
repeat.

The area feature value calculation processing means 3 searches for a feature point relating to the edge of the shape calculated by the edge feature value calculation processing means 2 for a target point included in a region surrounded by the edge, and finds the target point. Calculate the feature amount regarding the region.
This processing is performed as follows in the procedure shown in FIG. [Step 1] A feature value F (n) for a target point included in the processing area is acquired. [Step 2] F (n) of type = 1 are rearranged in descending order of height. [Step 3] F of type = 1 where sx = Up
(N) are sequentially obtained. [Step 4] Scan in the X direction from the target point (i, j), and calculate (i1, j) for which sx = Down first when type = 1. [Step 5] If sx has not been set in F (n) of each point between (i, j)-(i1, j), type
= 3, sx = Between. [Step 6] Scan in the Y direction from (i, j), and sy = Down first when type = 1 (i, j)
j1) is calculated. [Step 7] If sy is not set in F (n) of each point between (i, j)-(i, j1), type
= 3, sy = Between is set. [Step 8] (i, j), (i1, j), (i, j)
Set type = 2 in F (n) of 1). [Step 9] [Step 2] to [Step 8] are repeated until the processing is completed for all type = 1 F (n).

The shape extraction processing means 4 includes a feature quantity relating to the edge of the shape calculated by the edge feature quantity calculation processing means 2 and a feature quantity relating to the area of the target point calculated by the area feature quantity calculation processing means 3. , The shape is extracted. This processing is performed as follows in the procedure shown in FIG. [Step 1] A feature value F (n) and a threshold value for a target point included in the processing area are acquired. [Step 2] The points where sx and sy both have the same F (n) are classified into a set whose shortest distance therebetween satisfies a threshold value. [Step 3] For each calculated set, an approximate straight line Line (m) indicating the distribution of points included in the set is calculated by the least square method. [Step 4] F (n) of type = 3 is rearranged in descending order of height. [Step 5] F (n) of type = 3 is acquired in order, the classification type of F (n) of a point adjacent to the target point is sequentially examined, an area having the same type = 3 is extracted, and the area is extracted. Type = 4 is set to F (n) of all the constituent points. [Step 6] Line adjacent to the calculated area
(M) is obtained, each intersection is calculated, and the closed area
Extract eRgn (o). [Step 7] [Step 5] and [Step 6] are repeated until the processing is completed for all F (n) of type = 3.

Since the method of calculating a straight line by the least square method in [Step 3] is not particularly different from the conventional method, a detailed description thereof will be omitted here.

The gradient feature value calculation processing means 5 calculates a feature value relating to the gradient from the range data in the closed area calculated by the shape extraction processing means 4. This processing is performed as follows in the procedure shown in FIG. [Step 1] ShapeRgn included in the processing area
(O) Obtain range data of a target point included in the area. [Step 2] ShapeRgn (o) is rearranged in descending order of height. [Step 3] ShapeRgn (o) is sequentially acquired, and for the target point (i, j) included in the area,
The height difference between (i, j)-(i + 1, j) is calculated. [Step 4] It is checked whether the calculated height difference satisfies the threshold value, and the height relationship between the target points is stored as ↑, ↓, → and the classification feature amount. In the case of ↑, (5, Up, NULL,
i, j) and ↓, (5, Down, NULL, i,
j), →, (5, Plane, NULL, i,
j). [Step 5] For (i, j), (i, j) −
The height difference between (i, j + 1) is calculated. [Step 6] It is checked whether or not the calculated height difference satisfies a threshold value, and the height relationship between the target points is stored as ↑, ↓, → and the classification feature amount. In the case of ↑, (5, NULL, Up,
i, j) and ↓, (5, NULL, Down, i,
j), →, (5, NULL, Plane, i,
j). [Step 7] All target ShapeRgns
Until the processing for (o) is completed, [Step 3]-
[Step 6] is repeated. [Step 8] F (n) of type = 5 is obtained,
For (i, j), (i + 1, j), (i-1, j)
If the sx of F (n) is the same, the F (n) of (i, j)
Is changed. Sx = Plan for ↑, ↓, ↑
e, ↑, →, ↑, sx = Up, →, ↑, →, sx = Plane, →, ↓, →, sx = Pla
For ne, ↓, →, ↓, sx = Down, ↓, ↑, ↓
In this case, sx = Plane. [Step 9] F (n) of type = 5 is obtained,
For (i, j), (i, j + 1), (i, j-1)
F (n) of (i, j) is the same,
To change sy. For ↑, ↓, ↑, sy = Plan
e, ↑, →, ↓ are sy = Up, →, ↑, → are sy
= Plane, →, ↓, →, sy = Plane,
For ↓, →, ↓, sy = Down, and for ↓, ↑, ↓, sy = Plane.

This threshold value may be determined in advance as a value for each processing target area, or may be dynamically set in accordance with the variance of the distance information of the points having the acquired characteristic amounts.

The shape shaping processing means 6 calculates the gradient-related feature quantity calculated by the gradient feature quantity calculating processing means 5 from
The closed area is divided and a model of the shape is calculated. This processing is performed as follows in the procedure shown in FIG. [Step 1] ShapeRgn included in the processing area
(O), a target point constituting the region and a feature value F (n) for the target point included in the region are acquired. [Step 2] ShapeRgn (o) is sequentially acquired, and the object point (i, j) of sx = Up constituting the area is scanned in the X direction and classified into continuous areas having the same sx. I do. [Step 3] (i,
With respect to j), scanning is performed in the Y direction, and classified into continuous areas having the same sy. [Step 4] Adjacent classified areas are collected into nine sets. That is, (sx = Up, sy = U
p), (sx = Up, sy = Down), (sx = U
p, sy = Plane), (sx = PIane, sy =
Up), (sx = Plane, sy = Down), (s
x = Plane, sy = Plane), (sx = Dow)
n, sy = UP), (sx = Down, sy = Dow)
n), (sx = Down, sy = Plane). [Step 5] The outermost points forming the set of the same classification are extracted, and an approximate straight line indicating the distribution of the points is calculated by the least square method. [Step 6] The calculated straight line and ShapeRg
With respect to Line (m) constituting n (o), each inclination,
A combination of straight lines whose intercepts satisfy the threshold is calculated and integrated. [Step 7] Each intersection point of the integrated straight line is calculated, and a closed region is extracted. [Step 8] For each side of the calculated closed area, height information of the side is calculated so as to be a gradient of the classified area.

Since the method of calculating a straight line by the least square method in [Step 3] is not particularly different from the conventional method, a detailed description thereof will be omitted here.
In addition, this threshold value may be determined in advance as a value for each processing target area, or may be dynamically set according to the calculated inclination of the straight line and the variance of the intercept value.

Hereinafter, the above-described processing procedure will be specifically described in accordance with actual data.

In the input / output processing, processing is received from the operator, a processing target area is obtained, and the processing shifts to edge feature amount calculation processing. In the edge feature amount calculation processing, range data and a threshold value for a target point included in the processing area are obtained.
Here, for example, distance information Range of range data
When (i, j) is represented by an image as a gray value, 80 in FIG.
The range data as indicated by 1 is targeted. For this target point (i, j), (i-1, j)-(i, j)
, (I, j)-(i + 1, j), (i + 1, j)-
The height difference between (i + 2, j) is calculated, it is checked whether the height difference satisfies a threshold value, and the height relationship in the X direction between the target points is classified as ↑, ↓, →, and the classified height is calculated. The pattern of the relationship is checked, and if the pattern becomes → ↑ → or → ↓ →, the feature amount F (n) is stored. The same process is performed for the Y direction, and these processes are repeated for all (i, j) of interest. For example, F (n) =
A point having a feature of (1, Up, Up, i, j) is indicated by 901 in FIG. 9A, where F (n) = (1, Up, Dow).
9 (b) is a point having the characteristic of (n, i, j).
2, F (n) = (1, Down, Up, i,
The point having the characteristic of j) is 1001 in FIG.
The point having the characteristic of F (n) = (1, Down, Down, i, j) is as shown by 1002 in FIG.

Next, F (n) for the target point included in the processing area is obtained in the area feature amount calculation processing, and the type is obtained.
= 1 are rearranged in descending order of height. On the other hand, F (n) of type = 1 in which sx = Up is sequentially obtained, and scanning is performed in the X direction from (i, j), and the type is obtained.
= 1 and (i1, j) satisfying sx = Down first, and sx is added to the feature amount of each point between (i, j)-(i1, j).
Is not set yet, type = 3, sx = B
Set etween. For example, 1101 in FIG.
Are 901 and 902 in FIG. 9 and 1001 and 1 in FIG.
002 is superimposed and displayed in FIG.
Scan in the X direction from (i, j) indicated by 02,
FIG. 1 shows that sx = Down first when type = 1.
11, (i1, j) indicated by 1103, and at this time, (i, j)-(i) indicated by 1104 in FIG.
If sx has not been set in F (n) of each point between (1, j), type = 3, sx = Between is set. Further, scanning from (i, j) in the Y direction, t
First, sy = Down when type = 1 (i, j1)
Is calculated, and F of each point between (i, j)-(i, j1) is calculated.
If sy is not yet set in (n), type =
3. Set sy = Between. Finally, (i,
j), (i1, j) and (i, j1) have F (n) of type
e = 2, and set these processes to all type = 1
Is repeated until the process is completed for F (n).

Further, in the shape extraction processing, F (n) and the threshold value for the target point included in the processing area are obtained, and for the point where sx and sy have the same F (n), the shortest distance between them is obtained. The distance is classified into a set satisfying a threshold. For example, the point indicated by reference numeral 903 in FIG.
1203. Approximate straight line Lin indicating the distribution of points included in the set,
e (m) is calculated by the least squares method. For example, FIG.
For each set of 1202 and 1203 in FIG.
The straight line indicated by 1205 and 1206 in FIG. Next, F (n) of type = 3 is rearranged in descending order of height, acquired in order, and F (n) of a point adjacent to the target point is obtained.
Are examined in order, an area having the same type = 3 is extracted, and t (f) is assigned to F (n) of all points constituting the area.
Set ype = 4. For example, 130 in FIG.
A region as shown in FIG. 1 is extracted. On the contrary,
Line (m) adjacent to this region is acquired, each intersection is calculated, and a closed region ShapeRgn (o) is extracted. For example, with respect to 1301 in FIG.
ShapeRgn as shown at 1302 in 3 (b)
(O) is extracted. These processes are performed for all types
Repeat for F (n) = 3.

Further, in the gradient feature value calculation processing, ShapeRgn (o) included in the processing area and range data of the target point included in the processing area are acquired, and ShapeRgn is obtained.
(O) is sorted in descending order of height. This ShapeR
gn (o) are sequentially obtained, and a height difference between (i, j)-(i + 1, j) is calculated for a target point (i, j) included in the region, and the height difference is used as a threshold value. It is checked whether or not the condition is satisfied, and the height relationship between each target point is ↑, ↓, (i, j + 1).
A height difference between the target points is calculated, whether the height difference satisfies a threshold value is checked, a height relationship between the target points is classified as ↑, ↓, →, and a feature amount is stored. These processes are repeated for all target ShapeRgn (o). Next, F (n) of type = 5 is acquired, and s of (i + 1, j) and F (n) of (i-1, j) are obtained for (i, j).
If x is the same, sx of F (n) of (i, j) is changed, and sx of F (n) of (i, j + 1) and (i, j-1)
Are the same, the sy of F (n) of (i, j) is changed. For example, in ShapeRgn (o) as shown by 1302 in FIG. 13B, 1303 in FIG.
As a result of performing the processing assuming that the height of the region indicated by is the highest, the point having the characteristic of F (n) = (5, Up, Up, i, j) is indicated by 1401 in FIG. F (n)
A point having a feature of = (5, Plane, Plane, i, j) is as shown by 1402 in FIG.

Further, in the shape shaping process, ShapeRgn (o) included in the processing area, target points constituting the area, and F (n) corresponding to the target points included in the area are acquired, and the ShapeRgn (o) is obtained. ) Are obtained in order, and for a target point (i, j) of sx = Up constituting the area,
Scanning in the X direction is performed to classify into continuous regions having the same sx. Also, for (i, j) of sy = Up,
Scanning is performed in the Y direction to classify into continuous areas having the same sy. Adjacent ones of the classified areas are collected into nine sets, the outermost points forming the set of the same classification are extracted, an approximate straight line indicating the distribution of the points is calculated by the least square method, and the straight line and ShapeRgn For Line (m) constituting (o), a combination of straight lines whose slopes and intercepts satisfy the threshold is calculated and integrated. The intersections of these straight lines are calculated to extract a closed region, and the height information of each side of the calculated closed region is calculated so as to be a gradient of the classified region with respect to each side. For example, 130 in FIG.
For the region indicated by No. 3, 140 in FIG.
1. The area is classified into the area indicated by 1402 in FIG. 14B, and when the closed area of each area is calculated, the area becomes the area indicated by 1403 in FIG. Here, FIG.
The area indicated by 1404 in (c) is F (n) = (5, U
p, Up, i, j), which is the surface of the climb, and FIG.
The area indicated by 1405 in FIG. 4C is F (n) = (5,
Plane, Plane, i, j). Thus, a shape as shown by 1406 in FIG. 14D is calculated. Also, for example, 1 in FIG.
In the case of the straight lines 205 and 1206, since the values of the slope and intercept are close, they are integrated into one straight line, and the number of shape composing points is reduced.

Finally, in the input / output processing, the processing result is stored in the three-dimensional model database. Here, for example,
The three-dimensional model data as shown in FIG. 15 is stored.

It is to be noted that some or all of the processing functions of each unit shown in FIG. 1 can be realized using a computer, or that the computer can execute the processing procedures shown in FIGS. Needless to say, a program for realizing the processing function of each part in the computer or a program for causing the computer to execute the processing procedure can be stored in a storage medium readable by the computer, for example, an FD (Floppy Disk). : Registered trademark), MO, ROM, memory card, CD, DVD, removable disk, etc., and can be stored, provided, or distributed.

As described above, the present invention has been specifically described based on the embodiment. However, the present invention is not limited to the above-described embodiment, and may be variously modified without departing from the gist thereof. Needless to say.

[0045]

According to the present invention, since the feature calculated from the distance information obtained from the range data is used, the three-dimensional shape can be calculated and the three-dimensional model can be generated. can get.

Further, the height difference of four consecutive points including the target point is calculated from the distance information obtained from the range data, and when the height difference has a specific pattern, the pattern and position in that direction are characterized. Since the calculation is performed as an amount, noise caused by a measurement error of the range data can be reduced, and an effect that very efficient high-speed processing can be realized is obtained.

Further, a straight line forming a shape is calculated by using a set of points having feature amounts having the same height difference pattern in the coordinate axis direction such as the X direction and the Y direction as one unit, and the region feature amount calculation processing is performed. The straight lines constituting the same shape are extracted from the straight line group calculated using the feature amount calculated by the above, so that the noise in the shape extraction processing can be reduced, and the processing can be performed with high accuracy and high speed. The effect that can be obtained is obtained.

The height difference between two consecutive points is calculated from the distance information obtained from the range data with respect to the region constituting the shape calculated by the shape extraction processing, and the coordinate axis direction such as the X direction and the Y direction is calculated. Are classified into flat, climbing, and hollow, so that an effect of generating a model having a complicated gradient that is not a plane can be obtained.

Further, calculating a straight line constituting an area having the same gradient as the gradient classified in the gradient feature value calculating process, and calculating the straight line and the straight line calculated in the shape extracting process, Since the shape can be shaped by integrating those that can be regarded as the same straight line that satisfies the threshold value, it is possible to generate a model having a complicated shape,
In addition, there is obtained an effect that efficient and very high-speed processing can be realized.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an apparatus for realizing a three-dimensional model generation method according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an input / output processing flow according to the embodiment.

FIG. 3 is a diagram illustrating an edge feature amount calculation processing flow according to the embodiment.

FIG. 4 is a diagram illustrating a flow of a region feature amount calculation process in the embodiment.

FIG. 5 is a diagram illustrating a flow of a shape extraction process according to the embodiment.

FIG. 6 is a diagram illustrating a flow of a gradient feature amount calculation process according to the embodiment.

FIG. 7 is a diagram illustrating a flow of a shape shaping process in the embodiment.

FIG. 8 shows distance information Range in the embodiment.
FIG. 5 is a diagram illustrating an example of range data in which (i, j) is represented by an image as a gray value.

FIGS. 9A and 9B are diagrams illustrating an example of an edge feature amount calculation process according to the embodiment; FIG.

FIGS. 10A and 10B are diagrams illustrating a continuation of an example of the edge feature amount calculation process according to the embodiment; FIG.

FIG. 11 is a diagram illustrating an example of a region feature amount calculation process according to the embodiment.

FIGS. 12A and 12B are diagrams illustrating an example of a shape extraction process according to the embodiment;

FIGS. 13A and 13B are diagrams illustrating a continuation of an example of a shape extraction process according to the embodiment;

FIGS. 14A, 14B, 14C, and 14D are diagrams illustrating an example of a gradient feature amount calculation process and a shape shaping process according to the embodiment;

FIG. 15 is a diagram illustrating an example of three-dimensional model data generated in the embodiment.

[Description of Signs] 1 ... Input / output processing means 2 ... Edge feature quantity calculation processing means 3 ... Area feature quantity calculation processing means 4 ... Shape extraction processing means 5 ... Gradient feature quantity calculation processing means 6 ... Shape shaping processing means 7 ... 3 Dimensional model database 8 ... Range database

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Kenichi Ichikawa 2-3-1 Otemachi, Chiyoda-ku, Tokyo Within Nippon Telegraph and Telephone Corporation (72) Tomohiko Arikawa 2--3, Otemachi, Chiyoda-ku, Tokyo No.1 Nippon Telegraph and Telephone Corporation F-term (reference) 5B057 AA13 BA24 CA12 CB13 CC03 CH01 CH11 DA08 DB02 DC08 DC09 DC16 DC36 5L096 AA09 BA20 CA18 FA02 FA06 FA66 FA70 GA07

Claims (8)

[Claims] 1. A three-dimensional shape is calculated from range data,
A method for generating a three-dimensional model, comprising calculating a height difference between a target point and a point adjacent to the target point from distance information obtained from range data, and calculating a height difference between the target point and a point adjacent to the target point. An edge feature value calculation processing procedure for calculating a feature value related to an edge; and a feature point related to the edge of the calculated shape, and a target point included in a region surrounded by the edge is searched for. A region feature amount calculation processing procedure for calculating a feature amount; a shape extraction processing procedure for extracting a shape from the feature amount regarding the edge of the calculated shape and the calculated feature amount regarding the region of the target point; A gradient feature value calculation processing procedure for calculating a feature value related to a gradient from the range data in the closed region for a closed region having the determined shape, and a feature value related to the calculated gradient. Et the performed division of the closed region, the three-dimensional model generating method characterized by and a shape shaping processing procedure for generating a three-dimensional model to calculate the three-dimensional shape from the extracted shape. 2. In the edge feature amount calculation processing procedure, a height difference of four consecutive points including a target point is calculated from distance information obtained from range data, and when the height difference has a specific pattern, The three-dimensional model generation method according to claim 1, wherein at least a pattern and a position in the direction are calculated as a feature amount. 3. The region feature amount calculation processing procedure includes, based on a feature amount relating to an edge of the shape calculated in the edge feature amount calculation process procedure, a target point included in a region surrounded by a point having the feature amount. The method according to claim 1, wherein the search is performed. 4. In the shape extraction processing procedure, a straight line forming a shape is calculated using a set of points having a feature amount having the same height difference pattern in the coordinate axis direction as one unit, and the area feature amount calculation process 4. The method according to claim 1, wherein straight lines constituting the same shape are extracted from a straight line group calculated by using a feature amount relating to the area of the target point calculated in the procedure. 3
Dimension model generation method. 5. In the gradient feature amount calculation processing procedure, a height difference between two consecutive points is calculated from distance information obtained from range data for an area constituting the shape calculated in the shape extraction processing procedure. The method according to any one of claims 1 to 4, wherein the gradient in the coordinate axis direction is classified into flat, rising, and hollow. 6. In the shape shaping processing procedure, the gradient calculated in the gradient feature value calculation processing procedure is classified, and a straight line forming an area having the same gradient with respect to the classified gradient is calculated. And shaping the shape by integrating the straight line and the straight line constituting the shape calculated in the shape extraction processing procedure that can satisfy the threshold value and can be regarded as the same straight line. Claim 1
The method for generating a three-dimensional model according to any one of claims 1 to 5. 7. A three-dimensional shape is calculated from the range data,
An apparatus for generating a three-dimensional model, comprising calculating a height difference between a target point and a point adjacent to the target point from distance information obtained from range data, and calculating a height difference between the target point and a point adjacent to the target point. Edge feature value calculation processing means for calculating a feature value related to the edge; and a feature point related to the edge of the calculated shape, searching for a target point included in a region surrounded by the edge, and Area feature amount calculation processing means for calculating a feature amount; shape extraction processing means for extracting a shape from the feature amount regarding the edge of the calculated shape and the calculated feature amount regarding the region of the target point; Gradient feature value calculation processing means for calculating a feature value related to the gradient from the range data in the closed region for the closed region having the determined shape; and a feature value related to the calculated gradient. Et the performed division of the closed region, the three-dimensional model generating device characterized by comprising a shape shaping means for generating a three-dimensional model to calculate the three-dimensional shape from the extracted shape. 8. A program for causing a computer to execute a procedure in the three-dimensional model generating method according to claim 1 on a computer-readable recording medium. 3
A recording medium on which an execution program for a dimension model generation method is recorded.