JPH07174555A - Apparatus and method for photographic survey - Google Patents

Abstract

PURPOSE:To easily properly detect three-dimensional shape of an object by preventing miss matching. CONSTITUTION:A correlation window control part 20 is provided which can set a correlation window to a first photographic image stored in a photographic image memory 9 based on an optional matching reference point. A density value-detecting part 3 can detect a distribution state of brightness of a plurality of pixels within the correlation window for each of three chromaticities, and a search window control part 21 can set a search window corresponding to the correlation window to a second photographic image stored in the memory 9 in a manner to move on the second photographic image. Moreover, a density value-detecting part 5 which can detect a distribution state of brightness of a plurality of pixels in the search window for each of three chromaticities, is provided. A matching point-detecting part 6 detects a correlation coefficient between the distribution state of brightness for every chromaticity in the correlation window and that within the search window, thereby detecting a point corresponding to the matching reference point. A three-dimensional position-operating part 13 can detect a three-dimensional position SI of an object based on the matching reference point and the matching corresponding point. This apparatus is constituted of these parts.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a photogrammetric apparatus and a photogrammetric method suitable for detecting the three-dimensional shape of a surveyed object.

[0002]

2. Description of the Related Art Conventionally, a digital still camera capable of recording a stereoscopically photographed image as digital information,
Photogrammetry capable of quickly and easily detecting the three-dimensional shape of an object to be surveyed with a photogrammetric device such as EWS (engineering work station) capable of analyzing the digital information to analyze a three-dimensional shape. A method has been proposed. In the conventionally proposed photogrammetry method, first, a surveying object such as a building is stereo-photographed from two points, and a plurality of points (pass points) commonly photographed in two photographed images are obtained. Based on these two
The relative positional relationship between two photographic images at the time of shooting is determined (mutual orientation). Next, from a plurality of points (reference points) where the latitude, the longitude, the altitude, etc. are detected in advance at the points that were taken in common with these two photographic images, the two photographic images at the time of shooting were taken. Determine the relative positional relationship to the ground surface (ground location). As a result (that is, the result of the mutual orientation and the ground orientation), if the coordinate positions of the points corresponding to each other in the two photographic images are determined in each photographic image, the ground coordinates (latitude, longitude, A conversion formula in which one coordinate position at an altitude is determined is obtained. Then, a point in one of the photographic images where the position of the surveyed object in the ground coordinates is to be known is designated, and matching is performed by searching for a point in the other photographic image corresponding to this point. In this matching, one point of one photographic image is designated, and a region having a predetermined shape and area centering on the one point, that is, a correlation window is designated. Further, a search window having the same shape as the correlation window is set on the other photographic image so as to be movable on the image. Then, while moving the search window on the other photographic image, the respective image density values due to light and dark inside the correlation window and inside the search window are compared, and the search window when the similarity between both image density values is the highest. The center point of is the point on the other photographic image that corresponds to the one point on the other photographic image. Matching result, 2
Since it is possible to detect the coordinate position of each corresponding point on one photographic image plane, the coordinate position of the point to be obtained in ground coordinates can be obtained by the conversion formula.
By repeating the above matching, the three-dimensional shape of the survey target is detected.

[0003]

By the way, the comparison of the similarity between the image density values of the correlation window and the search window depends only on the image density value due to the light and dark irrelevant to the color (that is, the light and dark in the case of a black and white photograph). Therefore, it is likely that a plurality of search windows having image density values having a high degree of similarity with the image density value are present at a plurality of locations with respect to a correlation window having a certain image density value. It is evaluated that the image density values of the search windows at the positions that do not correspond correspond to each other, and mismatching is likely to occur. Mismatching makes it difficult to properly detect the three-dimensional shape of the surveyed object. The present invention has been made in view of the above circumstances, and an object thereof is to provide a photogrammetric apparatus and a photogrammetric method capable of facilitating proper detection of a three-dimensional shape of a surveyed object by preventing mismatching as much as possible.

[0004]

The first aspect of the present invention is a surveying object (60) existing in a three-dimensional space (CK).
A first photographic image (SG1) taken from a first position (PT1) of the three-dimensional space (CK) and the survey target (60) in the first three-dimensional space (CK). A three-dimensional shape (SI) in the three-dimensional space (CK) of the survey object (60) is obtained based on a second photographic image (SG2) taken from a second position (PT2) different from the position (PT1). In the photogrammetric device (1) for detecting, the photogrammetric device (1) has an image storage section (9) capable of storing the first photographic image (SG1) and the second photographic image (SG2), In the first photographic image (SG1) stored in the image storage section (9), the first photographic image (SG1)
1) A correlation window setting unit (20) capable of setting a correlation window (30) surrounding a predetermined area with reference to an arbitrary matching reference point (29A) is provided, and a plurality of correlation window setting units (20) are provided. Pixel (GS) brightness (α) distribution status (P1, P
2, P3) is provided for each of a plurality of types of chromaticity (R, G, B) of each pixel (GS), and a correlation window pixel distribution status detection unit (3) is provided, and the image storage unit (9) is provided. In the second photographic image (SG2) stored in, a search window setting unit (31) capable of setting a search window (31) corresponding to the correlation window (30) while moving on the second photographic image (SG2) ( 21) is provided, and the distribution status (Q1, Q2, Q3) of the brightness (β) of the plurality of pixels (GS) in the search window (31) is calculated for each pixel (G).
S) is provided with a search window pixel distribution status detection unit (5) capable of detecting each of a plurality of types of chromaticity (R, G, B), and each chromaticity (R, G, B) in the correlation window (30) ) For each luminosity (α) (P1, P2, P3) and for each chromaticity (R, G, B) within the search window (31) (Q) distribution (Q).
1, Q2, Q3) to determine the degree of similarity (γ) to detect the matching corresponding point (MT) of the second photographic image (SG2) corresponding to the matching reference point (29A) of the correlation window (30). Each matching photographic image of the matching reference point (29A) and the matching corresponding point (MT) corresponding to each other in the first and second photographic images (SG1, SG2). (SG1, SG
2) A three-dimensional shape detection unit (13) capable of detecting the three-dimensional shape (SI) in the three-dimensional space (CK) of the survey object (60) based on the position above is provided. The second invention of the present invention is the three-dimensional space (C
K) the survey object (60) taken from the first position (PT1) of the three-dimensional space (CK) and a first photographic image (SG1), and the survey object (60) Based on a second photographic image (SG2) taken from a second position (PT2) different from the first position (PT1) in the original space (CK), the three-dimensional image of the survey object (60). In the photogrammetric method for detecting a three-dimensional shape (SI) in a space (CK), the first photographic image (SG
1) A correlation window (30) surrounding a predetermined area is set on the basis of an arbitrary matching reference point (29A), and the brightness (α) of a plurality of pixels (GS) in the correlation window (30) is set. Distribution state (P1, P2, P3) is detected for each of a plurality of types of chromaticity (R, G, B) of each pixel (GS), and in the second photographic image (SG2), the correlation window ( The search window (31) corresponding to 30) is set so as to move on the second photographic image (SG2), and the search window (31) is moved by a predetermined movement amount (δ) every time the search window (31) moves. Distribution (Q1, Q2, Q) of brightness (β) of a plurality of pixels (GS) in the window (31)
3), a plurality of types of chromaticity (R,
G, B) for each chromaticity (R, G, B) within the correlation window (30) and distribution (P1, P) of lightness (α).
2, P3) and the lightness (β for each chromaticity (R, G, B) in the search window (31) detected each time the search window (31) moves by a predetermined movement amount (δ). ) Distribution situation (Q
1, Q2, Q3) to detect the similarity (γ),
The second photographic image (SG) of the search window (31) corresponding to the highest similarity (γ) among the detected similarities (γ)
From the position 2), the matching corresponding point (MT) of the second photographic image (SG2) corresponding to the matching reference point (29A) of the correlation window (30) is detected, and the first and second photographic images (SG1) are detected. , SG2) corresponding to the matching reference point (29A) and the matching corresponding point (MT).
Of the three-dimensional space (C) of the survey object (60) based on the position on each of the photographic images (SG1, SG2).
It is configured by detecting the three-dimensional shape (SI) in K). The numbers in parentheses () indicate the corresponding elements in the drawings for convenience, and therefore the present description is not limited to the description in the drawings. The same applies to the column of "action" below.

[0005]

With the above configuration, the correlation window (30) detected when performing matching in the form of detecting the position of the search window (31) including the image corresponding to the image in the correlation window (30).
The similarity (γ) between the image inside and the image inside the search window (31) is
When the photogrammetry device (1) of the first invention of the present invention is used, or when the photogrammetry method of the second invention is applied,
The brightness (α, α) for each chromaticity (R, G, B) possessed by a plurality of pixels (GS) in the correlation window (30) or the search window (31)
Distribution of β) (P1, P2, P3, Q1, Q2, Q
Detected by comparison of 3).

[0006]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a photogrammetric apparatus to which the present invention is applied, and FIG. 2 is a perspective view showing a positional relationship between a surveying object and first and second points where stereo imaging is performed. 3 is a conceptual input to the photogrammetric device of FIG.
A diagram showing reference coordinates KZ, camera coordinates CZ1 and CZ2,
4 and 5 are diagrams showing the first and second displays, and FIG. 6 is a diagram conceptually showing the matching procedure.

As shown in FIG. 1, the photogrammetric device 1 has a main control unit 2, and the main control unit 2 is provided with a correlation window density value detection unit 3 and a search window density value via a bus line 2a. Detection unit 5, matching point detection unit 6, display means 7, photographic image memory 9, mutual orientation calculation unit 10, ground orientation calculation unit 11, deviation correction calculation unit 12, three-dimensional position calculation unit 13, three-dimensional position memory unit. 15, output unit 16, input unit 17, lattice setting control unit 26, coordinate conversion control unit 19, correlation window control unit 20, search window control unit 21, cursor position recognition unit 22, projection conversion control unit 25, lattice setting control unit. 26 is connected. The correlation window density value detection unit 3 includes a correlation window red density value detection unit 3a, a correlation window green density value detection unit 3b, and a correlation window blue density value detection unit 3c.
The search window density value detection unit 5 includes a search window red density value detection unit 5a, a search window green density value detection unit 5b, and a search window blue density value detection unit 5c. Further, the display means 7 is provided with a display control section 7a and a work instruction display section 7b capable of displaying work instructions and commands to an operator (not shown) by characters or the like (not shown), and the display control section 7a. Has a first display 7c and a second display 7d
Are connected. The input means 17 is composed of an input operation section 17a and a cursor movement operation section 17b. Further, the photo image memory 9 is connected to the deck 23 into which the IC memory card 62a can be loaded.
The IC memory card 62a can also be loaded into the camera 62, which is a known color digital still camera. That is, a photographic image taken by the camera 62 loaded with the IC memory card 62a is recorded on the IC memory card 62a as image information configured by converting each pixel of the photographic image into digital information. Further, the digital information of each pixel of the photographed photographic image is optical density, that is, density value information relating to lightness, and the density value information of each pixel is red R, green G, blue B (that is, It consists of optical density information for each of the three chromaticities (three primary colors). Therefore, the digital information of one pixel is red R
Density information of green, density information of green G, and density value of blue B.

Since the photogrammetric device 1, the camera 62, and the like have the above-described configurations, the photogrammetric device 1 is used to survey the object 60 to be surveyed as follows. First, by using the camera 62 loaded with the IC memory card 62a, the survey object 60 such as a building or a road in the ground space CK
As shown in FIG. 2, images are taken from the first point PT1 and the second point PT2 which are two different points in the ground space CK (that is, stereo images are taken). The result of the shooting,
The first image information GJ1 and the second image information GJ2, which are digital information of each photographic image taken at the first point PT1 and the second point PT2, are recorded in the IC memory card 62a. It should be noted that the three-dimensional position in the ground space CK, that is, three reference points KJ1, KJ2, and KJ3 whose latitude, longitude, and altitude are known are previously detected and marked on the survey target 60, and the first point PT1 and These reference points KJ1 and KJ are obtained when photographing from any of the two points PT2.
2, KJ3 is photographed. Then, the first image information G
The IC memory card 62a on which J1 and the second image information GJ2 are recorded is removed from the camera 62 and loaded on the deck 23 of the photogrammetry apparatus 1. After loading, an operator (not shown) gives a command to start the analysis processing work via the input operation unit 17a of the input means 17. The instruction to start the analysis processing work is transmitted to the main control unit 2 via the bus line 2a, and the main control unit 2 receives the instruction and starts the analysis processing work. That is, first, the main control unit 2 commands the photographic image memory 9 to cause the deck 23 connected to the photographic image memory 9 to load the IC memory card 62a loaded in the deck 23 as shown in FIG. First image information GJ1 and second image information G
Read and store J2.

As shown in FIG. 3, a three-dimensional reference coordinate KZ consisting of the Xk axis, the Yk axis and the Zk axis is input to the photographic image memory 9, and the Xk axis component of the reference coordinate KZ, Yk. The axis component and the Zk axis component are the ground space C, respectively.
It corresponds to the latitude, longitude, and altitude of K. Further, the photographic image memory 9 is provided with two camera coordinates CZ1 and CZ2 that can have an arbitrary positional relationship with respect to the reference coordinates KZ. The camera coordinates CZ1 are the Xc1 axis, the Yc1 axis, and Zc.
The three-dimensional coordinate consisting of one axis, the camera coordinate CZ2 is Xc2.
It is a three-dimensional coordinate including an axis, a Yc2 axis, and a Zc2 axis. The camera coordinates CZ1 and CZ2 are the same fixed coordinates with respect to the camera 62, but the positional relationship between the ground space CK and the camera 62 when the image is taken from the first point PT1 is the position of the reference coordinates KZ and the camera coordinates CZ1. The relationship is shown, and the ground space CK and the camera 62 when taken from the second point PT2
The positional relationship between the reference coordinates KZ and the camera coordinates CZ2 can be shown. Also, the camera coordinate C
Of Z1 and CZ2, the plane in which the Zc1 axis component is represented by zero and the plane in which the Zc2 axis component is represented by zero are image planes GH1 and GH2 on which the photographic image photographed by the camera 62 is formed. ing. Therefore, when the photographic image memory 9 reads and stores the recorded first image information GJ1 and second image information GJ2 from the IC memory card 62a, the first image information GJ1 is stored in the image plane GH1 at the camera coordinate CZ1. Is input as the first photographic image SG1 and the second image information GJ2 is read and stored in the image plane GH2 of the camera coordinate CZ2 as the second photographic image SG2 (note that the reference coordinate KZ and the camera coordinate C
The relative positional relationship between the three members, Z1 and CZ2, has not yet been determined. ).

Upon completion of reading and storing in the photographic image memory 9, the photographic image memory 9 stores camera coordinates CZ1, CZ2 and camera coordinates C after the reading and storing is completed.
The first photo image SG1 on Z1 and the second photo image SG2 on the camera coordinate CZ2 are reported to the main control unit 2 by electric signals. The main controller 2 controls the camera coordinate CZ1 and the first photographic image SG1, and the camera coordinate CZ2 and the second photographic image SG2.
Is transmitted to the coordinate conversion control unit 19 and the first photographic image SG1 is transmitted.
And the second photographic image SG2 is coordinate-converted to be displayed on the first display 7c and the second display 7d. By the way, as shown in FIG. 4, plane coordinates HZ1 composed of the X1 axis and the Y1 axis are set on the first display 7c, and the X2 axis and the Y2 axis are set on the second display 7d.
A plane coordinate HZ2 composed of an axis is set. Therefore,
In the coordinate conversion control unit 19, an arbitrary point on the image plane GH1 in the camera coordinate CZ1 is set as the Xc1-axis component of the arbitrary point as the X1-axis component of the plane coordinate HZ1 and the Yc1-axis component as Y1.
The first photographic image S on the camera coordinate CZ1 is converted by the method of replacing each with the axial component and converting into the point of the plane coordinate HZ1.
G1 is transformed on the plane coordinate HZ1. Similarly, in the second photographic image SG2 on the image plane GH2 of the camera coordinates CZ2, an arbitrary point on the camera coordinates CZ2 is converted into an X2 axis component of the plane coordinates HZ2 by the Yc axis.
By replacing the 2-axis component with the Y2-axis component and converting it into the point of the plane coordinate HZ2, the camera coordinate CZ2
The second upper photographic image SG2 is converted to the plane coordinate HZ2. After the conversion, the main control unit 2 transmits both photographic images SG1 and SG2 of the coordinate conversion control unit 19 to the display control unit 7a of the display means 7, and the display control unit 7a displays the first photographic image SG1 and the second photographic image SG2. They are displayed on the plane coordinates HZ1 of the first display 7c and on the plane coordinates HZ2 of the second display 7d, respectively.

After the display on the first display 7c and the second display 7d is completed, the main control unit 2 starts the mutual orientation work by a known method. That is, the main controller 2
As described above, the reference coordinates KZ, the camera coordinates CZ1 and CZ2 read and stored in the photographic image memory 9, the first photographic image SG1 on the camera coordinates CZ1, and the camera coordinates C.
The second photographic image SG2 on Z2 is transmitted to the relative orientation calculation unit 10, and the main control unit 2 instructs the work instruction display unit 7b to instruct the input of the pass point (not shown). To perform. An operator (not shown) visually compares the first photographic image SG1 and the second photographic image SG2 displayed on the first display 7c and the second display 7d by reading the display on the work instruction display unit 7b, and the first photographic image is displayed. Five pass points PPa1-P from SG1
For Pa5, five pass points PPb1 to PPb5 are searched for from the second photographic image SG2. In addition, pass point P
Pa1 (PPa2 to PPa5) is one point (that is, one pixel GS) in the first photographic image SG1, the pass point PPb.
1 (PPb2 to PPb5) is one point (that is, one pixel GS) in the second photographic image SG2, and the corresponding pass point PPa1 (PPa2 to PPa5) and pass point P.
Pb1 (PPb2 to PPb5) together with survey object 60
It is possible to clearly determine that the same one point DP1 (DP2 to DP5) in the inside is photographed. Five pass points P
After finding out Pa1 to PPa5 and the five pass points PPb1 to PPb5 corresponding to these, the operator (not shown) makes pass points PPa1 corresponding to each other.
(PPa2-PPa5), PPb1 (PPb2-PPb
5) are paired and input sequentially. As shown in FIG. 4, the first display 7c and the second display 7d are provided with movable cursors 33 and 35 on the respective displays 7c and 7d via the cursor movement operation unit 17b of the input means 17. And the pass point PPa1
(PPa2-PPa5), PPb1 (PPb2-PPb
When inputting 5) as a pair, the cursors 33 and 35 are moved on the respective displays 7c and 7d via the cursor moving operation unit 17b, and the cursor 33 is moved to the pass point PPa1 (one of the pixels GS). PPa2-P
Pa5) and position the cursor 35 at the pass point P
Position to Pb1 (PPb2 to PPb5). Signals of completion of positioning of the cursors 33 and 35 from the cursor movement operation unit 17b are transmitted to the cursor position recognition unit 22, respectively, and the cursor position recognition unit 22 positions the cursors 33 and 35, that is, the pass point PPa1.
Coordinate positions of (PPa2 to PPa5) on the plane coordinate HZ1 and the pass point PPb1 (PPb2 to PPb5).
The coordinate position on the plane coordinate HZ2 of is recognized, and the recognized plane coordinate HZ1 and the coordinate position on the plane coordinate HZ2 are transmitted to the coordinate conversion control unit 19 and the pass point coordinate position Pa1 (at the camera coordinate CZ1). Pa2-Pa5)
And the path point coordinate position Pb1 (Pb2 to Pb5)
Is converted into a signal, and after conversion, it is transmitted to the main control unit 2. The main control unit 2 controls the pass point coordinate position Pa1 (Pa2 to Pa).
5) and the corresponding path point coordinate position Pb1
(Pb2 to Pb5) are transmitted to the mutual orientation calculation unit 10 in the form of a pair.

Next, the main control unit 2 includes the mutual orientation calculation unit 1
Causes 0 to perform a relative orientation operation. By the way, the photographed first photographic image SG1 and second photographic image SG2 are located at predetermined positions (on the Zc1 axis and the Zc2 axis in this embodiment) of the camera coordinates CZ1 and CZ2 from the respective points on the survey object 60. It can be considered that the images of the surveying object 60 are projected on the image planes GH1 and GH2 by the light beams that travel straight in a convergent manner toward the set projection center points O1 and O2, respectively. Therefore, conversely, the two projection center points O1 and O
2 is provided with a light source, and the first photographic image SG1 and the second photographic image SG2 are projected, the respective photographic images S
All the two light fluxes that have passed through the corresponding points in G1 and SG2 (that is, the points where the same point in the surveyed object 60 is photographed) should meet. Therefore, the projection center point O1 and the camera coordinates are applied by applying the known principle of projective geometry that, if at least five light fluxes of the corresponding light fluxes of the two stereoscopic pictures meet, the corresponding points all meet. Under the condition that the straight line connecting the path point coordinate positions Pa1 to Pa5 on the CZ1 and the straight line connecting the projection center point O2 and the pass point coordinate positions Pb1 to Pb5 on the camera coordinates CZ2 all meet, The relative inclination between the coordinate CZ1 and the camera coordinate CZ2 is calculated (that is, the mutual orientation calculation) and determined. The relative inclination of the calculated and determined camera coordinates CZ1 and CZ2 at the reference coordinate KZ is the relative inclination of the camera 62 in the ground space CK in the state of being photographed at the first point PT1 and the second point PT2, respectively. It corresponds to. In this way, the relative tilt of the camera 62 in the ground space CK in the state where images are taken at the first point PT1 and the second point PT2, respectively, is reproduced as a relative positional relationship between the camera coordinates CZ1 and the camera coordinates CZ2 at the reference coordinates KZ. Because it was done
As described above, in these camera coordinates CZ1 and camera coordinates CZ2, if light sources are respectively provided at the two projection center points O1 and O2 to project the respective photographic images SG1 and SG2, as shown in FIG. Image SG1,
In SG2, all the two light fluxes that have passed through the corresponding points meet, and it is possible to virtually form a model 61 that is relatively similar to the survey object 60 by the set of points that meet. This completes the mutual orientation work. After completion,
The main coordinates of the reference coordinates KZ, the camera coordinates CZ1 and CZ2 whose relative inclinations are determined by the mutual orientation calculation unit 10, and the first photographic image SG1 on the camera coordinates CZ1 and the second photographic image SG2 on the camera coordinates CZ2 are controlled. It is transmitted to the ground orientation calculation unit 11 via the unit 2.

At the same time as the transmission, the main control section 2 starts the ground location work by a known method. First, the main control unit 2 instructs the work instruction display unit 7b to perform a display (not shown) for instructing the input of the reference point. An operator not shown is
By looking at the display, the three reference points KJ1 to KJ3, which are marked in advance as described above, are the reference point images KJa1 to KJa1 captured in the first photographic image SG1.
KJa3, the reference point images KJb1 to KJb3 captured in the second photographic image SG2 are displayed on the first display 7c.
The first photographic image SG1 of No. 2 and the second photographic image SG2 of the second display 7d are searched for. After searching, the operator (not shown) moves the cursors 33 and 35 to the respective displays 7c and 7d via the cursor movement operation unit 17b.
The reference point image KJa1 (KJa2, KJa3) and the reference point image KJb1 (KJb2, KJb3) each made up of one corresponding pixel GS are sequentially positioned.
Along with the positioning, the operator (not shown) operates the reference point image K.
Ja1 (KJa2, KJa3) and corresponding reference point image K
The latitude, longitude, and altitude of Jb1 (KJb2, KJb3) detected in the ground space CK are set as the reference point coordinate position Kc1 (Kc2, Kc3) which is the coordinate position on the reference coordinate KZ corresponding to the ground space CK. Inputs are sequentially made via the input operation unit 17a of the input means 17. On the other hand, cursor 3
By positioning 3 and 35, the cursor position recognition unit 22 causes the reference point images KJa1 (KJa2, KJa).
3) The coordinate position on the plane coordinate HZ1 and the reference point image K
The coordinate position of the Jb1 (KJb2, KJb3) on the plane coordinate HZ2 is recognized, and the coordinate position on the plane coordinate HZ1 and the coordinate position on the plane coordinate HZ2 are further transmitted to the coordinate conversion control unit 19, The reference point image coordinate position Ka1 (Ka2, Ka3) at the camera coordinate CZ1 and
Reference point image coordinate position Kb1 (K
b2, b3) and transmitted to the main control unit 2.
The main controller 2 controls the reference point image coordinate position Ka1 (Ka2, Ka2, Ka2
3), the reference point image coordinate position Kb1 (Kb2, b3), and
Reference point coordinate position K input via the input operation unit 17a
The c1 (Kc2, Kc3) is transmitted to the ground orientation calculating section 11.

Next, the ground orientation calculation section 11 performs the ground orientation calculation. That is, the reference point image coordinate position Ka1 (Ka2, Ka3) on the camera coordinate CZ1 or the reference point coordinate position Kc1 (Kc2, Kc3) corresponding to the reference point image coordinate position Kb1 (Kb2, b3) on the camera coordinate CZ2 is ,
It is a virtual position on the model 61 obtained as a result of the mutual orientation work, and the position on the model 61 is determined relatively to the camera coordinates CZ1 and CZ2 as described above. Therefore, the reference point coordinate position Kc1 (Kc2,
Since Kc3) is a point on the reference coordinates KZ, the positional relationship between the model 61 and the reference coordinates KZ is specified, and the positional relationship between the reference coordinates KZ and the camera coordinates CZ1 or the camera coordinates CZ2 is specified. As described above, the model 61 specified on the reference coordinates KZ accurately corresponds to the actual survey object 60, and the positional relationship between the model 61 and the camera coordinates CZ1 and CZ2 is also specified. Therefore, if the coordinate positions of the points corresponding to each other in the camera coordinates CZ1 and CZ2 are determined, one point on the model 61 corresponding to these coordinate positions is determined in the form of the coordinate position in the reference coordinate KZ. . Therefore, the ground orientation calculation unit 1
1 constructs a predetermined conversion formula HS1 for converting each coordinate position of the corresponding points in the camera coordinates CZ1 and CZ2 into a three-dimensional coordinate position SI of the point in the reference coordinates KZ of the model 61, and the ground orientation calculating unit 11 Causes the conversion formula HS1 to be transmitted and stored in the three-dimensional position calculation unit 13 via the main control unit 2.

Next, the main control unit 2 has a ground orientation calculation unit 11
Further, the reference coordinates KZ, the camera coordinates CZ1, CZ2, and the first photographic image SG1 and the second photographic image SG2, whose relative positional relationships are determined with each other, are transmitted to the eccentricity correction calculation unit 12, and the eccentricity correction is performed. The calculation unit 12 is caused to perform the deviation correction work by a known method. That is, the correction planes SH1 and SH2 (where the projection center O
(The straight line connecting 1 and O2 and the correction planes SH1 and SH2 are parallel to each other) are set to the camera coordinates CZ1 and CZ2, respectively, as shown by the chain double-dashed line in FIG.
The first photographic image SG1 and the second photographic image SG2 on the image planes GH1 and GH2 are converted into correction planes SH1 and SH2 by converting points on the image planes GH1 and GH2, respectively.
Convert to H2. The conversion of the points on the image planes GH1 and GH2 into the correction planes SH1 and SH2 is performed by the image plane GH.
The points 1 and GH2 are converted into the intersections of the light fluxes connecting the projection centers O1 and O2 with these points and the correction planes SH1 and SH2 (that is, known projection transformation). In addition, on these correction planes SH1 and SH2, the Xs1 axis,
Modified plane coordinates SHZ1 consisting of Ys1 axis, Xs2 axis,
A corrected plane coordinate SHZ2 consisting of the Ys2 axis is set.
However, the Xs1 axis and the Xs2 axis are set to be the same straight line on the reference coordinate KZ. That is, in the photographic images SG1 and SG2 on the corrected plane coordinates SHZ1 and SHZ2, parallax between the photographic images SG1 and SG2 occurs only in the Xs1 axis direction or the Xs2 axis direction. Further, the deviation correction calculation unit 12 uses the correction plane coordinates SHZ1 and SZ.
A predetermined conversion formula HS2 for converting an arbitrary point on HZ2 into a corresponding point on the camera coordinates CZ1 and CZ2 is calculated.

Next, the deviation correction calculation unit 12 corrects the plane coordinates SHZ1, SHZ2 and the plane coordinates SHZ1, SZ.
The photographic images SG1 and SG2 converted on the HZ2 and the conversion formula HS2 are transmitted to the main control unit 2. The main control unit 2 transmits the first photographic image SG1 on the corrected plane coordinates SHZ1 and the first photographic image SG2 on the corrected plane coordinates SHZ2 to the coordinate conversion control unit 19. The coordinate conversion control unit 19 sets an arbitrary point on the corrected plane coordinate SHZ1 to the Xs1 axis component of the arbitrary point as the X1 axis component of the plane coordinate HZ1, and sets the Ys1 axis component to Y.
It is a method to convert by replacing each one axis component,
The first photographic image SG1 on the corrected plane coordinates SHZ1 is converted into the plane coordinates HZ1 and an arbitrary point on the corrected plane coordinates SHZ2 is converted into an Xs2 axis component of the arbitrary point as X of the plane coordinates HZ2.
The two photographic images SG1 and SG2 are converted into two-axis components by converting the Ys2-axis components into the Y2-axis components.
Is transmitted to the display controller 7a. The display controller 7a is shown in FIG.
As shown in FIG. 2, both of these photographic images SG1 and SG2 are displayed on the first display 7c and the second display 7d, respectively (first display 7c and second display 7).
Photo image S before the deviation correction work that was displayed immediately before in d
G1 and SG2 are erased. ). On the other hand, the main control unit 2 transmits the conversion formula HS2 to the coordinate conversion control unit 19 and the three-dimensional position calculation unit 13 to store it. The deviation correction work is completed as described above. As a result of the deviation correction work, as shown in FIG.
Photo image SG displayed on each display 7c, 7d
Between 1 and SG2, since parallax occurs only in the X1 axis direction corresponding to the Xs1 axis or in the X2 axis direction corresponding to the Xs2 axis, the value of the Y1 axis component of one point on the first photographic image SG1 and the The value of the Y2-axis component of one point on the second photographic image SG2 corresponding to one point is always equal.

After the deviation correction work is completed, the main control unit 2 starts the projective conversion formula detection work. The main control unit 2 receives the reference coordinates KZ, the camera coordinates CZ1 and CZ from the deviation correction calculation unit 12.
2. The corrected plane coordinates SHZ1 and SHZ2 are transmitted and input to the projective transformation control unit 25, and the five path point coordinate positions Pa1 to Pa5 on the camera coordinate CZ1 and the camera coordinate CZ2 used in the relative orientation calculation unit 10 are also transmitted. Of the five path point coordinate positions Pb1 to Pb5 are taken out in the form of pairs corresponding to each other, and these pairs are transmitted to the coordinate conversion control unit 19. Coordinate conversion control unit 19
Then, the transmitted pass point coordinate positions Pa1 to Pa5 on the camera coordinate CZ1 and the transmitted pass point coordinate positions Pb1 to Pb5 on the camera coordinate CZ2 are coordinate positions (i1, j on the corrected plane coordinate SHZ1 by a known projective transformation.
1), (i2, j2), (i3, j3), (i4, j
4), (i5, j5) and coordinate positions (l1, m1), (l2, m2), (l3, m) on the corrected plane coordinate SHZ2.
3), (l4, m4), and (l5, m5) are converted and transmitted to the projection conversion control unit 25. Projective transformation control unit 25
Then, i1 to i5 are substituted for x in the equation 1, which is an equation in which a, b, c, x, x ′, y, and y ′ are variables, and j1 is substituted for y.
~ J5 is substituted, l1 to l5 are substituted for x ', and m is substituted for y'.
By substituting 1 to m5, four equations having variables a, b, and c are formed, and a determinant having variables a, b, and c, that is, Equation 2 is formed from the four equations. Note that x and y in the equation 1 are X of an arbitrary point in the corrected plane coordinates SHZ1.
Corresponding to the s1 axis component and the Ys1 axis component, x ′ and y ′ correspond to the Xs2 axis component and the Ys2 axis component of the point on the corrected plane coordinate SHZ2 corresponding to the arbitrary point on the corrected plane coordinate SHZ1. Further, as described above, the corrected plane coordinates are set so that the Ys1 axis component of an arbitrary point on the corrected plane coordinate SHZ1 has the same value as the Ys2 axis component of the point on the corrected plane coordinate SHZ2 corresponding to the arbitrary point. Since SHZ1 and SHZ2 are deviated, y and y'are always equal.

[Equation 1]

[Equation 2] The equation 2 can be transformed into the equation 3.

[Equation 3] The matrix with the subscript t on the right shoulder represents the transposed matrix,
The matrix with the subscript -1 on the right shoulder represents the inverse matrix. Therefore, the projective transformation control unit 25 uses the variable a,
b and c are calculated and the calculated variables a, b and c are substituted into the above equation 1, and the corrected plane coordinates SHZ1 and the corrected plane in which the equation 1 is used as variables y, y ', x and x' It is determined as the projective transformation formula HS3 between the coordinates SHZ2.

Next, the main control section 2 starts the matching work, firstly commands the lattice setting control section 26, and sets a plurality of coordinates on the plane coordinates HZ1 displayed on the first display 7c as shown in FIG. Vertical line 27a and multiple horizontal lines 27b
Are provided in the form of forming a predetermined lattice stripe 27, and the vertical line 27
A reference intersection 29 is set at each intersection of a and the horizontal line 27b. After setting, the main control unit 2 causes the first setting operation of the correlation window 30. That is, the main control unit 2 commands the correlation window control unit 20 to set the plane coordinates H in which a plurality of reference intersections 29 are set.
One arbitrary reference intersection 29 in Z1 is selected, and the selected reference intersection 29 is set as the matching reference point 29A. Further, as shown in FIGS. 5 and 6, centering on the matching reference point 29A, a square correlation window 30 surrounding a predetermined area is set. The points 30 at the four corners of the correlation window 30
a, 30b, 30c, and 30d are coordinate positions (i31, j) on the plane coordinate HZ1 (or the corrected plane coordinate SHZ1).
31), (i32, j32), (i33, j33),
(I34, j34), and the matching reference point 29A is the coordinate position (i29, j29). First correlation window 3
After setting 0, the correlation window control unit 20 determines that the matching reference point 2
The coordinate position (i29, j29) of 9A is transmitted and stored in the three-dimensional position calculation unit 13. Then, the correlation window control unit 20
Digital information regarding a plurality of pixels GS included in the correlation window 30, that is, density value information NJ1 regarding lightness α for each of the three chromaticities of red R, green G, and blue B is extracted, and is output to the correlation window density value detection unit 3. To transmit. In the correlation window density value detection unit 3,
As schematically shown in FIG. 6, the information on the lightness α of the red R of the plurality of pixels GS included in the correlation window 30 is correlated with the information on the lightness α of the green G on the correlation window red density value detection unit 3a. Information regarding the window green density value detection unit 3b and the lightness α of blue B is separately transmitted to the correlation window blue density value detection unit 3c, and these density value detection units 3a, 3b,
3c, a plurality of pixels G included in the correlation window 30
Distributions P1, P of lightness α for each chromaticity R, G, B of S
2, P3 is detected. After detection, each of these chromaticities R,
The distribution states P1, P2, P3 of the brightness α for each of G and B are transmitted to the matching point detection unit 6. On the other hand, after setting the correlation window 30 for the first time, the correlation window control unit 20 sets the coordinate positions (i31, j31), (i32, j32) of the points 30a, 30b, 30c, 30d at the four corners of the correlation window 30, (I3
3, j33), (i34, j34) and the coordinate position (i29, j29) of the matching reference point 29A are output to the projective transformation control unit 25.

The projective transformation control unit 25 controls the coordinate positions (i3) of the points 30a, 30b, 30c and 30d at the four corners of the correlation window 30.
1, j31), (i32, j32), (i33, j3)
3), (i34, j34) and matching reference point 29
The coordinate position (i29, j29) of A is converted into the projective transformation formula H
Substituting for each S3, each corrected plane coordinate SHZ
Coordinate positions (i31, j31), (i32, j3)
2), (i33, j33), (i34, j34), (i
29, j29) and coordinate positions (l31, m31), (l32, m32), (l) on the corrected plane coordinate SHZ2.
33, m33), (l34, m34), (l29, m2)
9) is calculated. After the calculation, the projective transformation control unit 25 determines that these five coordinate positions (l31, m31), (l32, m3).
2), (l33, m33), (l34, m34), (l
29, m29) to the search window controller 21. The search window control unit 21 uses these coordinate positions (l31, m31),
(L32, m32), (l33, m33), (l34,
Based on m34) and (l29, m29), coordinate positions (l31, m31), (l3) on the corrected plane coordinate SHZ2.
2, m32), (l33, m33), (l34, m3)
4) parallelograms having four corner points 31a, 31b, 31c, 31d (because any two parallel lines on the corrected plane coordinate SHZ1 are converted to the corrected plane coordinate SHZ2 by the projective transformation formula HS3). In this case, the arbitrary two straight lines before conversion are parallel to each other in the corrected plane coordinate SHZ2. Therefore, on the corrected plane coordinate SHZ1, the line connecting the points 30a and 30b and the line connecting the points 30c and 30d are parallel to each other. Since the straight line connecting the points 30a and 30c and the straight line connecting the points 30b and 30d are parallel to each other, the points 31a and 31 on the corrected plane coordinate SHZ2 are
The straight line connecting b and the straight line connecting points 31c and 31d are parallel to each other, and the straight line connecting the points 31a and 31c and point 31
(Because the straight lines connecting b and 31d are parallel to each other), a search window 31 is set. In addition, the search window control unit 21 sets the coordinate position (l29, m29) as the search corresponding point 29B. The search corresponding point 29B is the center point of the parallelogram shape of the search window 31 (because of the corrected plane coordinates SHZ1.
When the arbitrary two points above and the midpoints of these two points are transformed into the modified plane coordinates SHZ2 by the projective transformation formula HS3, the point which was the midpoint of the arbitrary two points before transformation is
Since it is located at the midpoint between the two arbitrary points located on the plane coordinate HZ2 after conversion, the center point, which is the midpoint and intersection of the two diagonals of the square on the revised plane coordinate SHZ1, is parallel on the revised plane coordinate SHZ2. (Because it is the center of the quadrilateral, which is the midpoint and intersection of the two diagonals).

By the way, the above-mentioned projective transformation formula HS3 has the following meaning. That is, the model 61 at the reference coordinates KZ obtained by the mutual orientation work and the ground orientation work does not have a flat surface but a specific height difference. Therefore,
In order to determine the coordinate position of the point in the model 61 in the reference coordinate KZ, a point on the corrected plane coordinate SHZ1 corresponding to the coordinate position of the point in the model 61, as indicated by the above-mentioned conversion formulas HS1 and HS2. Coordinate position and model 6
Modified plane coordinates SHZ2 corresponding to the coordinate position of the point in 1
The coordinate position of the upper point had to be determined together. However, the five coordinate positions (i1
To i5, j1 to j5) and the corresponding five coordinate positions (11 to 15 and m1 to m5 on the corrected plane coordinates SHZ2).
5) and these five coordinate positions (i1 to i5, j1 to j5) obtained by the conversion expressions HS1 and HS2, or 5
Corresponding points T1 to T5 in the model 61 corresponding to two coordinate positions (11 to 15 and m1 to m5) are obtained, and these corresponding points T
When the virtual plane HH approximately including 1 to T5 is assumed on the reference coordinate KZ as shown in FIG. 3, the camera coordinate CZ1
The light source is placed at the upper projection center O1 and the corrected plane coordinates SHZ1
The upper first photographic image SG1 can be projected as the virtual photographic image SG3 on the virtual plane HH. That is, the model 61 having a relative height difference is set to the virtual photograph image SG3 having no relative height difference.
Can be replaced approximately. That is, the above-described projective transformation formula HS3 is a straight line that connects an arbitrary point on the corrected plane coordinate SHZ1 with a point on the virtual plane HH corresponding to the arbitrary point and the projection center O2 on the camera coordinate CZ2. , Which means conversion to an intersection with the corrected plane coordinate SHZ2. Therefore, the points 30a, 30b, 30c, and 30d at the four corners of the correlation window 30 of the corrected plane coordinates SHZ1 are converted into the projective transformation formula HS.
Points 31a, 31b of the corrected plane coordinates SHZ2 by 3,
31c, 31d and these points 31a, 31b, 3
By setting the search windows 31 having the four corners 1c and 31d, the points in the virtual photographic image SG3 on the virtual plane HH corresponding to the points included in the correlation window 30 are approximately replaced by the model 61. Since it corresponds to the points included in the search window 31, it is possible to maximize the consistency between the points included in the correlation window 30 and the points included in the search window 31. Therefore, in the comparison of the distribution of image density values, which will be described later,
Since it is possible to reduce the possibility that the similarity between images that should not correspond to each other will be evaluated higher than the images that should correspond to each other, it is possible to perform appropriate matching.

On the other hand, the virtual plane HH does not coincide with the model 61 having a difference in specific height as described above, and therefore the projective transformation formula HS3 is the corrected plane coordinate SHZ1.
Since it is an expression for converting any point on the corrected plane coordinate SHZ2 approximately corresponding to the arbitrary point, all the points on the corrected plane coordinate SHZ1 become corresponding points on the corrected plane coordinate SHZ2. It cannot be moved exactly. Therefore, as described above, the matching reference point 29A is the projective transformation expression HS.
3 is converted into a point on the corrected plane coordinate SHZ2, and the converted point is set as a search corresponding point 29B corresponding to the matching reference point 29A.
A and the search corresponding point 29B only correspond approximately and do not necessarily correspond exactly. Therefore, the matching point detection operation for obtaining the search corresponding point 29B on the corrected plane coordinate SHZ2 that exactly corresponds to the matching reference point 29A at the center of the correlation window 30 is performed below. by the way,
As described above, the corrected plane coordinates SHZ1 and SHZ2 are offset from each other, and both corrected plane coordinates SHZ1.
1, Y1 between both photographic images SG1 and SG2 on SHZ2
There is no parallax in the axis (Ys1 axis) direction or the Y2 axis (Ys2 axis) direction. That is, the point on the corrected plane coordinate SHZ2 that exactly corresponds to the matching reference point 29A is the projective transformation formula HS.
Search corresponding points 29B and Y2 converted and set by 3
The axis (Ys2 axis) components should have equal values. Therefore, the main control unit 2 commands the search window control unit 21 to set the search window 31 and the search corresponding point 29B to the corrected plane coordinates SHZ2.
In the above, the Xs2 axis is translated in the positive direction (or the negative direction) by a predetermined movement amount δ. The movement amount δ is 1
Equal to the width of one pixel GS in the X2 axis (Xs2 axis) direction,
The parallel movement by the movement amount δ means that the search corresponding point 2
9B means that the position of 9B moves in such a manner that the position of 9B sequentially changes to correspond to the pixel GS adjacent in the X2 axis (Xs2 axis) direction. Further, the search window control unit 21 causes the search corresponding point 29B to have a coordinate position (lk,
m29) is output to the matching point detection unit 16 for each movement amount δ movement. In addition, the search window control unit 21 shifts the movement amount δ and digital information regarding the plurality of pixels GS included in the search window 31, that is, the brightness of each of the three chromaticities of red R, green G, and blue B. The density value information NJ2 regarding β is extracted and transmitted to the search window density value detection unit 5. In the search window density value detection unit 5, as shown in FIG.
The brightness β of the red R among the plurality of pixels GS included in 1
Information regarding the green color G is displayed in the red density value detection unit 5a of the search window.
Information about the brightness β of the search window green density value detection unit 5b,
Information regarding the brightness β of blue B is separately transmitted to the search window blue density value detection unit 5c, and among these density value detection units 5a, 5b, and 5c, among the plurality of pixels GS included in the search window 31, The distribution statuses Q1, Q2, Q3 of the lightness β for each chromaticity R, G, B are detected. Also after detection,
The distribution situation Q1, Q of the lightness β for each of the chromaticities R, G, B
2 and Q3 are transmitted to the matching point detection unit 6.
That is, every time the search window 31 moves by the movement amount δ, the distribution situation Q of the brightness β of the plurality of pixels GS inside the search window 31.
1, Q2, Q3 are detected and transmitted to the matching point detection unit 6 each time they are detected.

Each time the distribution statuses Q1, Q2, Q3 are transmitted to the matching point detection unit 6, the main control unit 2 causes the matching point detection unit 6 to compare the distribution statuses.
That is, the matching point detection unit 6 determines the distribution states P1, P2, P3 of the brightness α of the plurality of pixels GS in the correlation window 30 that have already been transmitted, and the plurality of pixels GS in the search window 31 that are sequentially transmitted. Distribution of brightness β of Q1, Q2, Q3
The correlation coefficient γ indicating the degree of similarity between and is sequentially obtained. As shown in FIG. 6, the correlation coefficient γ is calculated based on the distribution conditions P1 and Q1.
Correlation coefficient γ1, between the distribution statuses P2, Q2
2. The correlation coefficient γ3 between the distribution states P3 and Q3 is calculated, and the product of these correlation coefficients γ1, γ2, and γ3 (γ1 ×
It is calculated as the cube root of γ2 × γ3). The correlation coefficient γ
Is calculated as the cube root of the product of the correlation coefficients γ1, γ2, γ3, the correlation coefficient γ is calculated as the correlation coefficient γ1, γ2, γ3.
And the correlation coefficients γ1, γ2, γ3
If even one of them is zero, the correlation coefficient γ becomes zero, which reflects that there is no correlation. The correlation coefficient γ that is sequentially obtained is the search corresponding point 29B corresponding to the correlation coefficient γ.
It is stored in the matching point detection unit 6 in a pair with the coordinate position (lk, m29) of (already transmitted to the matching point detection unit 6).

The search window 31 and the search corresponding point 29B are moved by moving the search window 31 by the movement amount δ and repeating the operation of detecting the correlation coefficient γ by the matching point detection unit 6 each time the movement is performed. After the corresponding point 29B is moved to a position within a predetermined range that may correspond to the matching reference point 29A exactly, the search window control unit 21 moves the search window 31. To finish. Upon completion, the main control unit 2 commands the matching point detection unit 6 to detect the matching corresponding point MT. That is, the matching point detection unit 6 selects the largest one among the plurality of stored correlation coefficients γ, and the search corresponding point 29B stored corresponding to this correlation coefficient γ.
The coordinate position (lk, m29) of the matching reference point 29A
Is detected as a matching corresponding point MT corresponding to. In general, the individuality of an image formed by a plurality of pixels in the correlation window or the search window is determined by the chromaticity of each of the plurality of pixels, rather than in the distribution of lightness irrelevant to the chromaticity of the plurality of pixels. It is more effective in each lightness distribution, so the similarity used for matching is the chromaticity of multiple pixels in the correlation window or search window, as in conventional photogrammetry. The correlation coefficient γ, which is the similarity used to perform matching as in the photogrammetry method according to the present invention, is detected in the correlation window 30 or the search window 3 rather than being detected only by comparing the distribution of lightness that is not related to
A plurality of pixels GS in the correlation window 30 can be detected by comparing distributions P1 to P3 and Q1 to Q3 of lightness α, β for each of the chromaticities R, G, and B of the plurality of pixels GS in one. The position of the search window 31 including a plurality of pixels GS forming an image corresponding to the image forming the position can be detected with higher reliability. That is, by applying the present invention, mismatching is prevented as much as possible, and proper detection of the three-dimensional shape of the survey object 60 is facilitated. After detecting the matching corresponding point MT, the matching point detecting unit 6 outputs the coordinate position (l, m29) of the matching corresponding point MT on the corrected plane coordinates SHZ2 to the three-dimensional position calculating unit 13.
Then, the three-dimensional position calculation unit 13 determines the corrected plane coordinates SHZ.
Matching reference point 2 corresponding between 1 and SHZ2
Based on 9A and matching corresponding points MT, reference coordinates K of the corresponding points on the model 61 corresponding to these matching reference points 29A and matching corresponding points MT after ground location work.
The three-dimensional position SI on Z is calculated by the conversion formulas HS1 and HS2 already input to the three-dimensional position calculation unit 13. The three-dimensional position calculation unit 13 stores the three-dimensional position SI in the three-dimensional position memory unit 15.

After that, the main control unit 2 performs the same procedure as the series of operations including the setting of the search window, the detection of the correlation coefficient, the detection of the matching point, and the detection of the three-dimensional position, which starts from the setting of the first correlation window. Then, a series of operations including the setting of the search window, the detection of the correlation coefficient, the detection of the matching point, and the detection of the three-dimensional position starting from the second, third, fourth, ... That is, the main control unit 2 commands the correlation window control unit 20 to select and select one of the reference intersection points 29 that has not been selected as the matching reference point 29A, as a new matching reference point 29A. A correlation window 30 is newly set around the matching reference point 29A. After setting, the correlation window control unit 20 causes the three-dimensional position calculation unit 13 to transmit and store the coordinate position (i29, j29) of the matching reference point 29A. Next, the correlation window control unit 20
Extracts the density value information NJ1 of a plurality of pixels GS included in the correlation window 30 and transmits it to the correlation window density value detection unit 3. In the correlation window density value detection unit 3, the correlation window red density value detection unit 3a detects the distribution status P1, the correlation window green density value detection unit 3b detects the distribution status P2, and the correlation window blue density value detection unit 3c. The distribution status P3 is detected. After the detection, these distribution states P1, P2, P3 are transmitted to the matching point detection unit 6. On the other hand,
After setting the correlation window 30, the correlation window control unit 20 sets the correlation window 30
Coordinate points (i31, j31), (i3) in the corrected plane coordinates SHZ1 of the points 30a, 30b, 30c, 30d at the four corners of
2, j32), (i33, j33), (i34, j3
4) and the coordinate position of the matching reference point 29A (i2
9, j29) is output to the projective transformation control unit 25. The projective transformation control unit 25 uses the points 30 a and 30 at the four corners of the correlation window 30.
coordinate positions (i31, j31) of b, 30c, 30d,
(I32, j32), (i33, j33), (i34,
j34) and the coordinate position of the matching reference point 29A (i
29, j29) are respectively substituted into the projective transformation formula HS3, and the coordinate positions (i31, j31), (i32, j32), (i33, in the corrected plane coordinates SHZ1 are respectively substituted.
j33), (i34, j34), and (i29, j29) corresponding to the coordinate position (l31,
m31), (l32, m32), (l33, m33),
(L34, m34) and (l29, m29) are calculated.
After the calculation, the projective transformation control unit 25 causes the five coordinate positions (l31, m31), (l32, m32), (l33,
m33), (l34, m34), and (l29, m29) are output to the search window control unit 21. The search window control unit 21 controls the coordinate positions (l31, m31), (l32, m3).
2), (l33, m33), (l34, m34), (l
29, m29), coordinate positions (l31, m31), (l32, m3) on the corrected plane coordinate SHZ2.
2) Set parallelogram search window 31 having four corner points 31a, 31b, 31c and 31d at (l33, m33) and (l34, m34). In addition, the search window control unit 21
The coordinate position (l29, m29) is set as the search corresponding point 29B. Then, the matching point detection operation is performed to find the search corresponding point 29B on the corrected plane coordinate SHZ2 that exactly corresponds to the matching reference point 29A at the center of the correlation window 30. First, the search window control unit 21 sets the search window 31 and the search corresponding point 29B on the corrected plane coordinates SHZ2 by X.
The s2 axis is translated in the positive direction (or the negative direction) by a predetermined movement amount δ. Further, the search window control unit 21 determines the coordinate position (l of the search corresponding point 29B on the corrected plane coordinate SHZ2).
k, m29) is output to the matching point detection unit 16 for each movement amount δ movement. In addition, the search window control unit 21
Each time the movement amount δ is moved, digital information regarding a plurality of pixels GS included in the search window 31, that is, density value information NJ2 regarding lightness β for each of three chromaticities of red R, green G, and blue B is extracted. , To the search window density value detection unit 5.
In the search window density value detection unit 5, the search window red density value detection unit 5
The distribution status Q1 is detected in a, the distribution status Q2 is detected in the search window green density value detection section 5b, and the distribution status Q3 is detected in the search window blue density value detection section 5c. After the detection, the distribution statuses Q1, Q2, Q3 are transmitted to the matching point detection unit 6. That is, every time the search window 31 moves by the movement amount δ, the distribution states Q1, Q2, and Q3 of the brightness β of the plurality of pixels GS inside the search window 31 are detected, and the matching point detection unit 6 is detected each time. Be transmitted to. Every time the distribution statuses Q1, Q2, Q3 are transmitted to the matching point detection unit 6, the main control unit 2 causes the matching point detection unit 6 to compare the distribution statuses. That is, the matching point detection unit 6 determines the distribution status P of the brightness α of the plurality of pixels GS in the already transmitted correlation window 30.
A correlation coefficient γ indicating the degree of similarity between 1, P2, P3 and the distribution statuses Q1, Q2, Q3 of the brightness β of the plurality of pixels GS in the search window 31 that are sequentially transmitted is sequentially obtained. The sequentially obtained correlation coefficient γ is stored in the matching point detection unit 6 as a pair with the coordinate position (lk, m29) of the search corresponding point 29B corresponding to the correlation coefficient γ. The search window 31 and the search corresponding point 29B are moved while moving the search window 31 by the movement amount δ and repeating the operation of detecting the correlation coefficient γ by the matching point detection unit 6 each time the movement is performed. Is moved to a position within a predetermined range that may correspond exactly to the matching reference point 29A, the search window control unit 21 then moves the search window 31 to the search window 31.
End the move. Upon completion, the main control unit 2 commands the matching point detection unit 6 to detect the matching corresponding point MT. That is, the matching point detection unit 6 selects the largest one from the plurality of stored correlation coefficients γ, and the coordinate position (lk, lk, of the search corresponding point 29B stored corresponding to the correlation coefficient γ). m29)
Is detected as the coordinate position (l, m29) of the matching corresponding point MT corresponding to the matching reference point 29A. After detecting the matching corresponding point MT, the matching point detecting unit 6
Is the coordinate position (l, m29) of the matching corresponding point MT.
Is output to the three-dimensional position calculation unit 13. Then, the three-dimensional position calculation unit 13 determines the matching reference points 29A based on the matching reference points 29A and the matching corresponding points MT corresponding between the corrected plane coordinates SHZ1 and SHZ2.
And the corresponding three-dimensional position SI on the reference coordinate KZ of the corresponding point on the model 61 after the ground orientation work corresponding to the matching corresponding point MT are already input to the three-dimensional position calculating unit 13 by the conversion equations HS1 and HS2 Calculate by The three-dimensional position calculation unit 13 stores the three-dimensional position SI in the three-dimensional position memory unit 15. The above correlation window settings, search window settings,
After repeating a series of operations including the detection of the correlation coefficient, the detection of the matching point, and the detection of the three-dimensional position, the correlation window control unit 20 never sets the matching reference point when trying to set the next correlation window. Since the reference intersection 29 not selected as 29A cannot be found on the corrected plane coordinate SHZ1, the correlation window control unit 20 transmits a signal indicating the matching work completion to the main control unit 2, and the main control unit 2 sends the signal. It is determined that the matching points have already been detected for all the reference points 29 by receiving, and therefore the three-dimensional positions SI of the corresponding points of the survey object 60 corresponding to all the reference points 29 have been obtained. Work instruction display section 7b
Display the matching work completion (not shown).

An operator (not shown) who has seen the display (not shown) of the matching work completion is
Input the 3D information output signal. The main control unit 2 outputs the three-dimensional position SI input to the three-dimensional position memory unit 15 from the output unit 16 based on the three-dimensional information output signal. The three-dimensional position SI is a coordinate position on the reference coordinate KZ of the model 61 after the ground orientation, and since the model 61 and the surveyed object 60 and the reference coordinate KZ and the ground space CK correspond to each other, The original position SI is the three-dimensional position of the point on the survey object 60 in the ground space CK, that is, the latitude, longitude, and altitude. The output unit 16 may output a numerical value or the like indicating the three-dimensional position SI in the form of a display, print it out, or output it in a form of recording it on another magnetic storage medium. With the above, all the surveying of the surveying object 60 using the photogrammetric device 1 is completed.

[0026]

As described above, according to the first aspect of the present invention, the surveying object such as the surveying object 60 existing in the three-dimensional space such as the ground space CK is set in the first three-dimensional space. First photographic image SG1 taken from the first position such as point PT1
And second photograph images of the surveyed object from a second position such as a second point PT2 different from the first position in the three-dimensional space. In a photogrammetric device that detects a three-dimensional shape such as a three-dimensional position SI in the three-dimensional space of the surveying object based on an image, the photogrammetric device stores the first photo image and the second photo image. In the first photographic image stored in the image storage unit, which has an image storage unit such as a possible photographic image memory 9, a matching reference point such as an arbitrary matching reference point 29A on the first photographic image is used as a reference. A correlation window setting unit such as a correlation window control unit 20 that can set a correlation window such as a correlation window 30 that surrounds a predetermined area is provided, and a plurality of pixels such as the pixels GS in the correlation window have a lightness such as a lightness α. Distribution situation P
Correlation window pixel distribution status detection by the correlation window density value detection unit 3 or the like capable of detecting the distribution status of 1, P2, P3, etc. for each chromaticity of a plurality of types of red R, green G, blue B, etc. that each pixel has And a search window control unit capable of setting a search window such as the search window 31 corresponding to the correlation window in the second photographic image stored in the image storage unit by moving the second photographic image on the second photographic image. Two
A search window setting unit such as 1 is provided to detect the distribution status of brightness such as brightness β of a plurality of pixels in the search window for each of a plurality of types of chromaticity of each pixel. A search window pixel distribution status detection unit such as a search window density value detection unit 5 is provided, and the brightness distribution status for each chromaticity in the correlation window and the brightness distribution status for each chromaticity in the search window are set. A matching point detecting unit 6 capable of determining the degree of similarity of the correlation coefficient γ and the like and detecting the matching corresponding point such as the matching corresponding point MT of the second photographic image corresponding to the matching reference point of the correlation window.
A matching point detection unit such as, and corresponding to the first and second photographic images, based on the position of the matching reference point and the matching corresponding point on each photographic image, the tertiary of the surveying object. A three-dimensional shape detection unit such as a three-dimensional position calculation unit 13 that can detect a three-dimensional shape in the original space is provided. Therefore, the individuality of the image formed by a plurality of pixels in the correlation window or the search window is determined by each of the chromaticities of the plurality of pixels, rather than in the lightness distribution state irrelevant to the chromaticity of the plurality of pixels. Since it is more effective in the distribution of lightness, the similarity used for matching is related to the chromaticity of multiple pixels in the correlation window or the search window, as in conventional photogrammetric equipment. Like the photogrammetric device according to the invention, rather than only detecting by comparing the distribution of no lightness,
The similarity used to perform matching is detected by comparing the distribution of lightness for each chromaticity of multiple pixels in the correlation window or search window, and the multiple pixels in the correlation window are configured. The position of the search window including a plurality of pixels forming an image corresponding to the image can be detected with higher reliability. That is, by using the photogrammetric device of the present invention, mismatching is prevented as much as possible,
Appropriate detection of the three-dimensional shape of the surveyed object becomes easy. A second invention of the present invention is to measure a surveying object such as a surveying object 60 existing in a three-dimensional space such as a ground space CK from a first position such as a first point PT1 in the three-dimensional space. The first photograph image such as the first photograph image SG1 and the second photograph obtained by photographing the surveyed object from a second position such as a second point PT2 different from the first position in the three-dimensional space. In a photogrammetric method for detecting a three-dimensional shape such as a three-dimensional position SI in the three-dimensional space of the surveying object based on a second photo image such as the image SG2, an arbitrary matching criterion on the first photo image A correlation window such as a correlation window 30 that surrounds a predetermined area is set on the basis of a matching reference point such as the point 29A, and a distribution state P1 of brightness such as brightness α of pixels such as a plurality of pixels GS in the correlation window. , P2, P3, etc. distribution status for each pixel A plurality of types of chromaticity such as red R, green G, and blue B are detected, and a search window such as a search window 31 corresponding to the correlation window in the second photographic image is moved on the second photographic image. Each time the search window moves a predetermined amount of movement δ or the like, the distribution of the lightness distributions Q1, Q2, Q3, etc. of the plurality of pixels within the search window such as the brightness β The situation is detected for each of a plurality of types of chromaticity of each pixel, and the distribution state of lightness for each chromaticity in the correlation window and the search detected each time the search window moves a predetermined amount of movement. Detecting the degree of similarity such as the correlation coefficient γ with the distribution of lightness for each chromaticity in the window, the second photographic image of the search window corresponding to the highest value of the detected degree of similarity From the position, the matching corresponding point MT or the like of the second photographic image corresponding to the matching reference point of the correlation window is matched. Detect corresponding points,
Corresponding in the first and second photographic image, based on the position of the matching reference point and the matching corresponding point on each photographic image, to detect the three-dimensional shape of the surveyed object in the three-dimensional space. Consists of Therefore,
The individuality of the image formed by a plurality of pixels in the correlation window or the search window is more Since it is more widely used in the distribution situation, the similarity used for matching, as in the conventional photogrammetry method, is the brightness that is independent of the chromaticity of multiple pixels in the correlation window or the search window. Rather than detecting only by comparing the distribution states of the above, as in the photogrammetry method according to the present invention, the degree of similarity used for matching is determined by the brightness of each chromaticity of a plurality of pixels in the correlation window or the search window. It is more reliable to detect the position of the search window including a plurality of pixels forming an image corresponding to the image formed by a plurality of pixels in the correlation window by comparing the distribution states of the. Kill. That is, by applying the present invention, mismatching is prevented as much as possible, and proper detection of the three-dimensional shape of the surveyed object is facilitated.

[Brief description of drawings]

FIG. 1 is a block diagram showing a photogrammetric apparatus to which the present invention is applied.

FIG. 2 is a perspective view showing a positional relationship between a surveyed object and first and second points where stereo imaging is performed.

FIG. 3 is a diagram showing conceptual reference coordinates KZ and camera coordinates CZ1 and CZ2 input to the photogrammetric apparatus of FIG. 1.

FIG. 4 is a diagram showing first and second displays.

FIG. 5 is a diagram showing first and second displays.

FIG. 6 is a diagram conceptually showing a matching procedure.

[Explanation of symbols]

1 ... Photogrammetry device 3 ... Correlation window pixel distribution status detection unit (correlation window density value detection unit) 5 ... Search window pixel distribution status detection unit (search window density value detection unit) 6 ... Matching point detection unit 9 ...... Image storage unit (photographic image memory) 13 ...... Three-dimensional shape detection unit (three-dimensional position calculation unit) 20 ...... Correlation window setting unit (correlation window control unit) 21 ...... Search window setting unit (Search window control unit) ) 29A ... Matching reference point 30 ... Correlation window 31 ... Search window 60 ... Surveying object CK ... Three-dimensional space (ground space) GS ... Pixel MT ... Matching corresponding point P1 ... Distribution situation P2 ... … Distribution status P3 …… Distribution status PT1 …… First position (first point) PT2 …… Second position (second point) Q1 …… Distribution status Q2 …… Distribution status Q3 …… Distribution status SG1 …… First photo image SG2 ... Second photo image SI ... … Three-dimensional shape (three-dimensional position) R …… Chromaticity (red) G …… Chromaticity (green) B …… Chromaticity (blue) α …… Brightness β …… Brightness γ …… Similarity (correlation coefficient ) Δ …… Amount of movement

Claims (2)

[Claims] 1. A first photographic image obtained by photographing a surveying object existing in a three-dimensional space from a first position in the three-dimensional space, and the surveying object in the first position in the three-dimensional space. In a photogrammetric device that detects a three-dimensional shape in the three-dimensional space of the surveying object based on a second photo image taken from a second position different from, the photogrammetric device is the first photo image and An image storage unit capable of storing the second photographic image, and in the first photographic image stored in the image storage unit, surrounds a predetermined area with reference to an arbitrary matching reference point on the first photographic image. A correlation window setting unit that can set a correlation window, and a correlation window pixel distribution state detection unit that can detect the distribution state of the brightness of a plurality of pixels in the correlation window for each of a plurality of types of chromaticity of each pixel. Is stored in the image storage unit. In the second photographic image, a search window setting unit capable of setting a search window corresponding to the correlation window in a manner of moving on the second photographic image is provided, and the brightness distribution status of a plurality of pixels in the search window Is provided with a search window pixel distribution status detection unit capable of detecting each of a plurality of types of chromaticity of each pixel, and the distribution status of lightness for each chromaticity within the correlation window, and for each chromaticity within the search window. Provide a matching point detection unit that can detect the matching correspondence point of the second photographic image corresponding to the matching reference point of the correlation window by determining the similarity with the distribution state of lightness, the first and second photographic images A corresponding three-dimensional shape detection unit capable of detecting the three-dimensional shape of the surveyed object in the three-dimensional space based on the positions of the matching reference point and the matching corresponding point on each photographic image. Configured photogrammetry Location. 2. A first photographic image obtained by photographing a surveying object existing in a three-dimensional space from a first position in the three-dimensional space, and the surveying object in the first position in the three-dimensional space. In a photogrammetric method for detecting a three-dimensional shape of the surveyed object in the three-dimensional space, based on a second photo image taken from a second position different from, any matching criterion on the first photo image. By setting a correlation window surrounding a predetermined area on the basis of points, the distribution state of the brightness of a plurality of pixels in the correlation window is detected for each of a plurality of types of chromaticity of each pixel, and the second photograph In the image, a search window corresponding to the correlation window is set so as to move on the second photographic image, and the brightness of a plurality of pixels in the search window is set every time the search window moves by a predetermined movement amount. The distribution status of each pixel for each type of chromaticity , The degree of lightness distribution for each chromaticity in the correlation window, and the similarity between the degree of lightness distribution for each chromaticity in the search window detected each time the search window moves a predetermined amount of movement, The matching corresponding point of the second photo image corresponding to the matching reference point of the correlation window is detected from the position of the second photo image of the search window corresponding to the highest similarity among the detected similarities. A three-dimensional shape of the surveyed object in the three-dimensional space, which is detected and based on the positions of the matching reference points and the matching corresponding points on the respective photographic images that correspond in the first and second photographic images. A photogrammetric method that detects and configures.