JP2000329552A - Three-dimensional map preparing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a three-dimensional feeling while facilitating the operation by approximating a building, and the like, with a polygon from three- dimensional coordinates determined through a coordinate process and sections determined through sectioning process, selecting a texture being attached to the polygon and then mapping. SOLUTION: At first, the image of a road or a building is picked up by means of a camera (S1), the camera parameters are calibrated (S2) and then the two-dimensional coordinates are determined at a point on a photograph representing the same point of a building in three-dimensional space (S3). Coordinates at a corresponding point in the three-dimensional space are then determined from the coordinates at the corresponding points on two photographs (S4). Subsequently, a figure showing the arrangement of a group of points is determined for a height data higher than a specified value (S5) and a building is sectioned using a digital map (S6). Furthermore, a photograph picking up the image of a three-dimensional polygon is made up by means of a computer (S8) and a texture is selected (S9). In a three-dimensional map thus obtained, the photograph of a building is varied depending on the viewing direction thereof and thereby a feeling of three-dimensional photograph can be imparted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention belongs to the technical field of a three-dimensional map creation method. In particular, the present invention relates to a technical field of a method of creating a three-dimensional map created using aerial photographs or the like.

[0002]

2. Description of the Related Art Conventionally, as a three-dimensional map, there are a three-dimensional map in which a building or the like is drawn three-dimensionally, and a map in which a drawing of a building or the like is cut out and stuck on the map so that it can be lifted. However, this type of three-dimensional map merely shows a view of a building or the like from a typical angle such as the front, and does not change even if the viewing direction changes, so there is no three-dimensional effect. In addition, there is an attempt to use a special lens such as a lenticular lens to make a map look three-dimensional.
A dimensional map creation method has not yet been proposed. It is also possible to stereoscopically view a map by attaching a special device to both eyes, such as a virtual space. However, since a special device must be attached, complexity is a problem.

[0003]

SUMMARY OF THE INVENTION The present invention has been made under the above-mentioned background, and does not require a special device, is inexpensive, is easy to operate, and provides a three-dimensional feeling to an observer. It is intended to provide a method of creating a given map.

[0004]

The present invention employs the following means to solve the above-mentioned problems. That is, claim 1
The described invention is a method of creating a three-dimensional map based on a plurality of aerial photographed photographs, wherein the map producing method includes a coordinate calculating step of obtaining three-dimensional coordinates of a point such as a building in the three-dimensional photograph from the photographed photograph. A partitioning step of determining a partition of the building or the like; a polygonalizing step of approximating a building or the like with a polygon from the three-dimensional coordinates obtained in the coordinate step and the partition obtained in the partitioning step; And a mapping step of pasting the texture of the selected photograph.

According to a second aspect of the present invention, in the first aspect of the present invention, the coordinate calculating step includes a rectification step of making an epi-curve of the photographed picture horizontal, and thereafter, a rectification step on the photographed picture. It is characterized by including a step of obtaining a corresponding point.

According to a third aspect of the present invention, in the first or second aspect of the present invention, the dividing step is performed using a digital map.

According to a fourth aspect of the present invention, in the first to third aspects of the present invention, if the division of the building or the like obtained in the division step does not form a convex shape, it is divided into a plurality of convex shapes. It is characterized by doing.

According to a fifth aspect of the present invention, in the first aspect of the present invention, the polygonalizing step comprises vertically raising a ground frame of the divided building or the like to form a side wall, The roof is constructed by obtaining the building height from the coordinate group obtained in the coordinate calculation step.

According to a sixth aspect of the present invention, in the first aspect of the present invention, in the polygonizing step, a roof of a building or the like is set to an average value of coordinate heights of point groups in a ground frame. It is characterized by being approximated by a flat roof.

According to a seventh aspect of the present invention, in the first aspect of the present invention, in the polygonalizing step, the roof height of the building or the like is determined by increasing the coordinate heights of point groups in the ground frame in order of size. It is characterized by being arranged side by side and being approximated by a flat roof whose height is the median value.

According to an eighth aspect of the present invention, in the first aspect of the present invention, in the polygonizing step, a plurality of predetermined shape models are designated on the building roof, and the polygon model is obtained in the coordinate calculating step. The height data of the coordinate group is projected on the x-plane and the y-plane, and the predetermined shape model is selected from those projected shapes and polygonized.

[0012] According to a ninth aspect of the present invention, in the first aspect of the present invention, the selecting step includes determining whether or not the portion to be mapped appears in the photographed photograph. 3 corresponding to the part to be mapped
A plurality of points are arranged on the surface of the three-dimensional polygon part, and whether or not there is another three-dimensional polygon other than the three-dimensional polygon on a half line passing through the center of the lens of the camera starting from the arranged point Is determined.

According to a tenth aspect of the present invention, in the first to ninth aspects of the present invention, the selecting step is performed when a part to be mapped to the polygon appears in a plurality of the photographed photographs. Select any one of the
It is characterized by performing mapping.

According to an eleventh aspect of the present invention, in the invention of the tenth aspect, the selecting step selects the largest one of the plurality of photographed images.

According to a twelfth aspect of the present invention, in the first to eleventh aspects, the selecting step selects a photographed photograph taken to take out a texture other than the aerial photographed photograph. And

According to a thirteenth aspect of the present invention, in the first or twelfth aspect, the mapping step comprises:
It is characterized in that the texture is pasted by reverse warping.

[0017]

FIG. 1 shows the steps of the production method of the present invention. FIG. 2 shows a schematic diagram of an apparatus for performing the method. FIG.
Shows a building arrangement for explaining the embodiment of the present invention.
Hereinafter, an embodiment of the invention will be described with reference to the drawings.

In FIG. 2, the present apparatus displays two digital cameras 11 and 12, a computer 13 which takes in data of photographed images from the cameras 11 and 12 and performs operations such as processing, and displays the result of the computer 13. Display 14
And a keyboard 15 and a mouse 16 for inputting data to the computer 13. The computer 13 includes a control operation device 18 and a storage device 19
And bus line 20 and input / output interface 2
1 is comprised.

In FIG. 1, step S1 is the camera 1
The roads and buildings are photographed in steps 1 and 12. Photographs are taken at different positions and angles from the air like an aerial photograph. Shooting requires at least two or more pictures, and the change in the texture of the building when the viewpoint changes as the number of pictures taken from various angles increases, but has the disadvantage that the processing time is long. Hereinafter, a case where two pictures are taken will be described. FIG. 3 shows a situation where an object 25 in a three-dimensional space where buildings and the like are arranged is photographed by the cameras 11 and 12. The camera 11 shoots from the left side, and the camera 12 shoots from the right side. Pictures 11a and 12a are pictures that are actually taken. The camera 26 is a camera for taking a picture viewed from an arbitrary direction and position, and does not actually take a picture. Picture 2
Reference numeral 6a denotes a photograph created by a computer, which differs depending on the viewing angle. The scenery of the building caused by the movement of the person 31 is represented by the photograph 26a.

Step S2 is a step of calibrating the camera parameters of the cameras 11 and 12. The camera parameters include all data related to the camera, such as the position, orientation, and angle of view of the camera, are represented by a (3 × 4) matrix, and include 12 unknown parameters. When the camera parameters are obtained, a projection transformation matrix from the object such as a building in the three-dimensional space to the photographed photographs 11a and 12a is known. The camera parameters can be obtained by actually measuring the geometrical optical properties of the camera. However, a method of calibrating the parameters by performing three-dimensional measurement on a plurality of points having known coordinates in a three-dimensional space is known. It is practical.

The calibration method is conventionally known, and here, a conventionally known calibration technique is used. That is, three-dimensional space such as a reference object or a building in which at least six points having known three-dimensional coordinate coordinates are arranged is photographed by the cameras 11 and 12, and coordinates of corresponding points on the photographed photograph are obtained. To obtain camera parameters by calculation. In addition, it is preferable that there are 10 to 20 points. In this case, the error is calculated so as to minimize the sum of squares of the error. As a result, 3
A point such as a building in the three-dimensional space is associated with a point on the photographed photograph.

FIG. 4 is an example of a photograph taken by the cameras 11 and 12. (A) is a photograph 11a taken by the camera 11 viewed from the left, and (B) is a photograph 12a taken by the camera 12 from the right. These photographs 11a, 1
The digital data 2a is stored in the storage device 19 of the computer 13. Note that the photographs 2a and 2a show the building 2
7, 28, 29 and the dome 30 are photographed.

In step S3, two-dimensional coordinates of the corresponding point of the photograph 11a and the photograph 12a, that is, the point of the photograph representing the same point of the building or the like (or the reference object) in the three-dimensional space are obtained. Corresponding points are determined automatically by using a correlation method. When the angle at the time of photographing is large like an aerial photograph, a method using a correlation method is suitable. Before finding a corresponding point, an operation called rectification is performed. This operation is an operation for converting an image so that the epi-curve becomes horizontal, and is a conventionally known technique. Detailed description is omitted. As a result, the point for which the correlation is to be determined has only to be searched in the horizontal direction, and the search operation is simplified. Note that the epipolar line refers to a straight line passing through a point in another photographed photograph corresponding to a point in one photographed photograph.

The correlation method is a conventionally known technique, in which a small window is set for each photographed photograph, one of the windows is fixed, and the other is moved to obtain a correlation between the windows. Is determined and determined as a corresponding point. The calculation for this is performed by the control calculation device 18. Result is storage device 1
9 recorded.

In step S4, the coordinates (x1, y1) and (x2, y2) (corresponding to the illustrations) of the corresponding points of the two photographed photos are taken from the coordinates (X, Y, Z) (of the corresponding points) in the three-dimensional space. (Not shown). As a result, coordinate points and heights on the ground such as a building are obtained. As a method of obtaining the corresponding point in the three-dimensional space, for example, it can be obtained by calculation using two projective transformation matrices and the coordinates of the corresponding point in the photographed image. Thereby, the position and height data of the visible point of the building are obtained. In step S5, as shown in FIG.
An arrangement diagram of a point group having height data (z coordinate) equal to or larger than a predetermined value is obtained. Hereinafter, this set of points is called a three-dimensional coordinate group, and CO
It is represented as P (cloud of point set).

In step S6, building segmentation is performed using a digital map. The digital map is provided with data of corner points on the ground of the buildings, that is, coordinates (x, y, 0) of the corner points for each building. Based on this data, each building is classified. This sorting may be performed by the operator using the mouse 16 while displaying the coordinates (x, y, 0) of the corner points on the display unit 14 or may be automatically performed by incorporating an automatic sorting program into the control calculation unit 18. Is also good. The boundary of the divided building should coincide with the boundary of the COP shown in FIG. 5 when the building stands vertically.

In step S6, each building is polygonized. That is, with respect to the buildings 27, 28, and 30, the boundary line on the ground of the building is raised vertically to form a side wall, and the height of the building is determined using the COP. Further, a building whose plane is not convex, such as the building 29, is appropriately divided, for example, as shown by a dotted line in FIG. Next, a method for determining the height of a building using the COP will be described. The roof of the building is regarded as a flat roof parallel to the horizontal plane, and CO
The average value of the height data is obtained for the section of each building represented by P. When calculating the average height from the COP, height data that is significantly apart is excluded as noise. Alternatively, a median filter may be used instead.
The median filter arranges the data in order of size and adopts the central value as the building height. By using this method, the operation of removing noise is omitted, and the processing is simplified.

From the data obtained as described above, each building is displayed as a three-dimensional polygon as shown in FIG.
FIG. 6 shows a three-dimensional model of a building or the like in a three-dimensional space, and a texture representing a window or the like is not attached. The building 29 is divided into polygons 34 and 35, and the building 3
0 is represented by a polygon 36 having a flat roof.
When various deformed roofs such as a dome-shaped hemisphere, a sloping roof, a roof having a ridgeline in the center, etc. are expressed as the roof display means, these are determined in advance as models and CO
You may make it analyze a P data and classify and express a deformed roof. For example, the COP data is converted to an x-plane, y
It is also possible to project on a plane and determine from their projected shapes.

In step S8, the computer 13 creates a photograph (hereinafter referred to as a model photograph) of a three-dimensional polygon model photographed by the camera 26 in FIG. The photograph 40 in FIG. 7 shows a part of the model photograph. In order to create a model photograph, the camera parameters of the camera 26 in FIG. 3 can be obtained by applying a predetermined matrix to a matrix indicating the camera parameters of the camera 11 or the camera 12. For example, when the camera 26 is rotating, it is obtained by multiplying a rotation matrix, and when the camera 26 is translated, it is obtained by multiplying a matrix for the translation. By applying the obtained matrix to coordinates such as corner points of the three-dimensional model, the coordinates of the model photograph can be obtained, and by connecting these, the model photograph 40 can be obtained. In addition, a part where buildings and the like overlap and cannot be seen is erased by a hidden line elimination method (known method). It is convenient to rotate and move the camera 26 by moving the mouse 16 up, down, left, and right so that changes in the map can be seen on the display 14.

In step S9, a texture to be attached to each part of the model photograph 40 is selected. In this selection step, in addition to the case where the photograph is selected from the photographed photographs 11a or 12a, a clearer photograph photographed (for example, from the ground) may be selected in order to take out the texture. In the former case, it is checked whether a portion corresponding to the portion to which the texture is to be attached exists in the photographed photograph 11a or 12a or both. As a method of checking, a plurality of points are randomly designated on the surface of a three-dimensional polygon corresponding to the polygon surface of the model photograph to which the texture is to be attached, and all designated points are connected to the lens centers of the cameras 11 and 12. The control arithmetic unit 18 calculates and checks whether another three-dimensional polygon exists on a half line (a half line from the designated point toward the lens center).

If both exist, the three-dimensional polygon surface to which the texture is to be attached is hidden behind the other three-dimensional polygons, and the photographed photographs 11a and 12a
Texture cannot be warped. In this case, the user selects not to perform mapping. When another polygon exists on the half line passing through the center of one camera lens but no other polygon exists on the half line passing through the center of the other camera lens, the texture is taken from the photograph taken by the other camera. Extract and warp. If no other polygon exists on a half line passing through the centers of both camera lenses, texture is extracted from one of the photographed images and warping is performed. For example, one having a larger area such as a wall surface is selected. This usually results in a more detailed texture.

Further, a photograph of a building or the like taken from the ground can be used without warping the texture from an aerial photograph or the like. By using photos taken from the ground, clearer photos can be mapped. It is also possible to attach a texture to a portion not appearing in the aerial photograph 11a or 12a.

A mapping procedure for attaching a texture to the quadrilaterals 42 (P1 to P4) constituting the polygon 32 of the model photograph 40 shown in FIG. 7 will be described below. First, a projection transformation matrix φ to a photographed photograph (for example, photograph 11a) for taking out the texture from the surface to which the texture is to be attached
Ask for. In the method for obtaining the projection transformation matrix φ, the projection transformation matrix can be obtained by multiplying a reference projection transformation matrix by a matrix such as rotation or translation. Also, a method using affine transformation or homographic transformation is known. If this known method is used, a texture can be extracted from a photograph taken from the ground or the like.

Inverse warping is performed using the obtained projective transformation matrix φ. That is, the quadrilateral 42 in front of the polygon 32
The coordinates (x,
The corresponding point is obtained (for example, on the photograph 11a) by multiplying y) by the projective transformation matrix φ. The pixel data of the obtained point on the photograph 11 a is read out from the storage device 19 and read into the quadrilateral 42.
As pixel data. The above procedure is performed for all figures of the model photograph 40. Thereby, the texture is attached to the model photograph 40. Step 11
Then, it is detected whether the viewpoint position and the angle have changed.
Actually, it is detected whether or not the position of the mouse 16 has moved, and if it has moved, the procedure of steps 8 to 10 is repeated.

As described above, the three-dimensional map created by the method of the present embodiment changes as the direction in which the photograph of a building or the like changes, so that it is as if a three-dimensional photograph is viewed. This has the effect of giving a sense to the observer. In addition, since a complicated device is not required to carry out the present method, there is an effect that an inexpensive three-dimensional map can be created, which can be said to be a practical means. Furthermore, once the three-dimensional map according to the present method is created, it is easy to reproduce the three-dimensional map, and the reproducing means can be produced or copied in large quantities.

The embodiments and examples of the present invention have been described in detail with reference to the drawings. However, the specific configuration is not limited to the examples, and changes in design and the like may be made without departing from the gist of the present invention. Even if there is, it is included in the present invention. For example, the order of the aerial photography in step 1 and the calibration in step 2 may be changed, or may be performed simultaneously. Also, it is not necessary to execute all the steps described in the present embodiment as they are, and some of them may be omitted or changed to another method.

[0037]

As described above, according to the configuration of the present invention, the three-dimensional map created by the present method changes as the direction in which a photograph of a building or the like changes, as if it were a three-dimensional photograph. This has the effect of giving the observer the sensation of watching the scene. In addition, since a complicated device is not required to implement the method, an inexpensive three-dimensional map can be created, which is a practical means.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a procedure according to an embodiment of the present invention.

FIG. 2 shows a configuration example of an apparatus for performing the method of the present invention.

FIG. 3 illustrates an arrangement configuration of a building and a camera according to the present embodiment.

FIG. 4 shows an example of a photograph taken by the camera of the embodiment.

FIG. 5 shows an example of a COP in the present embodiment.

FIG. 6 shows a polygonized building in the present embodiment.

FIG. 7 shows a procedure for selecting and pasting a texture of a photograph.

[Explanation of symbols]

 11, 12 Camera 13 Calculator 14 Display 15, 16 Input operation means 18 Control arithmetic unit 19 Storage memory 27-30 Building 32-36 Polygon

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Olivier Delahaya 5-19-9 Hiroo, Shibuya-ku, Tokyo Hiroo ON Building Gen-Tech Co., Ltd. (72) Inventor Takao Ichihashi 5-19-9, Hiroo, Shibuya-ku, Tokyo Hiroo ON Building Gen-Tech Co., Ltd. (72) Inventor Kazushi Suzuki 5-19-9 Hiroo, Shibuya-ku, Tokyo Hiroo ON Building Gen-Tech Co., Ltd. F-term (reference) 2C032 HB03 HB11 HB22 HC23 5B050 BA02 BA07 BA09 BA17 EA19 EA22 EA28 FA02 9A001 DD15 HH29 JJ11 JJ31

Claims (13)

[Claims] 1. A method for creating a three-dimensional map based on a plurality of aerial photographs, the method comprising: calculating a three-dimensional coordinate of a point such as a building in the three dimensions from the photographed photographs. A partitioning step of determining a section of the building or the like; a polygonalizing step of approximating a building or the like with a polygon from the three-dimensional coordinates obtained in the coordinate calculation step and the section obtained in the partitioning step; A three-dimensional map creation method, comprising: a selection step of selecting a texture; and a mapping step of pasting a texture of the selected photograph. 2. The method according to claim 1, wherein the coordinate calculating step includes a rectification step of making an epi-curve of the photographed picture horizontal, and a step of obtaining a corresponding point on the photographed picture thereafter. 3. The method for creating a three-dimensional map according to item 1. 3. The method according to claim 1, wherein the dividing step is performed using a digital map.
3. The method for creating a three-dimensional map according to item 1. 4. The method according to claim 1, wherein when the division of the building or the like obtained in the division step does not form a convex shape, the division is made into a plurality of convex shapes. 3D map creation method. 5. The polygon conversion step comprises forming a side wall by vertically raising a ground frame of the divided building or the like, and forming a roof by obtaining a building height from a coordinate group obtained in the coordinate calculation step. The method for creating a three-dimensional map according to any one of claims 1 to 4, wherein: 6. The polygon forming step according to claim 1, wherein the roof of the building or the like is approximated by a flat roof having an average value of coordinate heights of point groups in the ground frame. 3. The method for creating a three-dimensional map according to item 1. 7. The polygon conversion step is characterized in that a roof of a building or the like is arranged in the order of magnitude of coordinate heights of point groups in a ground frame, and is approximated by a flat roof having a median height. The three-dimensional map creation method according to any one of claims 1 to 5. 8. The polygonizing step specifies a plurality of predetermined shape models on a building roof and projects height data of a coordinate group obtained in the coordinate calculating step on an x-plane and a y-plane. The three-dimensional map creation method according to any one of claims 1 to 6, wherein the predetermined shape model is selected from the projected shapes and converted into a polygon. 9. The selecting step includes arranging a plurality of points on a surface of a three-dimensional polygon portion corresponding to the portion to be mapped, in order to determine whether the portion to be mapped appears in the photographed photograph. And determining whether or not there is another three-dimensional polygon other than the three-dimensional polygon on a half line passing through the center of the lens of the camera starting from the arranged point. The method for creating a three-dimensional map according to claim 1. 10. In the selecting step, when a portion to be mapped to the polygon appears in a plurality of the photographed photographs, any one of the photographed photos is selected and the mapping is performed. Any one of claims 1 to 8
3. The method for creating a three-dimensional map according to item 1. 11. The three-dimensional map creation method according to claim 9, wherein the selecting step selects a photograph having the largest photographed image among the plurality of photographed photographs. 12. The three-dimensional map according to claim 1, wherein the selecting step selects a photographed photograph taken for taking out a texture other than the aerial photographed photograph. How to make. 13. The three-dimensional map creation method according to claim 1, wherein in the mapping step, a texture is pasted by reverse warping.