JP2002342787A - Method and device for generating three-dimensional model and computer program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate a precise three-dimensional model. SOLUTION: This three-dimensional model generating device is constituted of a three-dimension reconstituting part 31 for generating the three-dimensional shape data of an object based on data obtained by measuring the object in a non-contact state, a multi-lens camera 3 for obtaining the two-dimensional picture of the object by photographing the object, a calculating part 32 for calculating the deviation of view points between the measurement of the object and the photographing of the object based on a two-dimensional picture FD1 obtained at photographing the object and a two-dimensional FD2 obtained by photographing the object, and a mapping part 33 for attaching the two-dimensional picture FD2 obtained by photographing the object to three-dimensional shape data DL by correcting the calculated deviation of the view points.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for generating a mapped three-dimensional model based on two images of a subject taken in stereo, and a computer program.

[0002]

2. Description of the Related Art Conventionally, there has been known a technique of generating three-dimensional shape data from a plurality of images (two-dimensional images) of a target object having parallax. In order to acquire a plurality of images having parallax, a multiview camera in which a plurality of cameras are integrated is often used.

In a multi-lens camera, external parameters (camera position and orientation) and internal parameters (focal length, pixel pitch, etc.) of each camera are calibrated in advance. An object is photographed by a multi-lens camera, corresponding points are detected from a plurality of obtained images, and three-dimensional reconstruction is performed based on the principle of triangulation to obtain three-dimensional shape data. In order to detect more sets of corresponding points, pattern light is projected onto an object to be photographed.

[0004] By mapping a two-dimensional image of an object onto the generated three-dimensional shape data, a faithful and good-looking three-dimensional model of the object is generated. When capturing an image for mapping, the pattern light is in the way, so it is performed separately from capturing for obtaining three-dimensional shape data.

Conventionally, when photographing an object,
Since the multi-lens camera is fixed by using a tripod mount or the like, if the object does not move, even if a plurality of shootings are performed, the image does not shift between them.

[0006]

However, if the object moves, there is a possibility that images obtained by a plurality of photographings will be shifted from each other.

Further, since it is troublesome to fix the multi-lens camera with the tripod mount, there is a demand for photographing the multi-lens camera by hand if possible. If it is hand-held, the object can be easily photographed.

[0008] However, when shooting with a handheld camera,
There is a high possibility that the multi-lens camera will inevitably move in a plurality of shootings, causing a shift between the obtained images.

[0009] Then, there is a possibility that a shift may occur between the three-dimensional shape data generated based on the image photographed by projecting the pattern light and the image photographed for mapping. If mapping is performed with the image shifted, an unnatural three-dimensional model will result.

In order to generate three-dimensional shape data, a three-dimensional measuring device for measuring an object in a non-contact manner by a light cutting method or the like may be used. However, in any case, when the photographing or measurement is performed by hand, the position or the viewpoint is determined between the photographing or measurement for acquiring the three-dimensional shape data and the photographing of the two-dimensional image for mapping. Mismatch occurs. Therefore, an accurate three-dimensional model cannot be generated.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and has as its object to generate an accurate three-dimensional model.

[0012]

A method according to the present invention comprises the steps of: measuring a target object in a non-contact manner to generate three-dimensional shape data of the target object; and photographing the target object to obtain a two-dimensional image of the target object. And comparing the two-dimensional image obtained at the time of measuring the object with the two-dimensional image obtained by photographing the object, and correcting the difference between them to obtain the image obtained by photographing the object. Attaching the obtained two-dimensional image to the three-dimensional shape data.

An apparatus according to the present invention comprises: means for generating three-dimensional shape data of an object based on data obtained by measuring the object in a non-contact manner; Based on a means for obtaining an image, and a two-dimensional image obtained by measuring the object and a two-dimensional image obtained by photographing the object, a viewpoint shift between the measurement of the object and the photographing of the object is calculated. Means for calculating the two-dimensional image obtained by photographing the object by correcting the obtained viewpoint displacement.
Means for attaching to the dimensional shape data.

According to another aspect, a first storage unit for storing a plurality of two-dimensional images having parallax about an object obtained by one photographing with a multi-view camera, and another one-dimensional image by the multi-view camera. A second storage unit for storing a two-dimensional image of the object obtained by the photographing, and a three-dimensional unit for generating three-dimensional shape data of the object based on the two-dimensional image stored in the first storage unit A reconstruction unit, a two-dimensional image stored in the second storage unit, and a two-dimensional image stored in the first storage unit. And a gluing unit that corrects the determined viewpoint displacement and attaches the two-dimensional image stored in the second storage unit to the three-dimensional shape data. Become.

In the present invention, the measurement for generating the three-dimensional shape data of the object includes the measurement of the object using a three-dimensional measuring device such as a light section method, and the photographing of the object using a multi-lens camera. And so on.

[0016]

FIG. 1 is a diagram showing a generating device 1 according to the present invention, and FIG. 2 is a block diagram showing a configuration example of a processing device 5.

In FIG. 1, a generating device 1 comprises a multi-lens camera 3 for photographing an object Q, and a processing device 5. The multi-lens camera 3 includes two digital cameras 3k,
Reference numeral 31 denotes a twin-lens camera connected to each other and integrated. Hereinafter, the members and elements related to one camera 3k may be denoted by “k” and the members and elements related to the other camera 3l may be denoted by “l”.

In the multi-lens camera 3, internal parameters and external parameters of each camera are calibrated in advance, and a projection matrix is obtained. By photographing the object Q once,
Two images FD of the object Q having parallax are obtained.

The multi-lens camera 3 is provided with a pattern light projection device 3p for projecting pattern light onto the object Q, and a continuous shooting control unit 3r for continuously executing shooting with pattern light and shooting without pattern light. Can be A memory for storing a large number of images FD, a memory for storing camera parameters,
A liquid crystal display for displaying the image FD, various operation buttons such as a release button, a flash, an input / output interface, a control unit, a power supply unit, and the like are provided.

The photographed image FD or an image obtained by editing the image FD can be transferred to the processing device 5 and other external devices via an input / output interface. Also,
By removing the memory storing the image FD, the image can be input to the external device. Details of the function of the multi-lens camera 3 will be described later.

The processing device 5 performs various processes described later on the image FD obtained by the multi-lens camera 3, performs calibration, and generates a three-dimensional model ML. In addition,
The function of the processing device 5 may be built in the multi-view camera 3 and the multi-view camera 3 may generate the three-dimensional model ML.

As shown in FIG. 2, the processing device 5 includes a device main body 10, a magnetic disk device 11, and a medium drive device 1.
2, a display device 13, a keyboard 14, a mouse 15, and the like.

The apparatus main body 10 includes a CPU, a RAM, and an RO.
M, a video RAM, an input / output port, and various controllers. Various functions described below are realized by the CPU executing programs stored in the RAM, the ROM, and the like.

The magnetic disk device 11 has an OS (Operat
ing System), a modeling program PR for generating a three-dimensional model ML, other programs, an input image (two-dimensional image data) FD, a generated three-dimensional model ML, other data, and the like are stored. . These programs and data are loaded into the RAM of the apparatus body 10 at appropriate times.

The modeling program PR includes programs for reliability determination processing, three-dimensional reconstruction processing, deviation calculation processing, mapping processing, and other processing.

The medium drive device 12 is a CD-ROM
(CD), a floppy disk FPD, a magneto-optical disk, a semiconductor memory HM such as a compact flash (registered trademark), and other recording media are accessed to read and write data or programs. An appropriate drive device is used depending on the type of the recording medium. The above-described modeling program PR can be installed from these recording media. An image FD or the like can also be input via a recording medium.

The display surface HG of the display device 13 includes:
The various data described above, the image FD, the image of the processing process by the modeling program PR, the generated three-dimensional model ML, and other data or images are displayed.

The keyboard 14 and the mouse 15 are used by the user to designate corresponding points on the image FD displayed on the display device 13. In addition, it is used for inputting various data to the apparatus main body 10 or giving commands.

The processing device 5 can be configured using a personal computer or a workstation. The programs and data described above can be obtained by receiving them via the network NW.

FIG. 3 is a view showing a state in which the object Q is continuously photographed by the multi-lens camera 3, FIG. 4 is a timing chart at the time of continuous photographing,
FIG. 5 is a diagram showing a flow of generation of the three-dimensional model ML by the first correction method, FIG. 6 is a diagram for explaining the first correction method, and FIG. 7 is a diagram showing generation of the three-dimensional model ML by the second correction method. FIG. 8 is a diagram showing a flow, and FIG. 8 is a diagram for explaining a second correction method.

In FIG. 4, when the release button of the multi-lens camera 3 is pressed, continuous shooting starts under the control of the continuous shooting control section 3r. In the first shooting, pattern light is projected. After time T1 from the first shooting, the second shooting
Performed without pattern light. Thereby, in each photographing, the object Q is controlled by the two cameras 3k and 3l.
Thus, two images FD having a parallax regarding are obtained.

As shown in FIG. 3, the position of the multi-lens camera 3 is shifted between the first shooting and the second shooting. Therefore, the viewpoint (the position of the camera) with respect to the object Q is shifted between the first image FD1 and the second image FD2.

In the first correction method, as shown in FIG.
Based on the two images FD1 obtained in the first shooting,
The three-dimensional reconstruction unit 31 generates three-dimensional shape data DL1. The mapping unit 33 maps the image FD2 obtained in the second shooting to the generated three-dimensional shape data DL1. At the time of the mapping, the multi-view camera 3
Is performed (registration) of the positional deviation of. In the shift calculating unit 32, the image FD obtained in the first shooting
Based on the image FD2 obtained in the first and second shots,
The positional deviation of the multi-lens camera 3 between the first and second shootings is obtained. The position where the image FD2 is to be attached is changed according to the displacement.

As shown in FIG. 6, for one vertex of the object Q, a point P1 (u1, v1) on the image FD1 corresponds to a point P2 (u2, v2) on the image FD2. However, the point P2 (u2, v2) on the image FD2 corresponds in position to the point P2a (u2a, v2a) on the image FD1. If the image FD2 is used as it is in the three-dimensional shape data D
When mapping to L1, a point P2 on the image FD2
(U2, v2) is a point PL2 of the three-dimensional shape data DL1
a (x2a, y2a, z2a).

However, the point P1 (u on the image FD1
1, v1) and the point P2a (u2a, v2a) are corrected for positional deviation based on the deviation between the point P2 (u2, v2) on the image FD2 and the point PL1 (x1) of the three-dimensional shape data DL1. , Y1, z1).

In the second correction method, as shown in FIG.
Similarly, three-dimensional reconstruction is performed on the two images FD2 obtained in the second shooting to generate three-dimensional shape data DL2. First and second three-dimensional shape data DL1, D
L2 is compared, and the displacement of the multi-lens camera 3 is obtained from the difference between the characteristic points. According to the displacement, the image FD2
Change the position where you attach.

As shown in FIG. 8, three-dimensional shape data DL
The two vertices PL2 (x2, y2, z2) correspond to the vertices PL1 (x1, y1, z1) of the three-dimensional shape data DL1. However, when these three-dimensional shape data DL1 and DL2 are superimposed, their positions are shifted.
The vertex PL2 (x2, y2, z2) of L2 matches the point PL2a (x2a, y2a, z2a) of the three-dimensional shape data DL1.

Therefore, if the image FD2 is
When the mapping is performed on the three-dimensional shape data DL1, the point P2 (u2, v2) on the image FD2 is glued to the point PL2a (x2a, y2a, z2a) of the three-dimensional shape data DL1.

However, by performing the correction based on the displacement between the three-dimensional shape data DL1 and DL2, the image FD2
The upper point P2 (u2, v2) is the three-dimensional shape data DL1
Is correctly glued to the vertex PL1 (x1, y1, z1) of.

Next, correction processing and generation of the three-dimensional model ML in the generation device 1 will be described with reference to a flowchart. FIG. 9 is a flowchart showing a flow of a three-dimensional model generation process in the generation device 1.

In FIG. 9, the object Q is photographed twice by the multi-lens camera 3 with and without pattern light (# 1).
1). The order of photographing with and without pattern light may be any order.

Using the image FD photographed with the pattern light or the image FD photographed without the pattern light, three-dimensional reconstruction is performed to obtain three-dimensional shape data DL (# 12). The position shift of the camera when the images with and without the pattern light are respectively photographed is obtained (# 13). The image FD is attached to the three-dimensional shape data DL by correcting the positional deviation (# 14).

Hereinafter, each step will be described in more detail. In the following, a specific one of the multi-lens cameras 3 will be described.
One camera may be called a "reference camera" and the other camera may be called a "reference camera."

In step # 12, corresponding points are searched for the image, and based on the correspondence, three-dimensional coordinates of each point are obtained based on the principle of triangulation. A known method such as a block correlation method or a method of solving a gradient equation can be used as a corresponding point search method.

In step # 13, in the first correction method, first, an image with pattern light is compared with an image for bonding (sometimes called a "texture image"). A point having a large feature amount (hereinafter, simply referred to as a "feature point") such as a point existing on an edge is associated with an image with pattern light and an image for bonding. Depending on the presence or absence of the pattern light, it may be difficult to determine the correspondence, but if the properties (color, shape, etc.) of the pattern to be projected are known, the data is used when searching for corresponding points. And effective.

Then, based on the positional relationship between the associated feature point and the point corresponding to this feature point on the three-dimensional shape data DL obtained in step # 12, the reference camera for shooting without pattern light is used. Find the projection matrix of. This projection matrix reflects a shift in camera position between when photographing with pattern light and when photographing without pattern light.

Generally, if the three-dimensional coordinates of six or more points on an image are known, the projection matrix P can be obtained by solving the simultaneous equations shown in the following equation (1). Thus, the projection matrix of the reference camera at the time of capturing the image for mounting is obtained.

[0048]

(Equation 1)

The projection matrix is described in "3.
Dimension Vision, Kyoritsu Shuppan, published April 30, 1998,
Chapters 2 and 6 can be referred to. further,
Even if the camera position shifts, the internal parameters do not change. Therefore, since the internal parameters can be calculated from the projection matrix of the reference camera at the time of shooting with a known pattern light,
Using the decomposition of the projection matrix P into the internal parameter component and the external parameter component shown in the following equation (2), it is possible to calculate the external parameters of the reference camera at the time of capturing the image for mounting.

P = AM (2) where P is a projection matrix, A is an internal parameter component, and M is an external parameter component.

[0051]

(Equation 2)

Here, R is a 3 × 3 orthogonal matrix, and t is a three-dimensional vector. Further, since P has a degree of freedom of a constant multiple, this equation means "equal except for a constant multiple".

Since the external parameter components of the projection matrix have the constraint conditions as shown in the above equation (2), errors can be reduced by using them.
In step # 14, by using the projection matrix of the reference camera at the time of photographing the image for mounting determined in step # 13, the image for mounting can be mounted without displacement to the three-dimensional shape data DL.

In step # 13, in the second correction method, first, on the two three-dimensional shape data DL1 and DL2, the coordinates of some of the above-mentioned characteristic points, usually six or more characteristic points are obtained. Using the obtained coordinates, a matrix representing the change (rotation, translation) of the position of the camera between the time of shooting with the pattern light and the time of shooting the image for bonding is calculated by using the simultaneous equation shown in the following equation (3). Find by solving.

[0055]

(Equation 3)

Since the matrix component also has the constraint conditions shown in the above equation (3), an error can be reduced by using the constraint condition. Using this and the projection matrix at the time of shooting with pattern light, the projection matrix at the time of shooting the image for mounting may be obtained. The equation for finding this is shown in the following equation (4).

P ′ = PM (4) where P ′ is the projection matrix of the reference camera when capturing the texture image (unknown), and P is the projection matrix of the reference camera when capturing the image with pattern light (known). , M indicate a matrix (known) representing the rotation and translation obtained by the equation (3).

In the above embodiment, the multi-lens camera 3 is 2
Although an eye camera was used, three or more images FD with parallax were acquired using a camera with three or more eyes, and these three images F
The reliability of the three-dimensional shape data may be determined based on D, and only highly reliable data may be used as a feature point.

That is, for example, a plurality of three-dimensional shape data obtained from the image of the reference camera and the image of each reference camera are compared, and a point having a high degree of coincidence of the three-dimensional coordinates for each of the three-dimensional shape data is determined. It is extracted as a highly likely point and used instead of a feature point.

In this case, a program for reliability determination processing stored in the magnetic disk device 11 is executed. An example of the content of the reliability determination processing is as follows. That is, the reliability is obtained from the feature on the image FD2 photographed without projecting the pattern light on the object Q. Further, when there are three images FD1 and FD3, the three-dimensional reconstruction unit 31 calculates the reliability according to the degree of coincidence of each corresponding point of a plurality of three-dimensional shape data generated from the three images. . Then, when comparing the two pieces of three-dimensional shape data DL1 and DL2, the shift calculating unit 32 adds an evaluation according to the reliability to each point of each of the pieces of three-dimensional shape data DL1 and DL and compares them.

Further, weighting is applied to each point having a certain degree of reliability or more, for example, each point having a degree of coincidence on the three-dimensional shape data which is equal to or more than a certain value, according to the degree of coincidence. And
When using the least squares method or the like, the weight may be reflected in the evaluation function.

In this way, the projection matrix at the time of photographing the image for mounting is obtained, and the image is mounted on the three-dimensional shape data using the matrix. Even if the position of the camera is deviated, the glued image does not deviate from the three-dimensional shape data. As a result, an accurate three-dimensional model ML can be generated.

In the embodiment described above, the pattern light is projected in the first photographing, but the pattern light may be projected in the second photographing without projecting the pattern light in the first photographing. If feature points can be extracted from the object Q, the pattern light need not be projected. 3
Two images F with parallax for generating two-dimensional shape data
Three-dimensional reconstruction was performed from D.
The three-dimensional shape data DL may be obtained directly from the target using a dimension measurement device.

In general, there is also known a method of performing three-dimensional measurement by coding each part of a subject with pattern light and detecting a position where each code pattern is received (space coding method). The present invention is also applicable to this spatial coding method. I mean,
As in the above-described embodiment, the imaging is performed once in a state where the pattern light is projected onto the target object by the spatial coding method and in a state where the pattern light is not projected, and a two-dimensional image is attached while correcting a shift between them. , A three-dimensional model of the object may be generated.

In the above embodiment, the structure of the whole or each part of the multi-lens camera 3, the processing device 5, or the generation device 1,
The shape, size, number, material, processing content or order, etc. can be appropriately changed in accordance with the gist of the present invention.

[0066]

According to the present invention, an accurate three-dimensional model can be generated.

[Brief description of the drawings]

FIG. 1 is a diagram showing a generation device according to the present invention.

FIG. 2 is a block diagram illustrating a configuration example of a processing apparatus.

FIG. 3 is a diagram showing a state in which an object is continuously photographed by a multi-lens camera.

FIG. 4 is a timing chart at the time of continuous shooting.

FIG. 5 is a diagram showing a flow of generation of a three-dimensional model by a first correction method.

FIG. 6 is a diagram illustrating a first correction method.

FIG. 7 is a diagram showing a flow of generating a three-dimensional model by a second correction method.

FIG. 8 is a diagram illustrating a second correction method.

FIG. 9 is a flowchart illustrating a flow of a process of generating a three-dimensional model in the generation device.

[Explanation of symbols]

REFERENCE SIGNS LIST 1 generation device 3 multi-lens camera 3k, 3l camera 3p pattern light projection device 3r continuous shooting control unit 5 processing device 11 magnetic disk device (first storage unit, second storage unit, third storage unit) 31 three-dimensional Reconstruction unit 32 Misalignment calculation unit 33 Mapping unit (bonding unit) FD, FD1, FD2 Image (2D image) DL 3D shape data ML 3D model

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B050 BA04 BA06 BA09 BA10 DA02 DA07 EA12 EA13 EA17 EA27 EA28 5B057 BA02 BA13 CA08 CA13 CA16 CB08 CB13 CB16 CD01 CH11 DA07 DB03 DB09 DC32 5B080 BA07 DA06 GA22 5L096 CA04 CA17 EA14

Claims (15)

[Claims] A step of generating three-dimensional shape data of the object by measuring the object in a non-contact manner; a step of photographing the object to obtain a two-dimensional image of the object; The two-dimensional image obtained by photographing the object is compared with the two-dimensional image obtained by photographing the object, and the two-dimensional image obtained by photographing the object is corrected. A method for generating a three-dimensional model, comprising the steps of: 2. An apparatus for generating a three-dimensional model of an object, comprising: means for generating three-dimensional shape data of the object based on data obtained by measuring the object in a non-contact manner; Means for photographing the object to obtain a two-dimensional image of the object; measuring the object based on the two-dimensional image obtained when the object is measured and the two-dimensional image obtained by photographing the object; Means for determining a shift in viewpoint between the shooting of the target object and a means for correcting a determined shift in the viewpoint of a two-dimensional image obtained by shooting the target object and attaching the corrected two-dimensional image to the three-dimensional shape data. An apparatus for generating a three-dimensional model. 3. An apparatus for generating a three-dimensional model of an object, wherein the first storage stores a plurality of two-dimensional images having parallax of the object obtained by one photographing with a multi-view camera. Unit, a second storage unit that stores a two-dimensional image of the object obtained by another one of the shootings by the multi-lens camera, and an object based on the two-dimensional image stored in the first storage unit. A three-dimensional reconstruction unit for generating three-dimensional shape data of an object; a two-dimensional image stored in the second storage unit;
1 based on the two-dimensional image stored in the storage unit
A shift calculating unit for calculating a shift of a viewpoint between the shooting of the one and the shooting of the other one, and correcting the obtained shift of the viewpoint on the two-dimensional image stored in the second storage unit to obtain the three-dimensional shape. An apparatus for generating a three-dimensional model, comprising: a bonding part for bonding to data; 4. The two-dimensional image stored in the first storage unit is obtained by projecting pattern light onto an object and photographed. The two-dimensional image stored in the second storage unit is The three-dimensional model generation device according to claim 3, wherein the image is photographed without projecting the pattern light onto the object. 5. A multi-lens camera, a pattern light projection device for projecting pattern light onto an object, and one photographing performed by projecting the pattern light and another photographing performed without projecting the pattern light. The three-dimensional model generation device according to claim 3, further comprising: a continuous shooting control unit that causes the multi-lens camera to execute the shooting continuously. 6. An apparatus for generating a three-dimensional model of an object, the apparatus storing a plurality of first two-dimensional images having parallax about the object obtained by one photographing with a multi-view camera. One storage unit; a second storage unit that stores a plurality of second two-dimensional images having a parallax with respect to the target object obtained by the other one shooting by the multi-eye camera; A three-dimensional reconstruction unit that generates three-dimensional shape data of the object based on the two-dimensional image; and a three-dimensional reconstruction unit that is generated by the three-dimensional reconstruction unit from the first two-dimensional image and the second two-dimensional image. A third storage unit that stores first three-dimensional shape data and second three-dimensional shape data; and compares the first three-dimensional shape data with the second three-dimensional shape data. A shift calculator for determining the shift of the second The original image, generating apparatus of a three-dimensional model of the determined offset correction to the characterized by comprising a, a pasting unit for pasting the first three-dimensional shape data. 7. The first two-dimensional image is photographed by projecting pattern light onto an object, and the second two-dimensional image is photographed without projecting pattern light onto the object. The apparatus for generating a three-dimensional model according to claim 6, wherein: 8. A reliability determining unit for determining a reliability of each point of the three-dimensional shape data, wherein the displacement calculating unit calculates the first three-dimensional shape data and the second three-dimensional shape data. The three-dimensional model generation device according to claim 6, wherein, when performing the comparison, each point of each three-dimensional shape data is evaluated by adding an evaluation according to the reliability. 9. The three-dimensional model generation device according to claim 7, wherein the reliability determination unit obtains the reliability from a feature on an image of the two-dimensional image photographed without projecting the pattern light onto the object. . 10. The three-dimensional reconstruction unit according to claim 3, wherein the plurality of first two-dimensional images and the plurality of second two-dimensional images are three each. The three-dimensional model generation device according to claim 7, wherein a reliability is obtained according to a degree of coincidence of each corresponding point of a plurality of three-dimensional shape data generated from an image. 11. A multi-lens camera, a pattern light projection device for projecting pattern light onto an object, and one photographing performed by projecting the pattern light and another photographing performed without projecting the pattern light. The three-dimensional model generation device according to claim 6, further comprising: a continuous shooting control unit that causes the multi-lens camera to execute the shooting continuously. 12. A computer program for generating a three-dimensional model of an object, wherein the computer program stores a plurality of two-dimensional images having parallax of the object obtained by one photographing with a multi-view camera. A first storage unit, a second storage unit that stores a two-dimensional image of the target object obtained by another one of shootings by the multi-lens camera, and a two-dimensional image stored in the first storage unit A three-dimensional reconstruction unit that generates three-dimensional shape data of the object based on the two-dimensional image stored in the second storage unit;
1 based on the two-dimensional image stored in the storage unit
A shift calculating unit for calculating a shift of a viewpoint between the shooting of the one and the shooting of the other one, and correcting the obtained shift of the viewpoint on the two-dimensional image stored in the second storage unit to obtain the three-dimensional shape. A computer program for causing a computer to execute a process for functionally realizing the function of a bonding portion for bonding data. 13. A computer program for generating a three-dimensional model of an object, comprising: a plurality of first two-dimensional images having parallax of the object obtained by one photographing with a multi-view camera. A second storage unit that stores a plurality of second two-dimensional images having a parallax with respect to an object obtained by another one of shootings by the multi-view camera; A three-dimensional reconstruction unit that generates three-dimensional shape data of an object based on a plurality of two-dimensional images having a plurality of two-dimensional images, and the three-dimensional reconstruction unit from the first two-dimensional image and the second two-dimensional image A third storage unit that stores the generated first three-dimensional shape data and second three-dimensional shape data, and compares the first three-dimensional shape data with the second three-dimensional shape data. , Find the shift between them A computer executes a process for functionally realizing a shift calculating unit, and a gluing unit for correcting the obtained shift of the second two-dimensional image to the first three-dimensional shape data. Computer program to make it work. 14. A computer-readable recording medium on which the computer program according to claim 12 is recorded. 15. A computer-readable recording medium on which the computer program according to claim 13 is recorded.