JPH0830194A - Method for forming geospecific texture - Google Patents

Abstract

PURPOSE:To form geospecific texture patterns by executing strain correction of aerial photographs. CONSTITUTION:The states of the plural sample points photographed by a camera when this camera is moved virtually over a map and the plural sample points on an actually taken photograph are compared. The position and attitude of the camera where and at which the photograph is taken are then retrieved. The information on the heights at the plural points more than the reference points is obtd. by interpolating the altitude values of the few reference points. Coordinate conversion to the aerial photograph is executed from the ground surface by assuming that the photograph is taken by the camera of the position and attitude determined by a coordinate conversion system to make the points on the ground surface and the points on the aerial photograph correspondent to each other. The image data of the aerial photograph is stored into a memory made correspondent to the ground surface.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of generating a geospecific texture pattern used in a window visual field generator in a training device such as a flight simulator, and more particularly to distortion correction of an aerial photograph.

[0002]

2. Description of the Related Art In a conventional out-of-window field-of-view generator of a flight simulator, a texture pattern for mapping on the ground surface has a uniform pattern repeated. This was not able to express the peculiarity of the area, and was lacking in reality. On the other hand, there is one that maps a geospecific texture pattern, which can generate a topographic image unique to the area.

The source of the geospecific texture pattern may be an aerial photograph of the ground surface taken from an aircraft. In that case, since the photograph is a central projection, there is distortion due to the elevation of the ground surface, and the photograph itself cannot be used as a texture pattern. In order to use a photograph as a texture pattern, the distortion must be corrected from the central projection image to the orthogonal projection image.

When photographing the ground surface with a camera, an arbitrary point on the ground surface is center-projected to a certain point on the film by a simple coordinate transformation. This coordinate conversion formula is given by the position and orientation of the aircraft (camera) that took the image. Furthermore, it is conceivable to use this coordinate conversion formula to perform coordinate conversion on the aerial photograph to convert it into an orthogonal projection image.

[0005]

However, there are the following problems in performing coordinate transformation from an aerial photograph using this coordinate transformation formula to transform it into an orthogonal projection image.

Problem 1 The position and attitude of the aircraft that took the aerial photograph must be clear, but these cannot be obtained from the aerial photograph by mathematical expressions.

Problem 2 Since the image of the aerial photograph has no information in the z direction (height direction), coordinate conversion cannot be performed only by the aerial photograph.

Problem 3 To solve Problem 2, even if the coordinate conversion from the ground surface to the aerial photograph is performed by the above coordinate conversion formula that associates the points on the ground surface with the points on the aerial photograph, very fine elevation data on the ground surface can be obtained. It is necessary and it is impossible to prepare this in advance. The number of points on the ground surface for obtaining this elevation data is related to the resolution of the geospecific pattern necessary for operation, and for an aircraft, for example, a resolution of 4 m square is required. This point of surface to be coordinate conversion aerial photographs of surface coordinate system X _w, in Y _w plane, X _w-direction, the Y _w direction, is the need to set each example 4m intervals.
The elevation value Z _ow, which is the Z _w component of each point on the ground surface, cannot be given by the function of its X and Y components, X _ow and Y _ow. When generating a specific texture pattern, the amount of elevation data at intervals of 4 m is enormous.

The present invention solves the above problems,
It is an object of the present invention to provide a method for generating a geospecific texture by semi-automatically correcting distortion of an aerial photograph.

[0010]

A method for generating a geospecific texture pattern according to the present invention is for generating a geospecific texture pattern from an aerial photograph, and the aerial photograph is obtained as image data and displayed on the screen. A first process of pointing the arbitrary aerial sample points on the aerial photograph displayed on the screen with a mouse and obtaining the position as a photographed image position; The second process of obtaining the position on the map, which is displayed on the screen with the mouse so as to correspond to the point, as the ground position including the elevation value, and the position and orientation of the virtual camera are virtually changed within a predetermined range. Let me
Coordinates of the ground surface position in the second process corresponding to the arbitrary plurality of sample points as a virtual photograph position by a conversion formula for converting a point on the ground surface in the ground coordinate system when the image is taken by the camera into a film coordinate system. A third step of converting and retrieving a virtual photograph position closest to the photographed image position, and regarding the virtual camera position and posture at that time as the camera position and posture when the aerial photograph was taken; Equipped with a memory for distortion correction aerial photography in which storage elements are arranged corresponding to multiple points on the surface coordinate system, by interpolating the elevation value of the reference point set less than the number of the multiple points on the surface coordinate system by Lagrange interpolation The interpolated elevation values of the plurality of points are obtained, and the ground position of the interpolated elevation values of the obtained plurality of points is determined by the camera in the position and the posture considered in the third step. The distortion-corrected photograph position is converted into a film coordinate system as a shaded object using the conversion formula, and the image data in the first process at the distortion-corrected photograph position is converted into the distortion-corrected aerial image corresponding to the plurality of points. And a fourth step of storing in a photographic memory.

[0011]

In the method of generating a geospecific texture pattern according to the present invention, the following operations are performed. First, an aerial photograph is made into a file as image data, for example. Then, an arbitrary plurality of sample points are selected from this aerial photograph, and the position thereof is obtained as a real photograph position. On the photograph, a certain fixed point is set as a reference position, and the position of each point on the photograph is measured from there. In addition, a point corresponding to the arbitrary plural sample points is searched for on the map, and the position is obtained as the ground surface position together with the elevation value of the point. On the map, a certain fixed point is set as the reference position, and the surface position is measured from that point.

The aerial photograph should be displayed on the screen,
On the displayed screen, the user designates the arbitrary plural sample points on the screen. The position on the photograph can be obtained as the photographed photograph position by pointing the mouse. Further, the map is simultaneously displayed on the same screen on which the aerial photograph is displayed, and the screen is designated by the mouse so as to correspond to the arbitrary plural sample points on the displayed screen. It is possible to obtain the position on the map as the ground surface position by inputting the elevation value for the designated position and pointing the screen of the mouse.

Next, it is assumed that the camera is above the terrain represented by the map, and that the camera virtually moves and changes its attitude to take an image.
When the surface of the earth is photographed when the camera is located at an arbitrary position from the reference position of the surface of the earth, each point on the surface of the earth is coordinate-converted into a corresponding point in the film coordinate system by a conversion formula. This converted position will be called a virtual photograph position. Therefore, every time the virtual camera moves, the virtual photograph position of the point on the map corresponding to the arbitrary plural sample points changes in the film coordinate system. The virtual photograph position and the photographed photograph position of the arbitrary multiple sample points are compared for each predetermined point (each lattice point of the cube centered on the virtual camera position that is linearly approximated by calculation) during the movement of this camera, and the error The fewest, that is, the virtual photograph position closest to the photographed photograph position is searched. The virtual position and orientation of the camera at that time are regarded as the position and orientation of the camera when the aerial photograph was taken. There is no information about the position and orientation of the camera in the aerial photograph, but if the positions of multiple sample points are close to each other, the position and orientation of the camera in the unknown aerial photograph will be unknown. It can be regarded as the same as that of a known camera.

Here, an area on the map plane where an aerial photograph is taken is subdivided into a plurality of predetermined areas, for example, in a grid pattern, and a memory in which storage elements are associated with the plurality of subdivided points is prepared. This memory is called a distortion correction aerial photograph memory.
This subdivision is carried out so that the distortion-corrected aerial photography memory has sufficient resolution to reproduce the aerial photography image data.

Further, the reference point is set to be smaller than the number of the plurality of points, and the altitude value of the point is recorded.
Since the elevation value of this reference point is not given for all the regions on the map, the elevation values of a plurality of points subdivided into a grid are obtained as described above. This altitude value is obtained as an interpolated altitude value by interpolating the altitude value at the sample point. This interpolation can be realized by Lagrange interpolation.

Next, it is assumed that the surface of the interpolated elevation value at each point is photographed by the camera considered to be in the above-mentioned position and orientation. The ground surface at each point is converted into a film coordinate system as a distortion-corrected photographic position by the conversion formula described above. As described earlier, the aerial photograph is obtained as image data, and the image data at the distortion corrected photograph position is read.
The read image data is obtained by photographing the surface of the interpolated altitude value at each point, and is stored in the distortion correction aerial photograph memory provided corresponding to each point at which the interpolated altitude value is obtained. Since the memory for distortion correction aerial photography corresponds to storage elements at multiple points that subdivide the area on the map, the terrain represented by the image data stored there is an orthographic projection similar to the map representation. It will be what you did. In this way, the image data of the aerial photograph is distortion-corrected to the orthographic projection and stored in the distortion-corrected aerial photograph memory, and the geospecific texture pattern is generated.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a method for generating a geospecific texture pattern according to the present invention will be described below with reference to the drawings. An example will be described below on the assumption that the altitude at which an aerial photograph is taken is about 1000 m. FIG. 1 shows a procedure for generating a geospecific texture from an aerial photograph and necessary data and files, which are executed by a processor (not shown). In FIG. 1, (P1) represents the process of digital sampling of aerial photographs and maps. A plurality of aerial photographs of the target area (gaming area) and a map of the area (1: 25000 in the embodiment) are sampled with a scanner at a detail level of 100 dot / inch, and a photo file and a map file are sampled. create.

In FIG. 1, (P2) shows EWS (E) with the information necessary to search the position and orientation of the camera (mounted on the aircraft) that photographed each photo file.
ngineering Workstation: Shows the process of giving on the screen of a small personal computer, etc., by which the designer himself can interactively proceed with the design work by the man-machine interface with excellent interactivity. here,
The information is the following (a) to (d).

(In the present specification, vector display, for example, the vector r is displayed as <r> or displayed in white. The vector in each figure is displayed in bold or white.)

(A) Pixel coordinates <r> _{photorg [n]} (n = 1,2,3,4) of the standard points at the four ends of the aerial photograph in which the contents of the photograph file are displayed on the EWS screen. Here, the pixel coordinates are coordinates provided on the EWS screen.

(B) Pixel coordinates <r> of arbitrary two points on the map displaying the contents of the map file on the EWS screen
_{maporg [n]} (n = 1, 2) and the latitude and longitude of each point.
Here, the latitude and longitude are represented by the surface coordinate system.

(C) Pixel coordinates of arbitrary 5 sample points on the aerial photograph <r> _{photcrsp [n ]} (n = 1, 2,
3, 4, 5).

(D) Pixel coordinates <r> _{mapcrsp [n]} of points on the map corresponding to each sample point specified in (c)
(N = 1, 2, 3, 4, 5) and the elevation value of each point.

These pieces of information are used to specify a screen with a mouse,
And, for example, by inputting a numerical value by keyboard operation, it is saved in the photo file and the map file. FIG. 2 shows a window displayed on the screen of the EWS for making these inputs. The photo display window 21 and the map display window 22 each have a display area of the same size,
The upper left of each is the origin of pixel coordinates. When the operator points the sample point in (c) on the photograph in the photograph display window 21 with the mouse, the pixel coordinates <r> _{photcrsp [n]} are measured by the processing of the processor (not shown ₎ . This is the process of (P2.1) in FIG. 1, and the position measured by this is taken as the real photograph position. Further, as in (d) above, the map display window 2 is viewed while looking at the point on the photo screen designated in the photo display window 21.
Find the point corresponding to the sample point from the map of 2 and point it with the mouse, and the pixel coordinate <r>
_{mapcrsp [n]} is measured. At this time, the altitude value of the designated sample point is key-input while referring to the display of the numerical value input window 25. This is the process of (P2.2) in FIG. 1, and the position measured by this is the ground surface position. Other windows displayed on the EWS screen include a file name input window 23 and an alert display window 24.

In FIG. 1, (P3) is the pixel coordinate <r> of the sample point on the photograph input in (P2).
_This is the process of searching the position and orientation of the camera that took the aerial photograph from _{photcrsp [n]} and the point <r> _{mapcrsp [n]} on the map. The process flow of this process is shown in FIGS.

FIG. 5 is a diagram for explaining the projection of points on the ground surface onto a film when the ground surface is photographed by a camera at an arbitrary position and posture. At this time, the coordinate conversion of the points on the surface of the surface coordinate system to the film coordinate system is performed by the following equation (1).

[0027]

[Equation 1]

In the above formula (1), T is represented by the formula (2),

[0029]

[Equation 2]

And also

<R> _f : the position vector in the film coordinate system of the point projected on the film,

Z _oc : Z component of <r> _oc ,

D _f : focal length of the camera (described in aerial photography),

<R> _ow : a position vector of a point on the ground surface in the ground coordinate system,

<R> _cw : camera position vector in the surface coordinate system,

[0036] _{<Z> c / Z c:} Zc in the camera coordinate system
Axial unit vector,

T: Rotation conversion matrix according to the attitude of the camera mounted on the aircraft,

Ψ: rotation angle around the Z axis of the camera mounted on the aircraft,

Θ: rotation angle of the camera mounted on the aircraft around the Y axis,

Φ: The rotation angle of the camera mounted on the aircraft around the X axis.

In FIGS. 3 and 4, (S1) to (S)
In 5), the position <r> _cw of the virtual camera is roughly estimated. In (S1), the camera position search center <r> _cwo is set. This setting is performed by converting the pixel coordinates <r> _{mapcrsp [n]} on the map corresponding to a plurality of sample points from the pixel coordinates to the ground coordinate system and taking the average value thereof.
The conversion from the pixel coordinates of this <r> _{mapcrsp [n]} to the surface coordinate system is based on the latitude, longitude and pixel coordinates of any two points on the map described above, using the <r> _{maporg [n]} as a reference point. It is converted to the map, that is, the ground coordinate system, by the difference in the pixel coordinates of. (S2) is the setting of the attitude search center of the camera mounted on the aircraft, and ψ _o = 0, θ _o = 0, φ _o =
Set to 0. In (S3), the rotation conversion matrix T at the search center is calculated by the equation (2). (S4), (S5)
Is the position of the camera or the aircraft carrying it, d
While virtually moving at intervals of <r> _cw , the coordinates of the sample points on the map are transformed by the formula (1), and the difference from the coordinates of the sample points on the photograph is found to be <r> _cw . It is a process. The movement of the camera position is centered around the search center,
This is done inside a cube measuring 4000m on a side. Where d <
Let r> _cw = (100m, 100m, 100m). That is, a cube with a side of 4000 m is divided into 40 × 40 × 40 cubes, and parameters k, l, m (integers of -20 to 20) are specified to specify this small cube. Then, in (S4.1), the position of the camera is set by the equation (3).

[0042]

(Equation 3)

In (S4.2), the sampling point <r> on the map
<cr> _{f [n]} is calculated by converting the _{mapcrsp [n]} into film coordinates by the formula (1). In this case, <r> _{f [n]} is a virtual photograph position because it is a point taken by the virtual camera.
In (S4.3), the above (S4.3.
2) The difference between the calculated value <r> _f obtained by coordinate conversion in step 2) and the measured value on the photograph (pixel coordinates of the information (d) of (P2) in FIG. 1) <r> _{photcrsp [n ]} is _calculated. , The position of the camera <r> for the sample point with the larger squared value
_{The difference in cw [k] [l] [m]} is D [k] [l] [m].

Next, in (S5), among the parameter sets {k, l, m} corresponding to the respective points, the set {k _o , which minimizes the difference D [k] [l] [m] is set. Search for l _o , m _o }.

In this way, the camera position <r> _cw is roughly estimated in (S1) to (S5), and the position is _{calculated as} the difference D [k] [l].
It has been determined that the set {k _o , l _o , m _o } that minimizes [m].

Next, in (S6) to (S12), a more accurate search is performed to determine the position and orientation of the camera. The position search center is (S6) and <r> _cwo = <r>
_{cw [ko] [lo ] [mo]} . That is, <r> _{cw [ko] [lo] [mo]} searched in (S5) above is used as a new search center <r> _cwo.
Reset as.

In (S7), the search interval for position and rotation angle is set. Rotation angle dψ = dθ = dφ = 2de
g, position d <r> _wc = (50m, 50m, 50
m).

In the process of (S8), while moving at intervals of d <r> _cw and changing the posture angles ψ, θ, and φ, the calculation result of the coordinate conversion of the equation (1) and the measured value on the photograph ( Pixel coordinates of information (d) of (P2) in FIG. 1 <r
> _{Mapcrsp [n]} ). The posture angle is ± 20 × 2 deg depending on the integers given to the parameters j, k, and l.
Changes by 2 deg, and the position is changed by parameters m1, m2,
By the integer given to m3, the vertical and horizontal heights are moved by 50 m each by ± 20 × 50 m from the reset position.

In the process of (S8.1), the position of the camera and the rotation angle are set by the equation (4). That is,

[0050]

[Equation 4]

It is

At (S8.2), the angles ψ, θ, and φ set as described above are given to the equation (2), and the rotation matrix T is calculated. Further, in (S8.3), the sample point <r> _{mapcrsp [n]} on the map is coordinate-converted into film coordinates by the equation (1), and <r> _{f [n]} is calculated. <R in this case
> _{F [n]} is a virtual photograph position because it is a point taken by the virtual camera. In (S8.4), for each sample point, the difference between the value <r> _f obtained by coordinate conversion in (S8.3) and the pixel coordinate <r> _{photcrsp [n]} on the photograph is calculated, and the difference is _calculated. A sample point having a large squared value is defined as a difference D [j] [k] [l] [m1] [m2] [m3] at the camera position <r> _cw .

Next, in (S9), the difference D [j] [k] among the parameter sets {j, k, l, m1, m2, m3} is used.
The set {j _o , k _o , l _o , m that minimizes [l] [m1] [m2] [m3]
Search for 1 _o , m2 _o , m3 _o }. In (S10), it is determined whether or not the retrieved difference D [j _o ] [k _o ] [l _o ] [m1 _o ] [m2 _o ] [m3 _o ] is within the allowable error range, and the allowable difference is determined. Within the margin of error ie

[0054]

[Equation 5] D [j _o ] [k _o ] [l _o ] [m1 _o ] [m2 _o ] [m3 _o ] ≦ DMAX (5)

If r is satisfied, then <r> _cw , ψ,
Let θ and φ be the position and orientation of the camera when taking a picture.

When the above equation (5) is not satisfied,
As in (S11), the search center of the position and the rotation angle is reset. That is, [j _o ] [k _o ] searched in (S9) above.
The position and posture angle determined by [l _o ] [m1 _o ] [m2 _o ] [m3 _o ] are set as a new search center as in Expression (6).

[0057]

(Equation 6)

In step S12, the position and rotation angle search intervals are reset. Rotation angle dψ = dψ / 2, dθ =
dθ / 2, dφ = dφ / 2, and position d <r> _wc =
A finer change is made as d <r> _wc / 2. After setting in this way, the process returns to (S8) and is the same as above (S10).
Proceed to.

As described above, the difference D [j _o ] [k _o ] in (S10).
If [l _o ] [m1 _o ] [m2 _o ] [m3 _o ] is determined to be within the tolerance, this [j _o ] [k _o ] [l _o ] [m1 _o ] [ The position and posture angle determined by m2 _o ] [m3 _o ] are set as the position and posture angle at which the aerial photograph was taken, and the process proceeds to the distortion correction of (P4) in FIG. (P4) is
The position and orientation angle of the camera obtained in (P3) <r> _cw , ψ,
This is a process of correcting the distortion of the aerial photograph using θ and φ. Since the aerial photograph is a central projection as shown in FIG. 6, it is distorted by the elevation of the ground surface and is not an orthographic projection like a map. Therefore, the points of the aerial photograph are orthographically projected on the map to correct the distortion. As will be described later, a distortion-corrected aerial photograph memory 61 in which storage elements are associated with a plurality of predetermined points in a region on a map plane where an aerial photograph is taken is prepared,
The image data of the points of the aerial photograph after the distortion correction is stored in the distortion-corrected aerial photograph memory 61. Here, the distortion-corrected aerial photograph memory 61 is composed of a two-dimensional array of storage elements capable of recording a certain color, and this storage element is called a texel. The number of texels in the distortion-corrected aerial photography memory is about 1000 × 1000 or more. Each texel stores the intensity of red, green, and blue, which are color components, in 8 bits each. In FIG. 6, reference numeral 62 is a film.

Regarding the processing of (P4), X in the ground coordinate system of the map corresponding to the topography where the aerial photograph is taken.
In the w and Yw planes, in both the Xw axis direction and the Yw axis direction,
A mesh type elevation data group in which elevation values of reference points set at equal intervals (100 m) are recorded is prepared. FIG. 7 is a diagram schematically showing this elevation data group. Reference numeral 71 is a mesh that divides the ground surface into stitches, and 72 is elevation data of a reference point.

FIG. 8 shows a processing flow of distortion correction of (P4). In FIG. 8, (T1) is a process of searching an effective area in which coordinate conversion is performed in the surface coordinate system. The mesh elevation data is subjected to coordinate conversion by the formula (1) to search for an area that corresponds to the effective range of the photograph, that is, an effective area. The effective range of the photograph is a rectangle connecting standard points at the four ends of the photograph. The purpose of this search is to correct the distortion of one aerial photograph by using a distortion correction aerial photograph memory for storing the image data after the distortion correction in an appropriate area of the ground coordinate system Xw, Yw plane. That is, this is to correspond to the effective area. The effective area of this surface coordinate system is the area shown in this aerial photograph, and if the point <r> _ow in this area is coordinate-converted, it will be mapped within the effective range of the photograph. > When the coordinate conversion of _ow is performed, it is meaningless because it is mapped outside the effective range of the photograph (outside the frame of the photograph). Therefore, the effective area is determined by converting the altitude data around the camera position by the equation (1) and checking whether it is mapped within the effective range of the photograph or outside the effective range of the photograph. . When the effective area is determined, the distortion-corrected aerial photograph memory is associated therewith (see FIG. 9), so that the texel addressed by (p, q) = (0, 0) of the distortion-corrected aerial photograph memory 61 is set. position vector at the corresponding ground coordinate system, southwestern edge of the effective area 91 obtained (the Y _w direction north, X _w direction and east.) becomes the coordinates, it is a _<r> memow. Further, the length of one side of the effective area 91, the value obtained by dividing becomes texel spacing d _texel in ground coordinate system texels number of p direction of the distortion correction aerial photographic memory 61.

In FIG. 8, (T2) is a process of calculating the rotation conversion matrix T from the attitude angle of the camera. This posture angle is obtained in (P3) and is substituted into the equation (2).

In FIG. 8, (T3) is the process of distortion correction, and in (T3.1), the texel (p, q) of the memory for distortion correction aerial photography is set to <r> _texelw in the ground coordinate system.
Convert to. The texel (p, q) is coordinate-transformed by the equation (7).

[0064]

(Equation 7)

The vectors of <r> _texelw and <r> _memow are two-dimensional. FIG. 10 shows the relationship between each vector and the conversion of texel (p, q) into the surface coordinate system.

In FIG. 8, (T3.2) is <r>.
_This is the process of performing Lagrange interpolation to give the elevation value in _texelw . Regarding the interpolation of the elevation value, FIG.
This will be described with reference to FIG. In FIG. 11, the mesh 71
The intersection is a point where the elevation data is known, that is, a reference point. The point R is a point where the elevation value is to be interpolated, and is a point of <r> _texelw corresponding to each texel. First, using the elevation data, the Lagrange interpolation formula is applied in the X direction and the Y direction, and the interpolation intermediate values A1, A2, A3, A4, B1, B
Calculate 2, B3, B4. The elevation value ZA of the point R is obtained from A1, A2, A3, A4 by Lagrange interpolation. B1,
Similarly, the elevation value ZB of the point R is obtained from B2, B3, and B4,
Let (ZA + ZB) / 2 be the elevation value of point R. The Lagrange's interpolation formula uses the following formula (8).

[0067]

(Equation 8)

Where q _n is the X or Y coordinate value of a point of known altitude,

Z (q _n ): elevation value at the point q _m ,

A: X or Y coordinate value for which altitude is to be obtained

It is

In FIG. 1, (P5) is a process of joining aerial photographs after distortion correction. In order to give a geospecific texture pattern for the entire gaming area, it is necessary to connect dozens of distortion-corrected photographs.
These dozens of distortion-corrected photographs are stored in respective files, and these files are combined into one continuous photograph file so that the ground coordinates associated with the texels of the distortion-corrected aerial photograph memory are continuous.

(P6) is a process of dividing the connected distortion-corrected photographs into geospecific texture files. This file is a file that stores a geospecific texture pattern, and the data stored in this file is used as a texture pattern in a video processor. Therefore, this geospecific texture file has a format required by the video processor for mapping. Further, since the pseudo-visual field generator geographically divides the gaming area into several blocks and holds data for each block, it is necessary to divide the geospecific texture pattern corresponding to this block. Since the photo file after distortion correction does not support this block, (P
After performing the connection in 5), it is divided into blocks again in (P6).

[0074]

As described above, according to the present invention,
Searching for the camera position where the difference between the positions of the multiple sampling points on the photograph and the positions of the coordinates of the multiple sampling points on the map assumed to be captured by the camera assumed to move on the map is the smallest. With, it is possible to know the position and orientation of the camera that took the aerial photograph, and by interpolating the elevation values of the few reference points, the height information of a plurality of points greater than the reference points is obtained, and the points on the ground surface and the air are obtained. It is possible to convert the coordinates of the aerial photo into the aerial photo as if the image was taken by the camera with the above-mentioned fixed position and orientation by the coordinate conversion formula that associates the points on the photo with the image data of the aerial photo. Since it can be stored in the corresponding memory and orthographically projected, the distortion correction of the aerial photograph can be performed semi-automatically to generate the geospecific texture pattern.

[Brief description of drawings]

FIG. 1 is a diagram illustrating a procedure for generating a geospecific texture from an aerial photograph.

FIG. 2 is a diagram illustrating a window displayed on a screen of EWS.

FIG. 3 is a diagram illustrating a processing flow of a process of searching the position and attitude of an aircraft, showing a first half part.

FIG. 4 is a diagram illustrating a processing flow of a process of retrieving the position and attitude of an aircraft, and shows the latter half.

FIG. 5 is a diagram illustrating the projection of points on the ground surface onto a film when the ground surface is photographed by a camera in an arbitrary position and posture.

FIG. 6 is a diagram for explaining the central projection of the ground surface onto the film, and the orthogonal projection of the distortion-corrected aerial photography memory corresponding to the map plane and the ground surface.

FIG. 7 is a diagram illustrating a mesh type elevation data group in which elevation values at reference points are recorded.

FIG. 8 is a diagram illustrating a processing flow of distortion correction.

FIG. 9 is a diagram illustrating the correspondence between the effective area of the surface coordinate system and the distortion correction aerial photograph memory.

FIG. 10 is a diagram showing the conversion of texels (p, q) into the surface coordinate system and the relationship between the vectors.

FIG. 11 is a diagram illustrating interpolation of an altitude value.

FIG. 12 is a diagram illustrating two-dimensional Lagrange interpolation.

[Explanation of symbols]

P1 ... Digital sampling process, P2 ... Photo map display process, P3 ... Attitude search process, P4 ... Distortion correction process.

Claims (1)

[Claims] 1. A method for generating a geospecific texture pattern from an aerial photograph, wherein an aerial photograph is obtained as image data, and a plurality of arbitrary sample points on the aerial photograph displayed on the screen are indicated on the screen by a mouse. The first step of obtaining the position as a photographed image position, and the position on the map that is displayed on the same screen as the screen at the same time and is pointed by the mouse so as to correspond to the arbitrary multiple sample points And a second process for obtaining a ground surface position including the above, and a second position corresponding to the arbitrary plural sample points by virtually changing the position and orientation of the virtual camera within a predetermined range.
The ground surface position in the process of, by the conversion formula for converting the points on the ground surface in the ground surface coordinate system when taken by the camera into the film coordinate system, coordinate conversion as a virtual photograph position,
A virtual photo position closest to the real photo position is searched, and the virtual camera position and posture at that time are regarded as the camera position and posture when the aerial photo was taken.
And a memory for distortion correction aerial photography in which memory elements are arranged corresponding to multiple points on the surface coordinate system, and the elevation value of the reference point set to be less than the number of the multiple points on the surface coordinate system is calculated by Lagrange interpolation. Interpolation elevation values of the plurality of points are obtained by interpolation, and the conversion is performed by assuming that the ground surface positions of the obtained interpolation elevation values of the plurality of points are taken by the camera in the position and posture considered in the third step. A distortion-corrected photograph position is converted into a film coordinate system using a formula, and the image data in the first step at the distortion-corrected photograph position is stored in the distortion-corrected aerial photograph memory corresponding to the plurality of points. 4. The method for generating geospecific texture, comprising the steps of 4).