JPH10302068A - Stereoscopic model preparation device, stereoscopic model preparation method and medium recording stereoscopic model preparation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prepare a more accurate stereoscopic shape by preparing the stereoscopic shape formed by gathering a partial area turned to an established shape from a judgement processing and a temporary stereoscopic shape. SOLUTION: The observability of the partial areas 3-6 is evaluated by the number (x) of directions capable of observation without hiding the partial areas 3-6 by the temporary stereoscopic shape 2 or hiding the temporary stereoscopic shape 2 by the partial areas 3-6, that is regardless of the temporary stereoscopic shape 2, at the time of assuming that the temporary stereoscopic shape 2 is to be observed (projected) from the plural directions. Then, in the case that the obtained presence estimation rate of the partial areas 3-6 of the peripheral area of the temporary stereoscopic shape 2 becomes more than a prescribed threshold value, it is judged that the partial areas 3 6 are stereoscopic shape present areas (established shape parts). In such a manner, whether or not acquisition is to be performed as the respective partial areas 3-6 of the peripheral area of the temporary stereoscopic shape 2 is judged, the partial areas 3-6 judged as to be acquired and the temporary stereoscopic shape 2 are gathered and the stereoscopic shape from an object 1 is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a three-dimensional model creation device, a three-dimensional model creation method, and a medium in which a three-dimensional model creation program is recorded.

[0002]

2. Description of the Related Art As a conventional method of creating a three-dimensional model, a target object is photographed from a plurality of directions using a photographing machine such as a television camera, and contour information (silhouette image) of an object image obtained from the photographed object is obtained. A method of reconstructing a three-dimensional shape (three-dimensional model) as three-dimensional coordinates in a computer as a voxel (VOXEL) model by a cone-penetration method or a voting method (voting method) is known.

[0003]

However, in the method using the cone-penetration method, even if an error exists in one of the silhouette images, a large difference occurs between the created three-dimensional shape and the target object. . For example, if the target object has a common color scheme with the photographing background when viewed from a certain direction, an undesired missing portion (a hole, a notch, etc.) may be present in the silhouette image obtained by photographing from this direction. As a result, there is a problem that a three-dimensional shape having a missing portion caused by the undesired missing portion is created.

In the voting method, the threshold value for the number of votes can be set to reduce the occurrence of missing portions due to undesired missing portions in the silhouette image from the three-dimensional shape. In the method, when the appearance of the target object is mainly composed of a planar shape (for example, in the case of a cube or the like), there is a problem that a three-dimensional shape in which the central portion of the planar shape in the target object is expanded is created. Occurs.

The present invention has been made in view of the above-described problems, and provides a three-dimensional model creating apparatus, a three-dimensional model creating method, and a medium in which a three-dimensional model creating program capable of creating a more accurate three-dimensional shape is recorded. That is the purpose.

[0006]

SUMMARY OF THE INVENTION The present invention relates to a three-dimensional model creating apparatus for creating a three-dimensional model of a target object, wherein the temporary three-dimensional shape as a definite shape part is determined based on a plurality of object images of the target object. The provisional three-dimensional shape creating means for creating the three-dimensional shape, and considering the existence of the provisional three-dimensional shape, with respect to the partial region around the provisional three-dimensional shape, determines whether or not the partial region is a definite shape portion to the reliability of the object image. Determination processing means for performing a determination process for determining based on the three-dimensional shape, and a three-dimensional shape creation means for creating a three-dimensional shape by combining the partial region and the temporary three-dimensional shape as a final shape portion from the determination process Features.

According to the present invention, a tentative three-dimensional shape determined as a determined shape portion based on the reliability of an object image while considering the existence of the tentative three-dimensional shape as a determined shape portion and the existence of the tentative three-dimensional shape as the determined shape portion. Since the three-dimensional shape is obtained by combining the partial regions around the three-dimensional shape, it is possible to suppress undesired omission of the three-dimensional shape due to an undesired omission portion present in the object image,
Even when the appearance of the target object is mainly composed of a planar shape, it is possible to suppress the central portion of the planar shape portion of the target object from expanding.

Further, the present invention is a three-dimensional model creating apparatus for creating a three-dimensional model of a target object, wherein a temporary three-dimensional shape as a fixed shape part is created based on a plurality of object images of the target object. The three-dimensional shape creating means may determine, for each of the plurality of directions, the number of directions in which the partial region forming the periphery of the temporary three-dimensional shape can be observed without covering the temporary solid shape. Observability determining means, and for each of the partial areas, the object in which different parts can be found in the object image of the target object corresponding to the partial area and the object image of the temporary three-dimensional shape corresponding to the object image A difference part number obtaining means for obtaining the number of sets of the object image of the object and the object image of the temporary three-dimensional shape, and for each of the partial regions, a value of a ratio or a difference between the number of the directions and the number of the sets,
A determined shape portion determining means for determining whether or not the partial region is a defined shape portion by determining this value with a predetermined threshold value; and a three-dimensional shape combining the temporary solid shape and the determined partial shape portion. And a three-dimensional shape creating means for creating a shape.

In these inventions, in order to obtain object images relating to the plurality of target objects, for example, the target object is photographed in a plurality of directions including a background, for example, in a circumferential direction, and the background is photographed. And a difference between a background object image obtained by shooting the target object including the background by the shooting means and a background image obtained by shooting only the background by the shooting means, to obtain a plurality of object images. A silhouette creating means for creating may be provided.

The present invention relates to, for example, a partial region,
With respect to the number of circumferential directions in which a partial region surrounding the temporary three-dimensional shape can be observed without covering the temporary three-dimensional shape, the object image corresponding to the target object corresponding to the partial region and the object image corresponded (that is, A ratio of the number of pairs of the object image related to the target object and the object image related to the temporary three-dimensional shape in which a different part is found between the object image related to the temporary three-dimensional shape and the object image related to the partial three-dimensional shape; In the case where the existence estimation rate at which the part is present is equal to or greater than a predetermined threshold value, the certain partial area is determined as a confirmed shape part.

Then, since the three-dimensional shape is obtained by combining the previously determined temporary three-dimensional shape as the definite shape portion and the partial region around the temporary three-dimensional shape determined as the definite shape portion, undesired three-dimensional shapes existing in the object image are obtained. Unwanted loss of the three-dimensional shape due to the missing part can be suppressed, and even if the appearance of the target object is mainly composed of a planar shape, the central part of the flat part of the target object expands. Can be suppressed.

When the object is rotated from a plurality of directions (circumferential directions) including the background by the photographing means, for example, when the object itself is rotated, one photographing of the background is sufficient. However, in the case where the photographing means photographs around the target object, background photographing in a plurality of directions is necessary.

The present invention is a method for creating a three-dimensional model of a target object, comprising the steps of: using a plurality of object images relating to the target object, forming a temporary three-dimensional shape as a definite shape part; While considering the existence of the temporary three-dimensional shape,
A step of performing a determination process for determining whether or not the partial region is a definite shape portion based on the reliability of the object image with respect to the partial region around the temporary three-dimensional shape; And forming a three-dimensional shape by combining the partial region and the provisional three-dimensional shape.

According to the present invention, a provisional three-dimensional shape which is determined as a confirmed shape portion based on the reliability of the object image while considering the existence of the provisional three-dimensional shape as the confirmed shape portion and the existence of the provisional three-dimensional shape as the confirmed shape portion. Since the three-dimensional shape is obtained by combining the partial regions around the shape, it is possible to suppress the undesired loss of the three-dimensional shape due to the undesired missing portion present in the object image, and to reduce the appearance of the target object mainly from the planar shape. Even in the case where the target object is configured, it is possible to suppress the central portion of the portion having the planar shape in the target object from expanding.

[0015] The present invention is a three-dimensional model creation method for creating a three-dimensional model of a target object, comprising the steps of: preparing a plurality of object images of the target object; A process of creating a three-dimensional shape,
A step of individually creating an object image relating to the temporary three-dimensional shape corresponding to each of the object images from the temporary three-dimensional shape; And the step of determining the number of circumferential directions that can be observed without covering each of the partial regions constituting the periphery, and for each of the partial regions, an object image of the target object corresponding to the partial region, A step of determining the number of sets of the object image relating to the target object and the object image relating to the provisional three-dimensional shape in which a different part can be found in the corresponding object image relating to the provisional three-dimensional shape; Determining the value of the ratio or difference between the number of directions and the number of sets, determining this value with a predetermined threshold value, thereby determining whether or not the partial region is a definite shape portion; and Defined shape part and size Characterized in that it comprises by the steps of creating a three-dimensional shape matching the partial area, a.

According to the present invention, for example, an object image of the target object corresponding to a partial region is defined with respect to the number of circumferential regions that can be observed without covering a partial region surrounding the temporary solid shape with the temporary solid shape. The ratio of the number of sets of the object image related to the target object and the object image related to the provisional three-dimensional shape in which a different part can be found in the object image related to the provisional three-dimensional shape corresponding to the object image (that is, an undesired missing part exists). Is smaller than or equal to a predetermined threshold value, the partial region is determined to be a confirmed shape portion.

The three-dimensional shape is obtained by combining the previously determined temporary three-dimensional shape as the final shape portion and the partial area around the temporary three-dimensional shape determined to be the final shape portion. Unwanted loss of the three-dimensional shape due to the missing part can be suppressed, and even if the appearance of the target object is mainly composed of a planar shape, the central part of the flat part of the target object expands. Can be suppressed.

In particular, it is characterized in that the provisional three-dimensional shape is prepared using a cone-penetration method. In this case, since the reliability of the provisional three-dimensional shape as the final shape portion is increased, even when the appearance of the target object is mainly composed of a planar shape, the central portion of the planar shape portion of the target object expands. Can be better suppressed.

The plurality of object images relating to the target object can be obtained by, for example, photographing the target object including the background from a plurality of directions (a plurality of circumferential directions) around the target object by a photographing means such as a camera. The obtained background object image is prepared, the background image obtained by photographing the background is prepared, and the corresponding background image is subtracted from each of the background object images. It is preferable that the plurality of directions around the target object be substantially even around the target object.

Further, the number of circumferential directions which can be observed without overlapping the temporary solid shape with the partial region constituting the periphery of the provisional three-dimensional shape is determined for each of the partial regions constituting the periphery. It is preferable that the plurality of circumferential directions in the process be substantially uniform around the target object, and it is preferable that the plurality of circumferential directions be the same as the plurality of directions around the target object at the time of the photographing.

In addition, it is preferable that the tentative three-dimensional shape can be subjected to a voting process in which, for example, a definite shape portion is obtained in a voxel space based on an object image of the target object when all votes are acquired. The number of votes near the total number of votes may be used.

Further, for each of the partial regions, the target object in which different parts can be found between the object image of the target object corresponding to the partial region and the object image of the temporary three-dimensional shape corresponding to the object image The number of sets of each partial region in the step of obtaining the number of sets of the object image according to the above and the temporary solid shape is, for example, the number of votes in the voting process from the different portion to the partial region in the voxel space. Can be obtained as

Further, for each of the partial regions, a ratio between the number in the circumferential direction and the number of sets is obtained, and the ratio is determined with a predetermined threshold value to determine whether or not the partial region is a fixed shape portion. When determining, for example, when the ratio of the number of sets to the number in the circumferential direction is equal to or greater than a predetermined threshold value, the partial region may be determined to be a confirmed shape portion. In this case, of a plurality of circumferential directions, the number of circumferential directions that can be observed without overlapping the temporary solid shape with the partial region forming the temporary solid shape is determined for each of the partial regions forming the circumference. When the plurality of circumferential directions in the obtaining step are the same as the plurality of directions around the target object, the predetermined threshold value may be, for example, 0.5.

Also, for each of the partial areas, a value of a difference between the number in the circumferential direction and the number of sets is obtained, and this value is determined with a predetermined threshold value to determine whether or not the partial area is a fixed shape part. When it is determined that the partial area is set to 1 and the (the number in the circumferential direction−the number of sets) is equal to or less than the threshold, the partial region may be determined to be the confirmed shape portion.

[0025]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment according to the present invention will be described below with reference to the drawings. First, the principle of the method according to the present embodiment will be described with reference to FIGS. Here, a case where the target object is a cube, and photographing, projection, and the like use an orthographic projection method for simplicity of description will be described.

First, the target object 1 shown in FIG. 1 is photographed from a plurality of directions (a plurality of circumferential directions) around an arbitrary axis (here, a central axis 1), and the first object related to the target object 1 from each of the plurality of directions is taken. One silhouette image (object image) a1 to h1 is obtained, and a tentative three-dimensional shape (determined shape portion) from the target object 1 shown in FIG. 2 is obtained by the cone-penetration method using the plurality of first silhouette images a1 to h1. Get 2. Here, for the sake of simplicity, the plurality of directions will be described as eight directions (directions A to H).

Next, the temporary three-dimensional shape 2 may have an undesired missing portion (error portion) as compared with the target object 1. Obtain the existence estimation rate. Here,
A silhouette error 7 corresponding to an undesired missing portion exists in the first silhouette image e1 in FIG.

Hereinafter, the method of obtaining the existence estimation rate will be described focusing on the partial area 3 in the surrounding area of the provisional three-dimensional shape.
The existence estimation rate is represented by a ratio of the number of possible votes for the partial area to the observability of the partial area.

First, a method for determining the observability of the partial area 3 will be described.

The observability is determined by observing (projecting) the provisional three-dimensional shape 2 from a plurality of directions.
Is not hidden by the temporary three-dimensional shape 2 and the temporary three-dimensional shape 2 is
Is evaluated by the number (X) of directions that can be observed without being hidden, that is, regardless of the provisional three-dimensional shape 2 (the partial region 3 and the provisional three-dimensional shape 2 do not overlap).

In this case, the partial region 3 is observed irrespective of the provisional three-dimensional shape 2 in four directions of the direction A, the direction C, the direction E, and the direction G.
Becomes

Next, a method (voting method) for obtaining the number of possible votes (Y) for the partial area will be described.

First, as shown in FIG. 3, it is assumed that the tentative three-dimensional shape 2 is projected from the plurality of directions in the same manner as in obtaining the first silhouette images a1 to h1. Two silhouette images (object images: referred to as core silhouette images) a2 to h2 are obtained. Here, each of the first silhouette images a1 to h1 has a corresponding relationship with each of the second silhouette images a2 to h2. That is, when the provisional three-dimensional shape 2 matches the target object 1, each of the first silhouette images a1 to h1 becomes
The second silhouette images a2 to h2 are obtained from the provisional three-dimensional shape 2 so as to coincide with h2.

Next, each of the second silhouette images a2 to h2 corresponding to the first silhouette images a1 to h1 is compared with each other, and a difference between the first silhouette image and the second silhouette image which are not coincident is extracted. . Here, the first silhouette image a1 and the corresponding second silhouette image a2 are different, and in FIG.
There is only one.

Then, the number of different portions in which the partial area can be observed (voting on the partial area) is counted as the number of possible votes of the partial area. The number of possible votes (Y) for the partial area 3 is 1.

The number of the above directions (X =
Using 4) and the number of possible votes obtained as described above (Y = 1), the existence estimation rate Y / X = 1/4 in which the partial area 3 is an undesired missing part is obtained.

Similarly, since X is 2 and Y is 1 for the partial region (undesired missing portion) 4, the existence estimation rate is Y / X = 1/2, and for the partial regions 5 and 6, In both cases, Y / X = 0/2, and other partial regions are similarly obtained.

When the estimated existence rate Y / X of the partial area around the temporary three-dimensional shape 2 obtained as described above is equal to or greater than a predetermined threshold value, the partial area is determined as the three-dimensional shape existing area. (Determined shape portion). In this way, it is determined whether or not to acquire each partial region of the surrounding region of the temporary three-dimensional shape 2 as a three-dimensional shape, and the partial region determined to be acquired is combined with the temporary three-dimensional shape 2 to obtain a three-dimensional shape from the target object 1. Get.

In such a method, even when the target object has a plane portion such as a rectangular parallelepiped or a cube, the portion corresponding to the plane portion in the three-dimensional model is obtained by using the provisional three-dimensional shape obtained by the cone-penetration method. It can be created well, and for example, by judging with a predetermined threshold the existence estimation rate of undesired missing parts obtained from the number of possible votes obtained by voting processing and observability, Desired missing portions can be relieved.

It should be noted that the cone-penetration method referred to in this specification also includes a voting method in which, when the total number of votes has been obtained, a three-dimensional shape exists.

Next, the present embodiment will be described in more detail. Here, the target object is a cube,
The case where the perspective projection method is used for photographing and projection will be described.

FIG. 4 is a schematic diagram showing the overall configuration of the three-dimensional model creation device of the present embodiment. In FIG. 4, reference numeral 11 denotes a turntable that can be rotated with the target object placed thereon, 13 denotes a camera (imaging device) for shooting the target object to obtain a silhouette image of the target object, and 15 denotes an image captured by the camera 13 It is a computer for extracting a silhouette image of the target object from the obtained image and creating a three-dimensional model based on the silhouette image.

FIG. 5 is a block diagram showing an outline of the above-mentioned three-dimensional model creating apparatus. In FIG. 5, 19 is a turntable 11
A photographing unit including the photographing device 13; an image storage unit 21 for storing an image photographed by the photographing unit 19; a calculation / control unit 23 for performing various arithmetic processing and control; 25 a shape storage unit;
Reference numeral 7 denotes a color information storage unit. Here, the image storage unit 21,
The calculation / control unit 23, the shape storage unit 25, and the color information storage unit 27 are built in the computer 15.

FIG. 6 is a diagram for explaining the flow of processing for creating a three-dimensional model by the three-dimensional model creating apparatus. FIG. 7 is a diagram for explaining the specific contents of the processing in the three-dimensional model creation device of FIG. 4, and FIG.
FIG. 7B is a diagram for explaining the shooting of the target object / background in step S2 of FIG. 7, and FIG. 7B is a diagram for explaining the creation of the first silhouette image in step S3 of FIG.
FIG. 7C is a view for explaining the first voting process (cone-penetration method) in step S4 of FIG. 6, and FIG. 7D.
FIG. 7 is a diagram for explaining a process of calculating an existence estimation rate of an undesirable missing portion around the temporary three-dimensional shape obtained by performing the first boarding process in step S5 of FIG. 6;
FIG. 7E is a diagram for explaining the process of determining the three-dimensional shape in step S6, FIG. 7F is a diagram for creating a polygon in step S7 of FIG. 6, and FIG. 7G is a diagram of step S8 in FIG. FIG. 4 is a diagram for explaining a texture map in FIG.

First, calibration is performed as step S1. That is, in the present embodiment, the camera 13
Is determined (perspective ratio). In the present embodiment, the camera 13 is fixedly arranged so that the target object can be photographed in the horizontal direction.

Next, as step S2, the background and the target object 33 are photographed using the camera 13. That is, the background image A0 is obtained by photographing the background without placing the target object on the turntable 11, and the target object 33 is placed on the turntable 11, and the turntable 11 is set to a plurality of predetermined rotational positions (target object). Is rotated by a predetermined angle around its central axis), and the camera 13 is fixed as described above to fix the relative positional relationship with the turntable 11 so as to photograph the target object 33 together with the background. ~ An is obtained. The background image A0 and the object images A1 to A1
An is stored in the image storage unit 21. In the present embodiment, the target object 33 is rotated every 10 ° to obtain background object images A1 to A36. Hereinafter, a case in which a three-dimensional model is created based on the background object images A1 to A36 thus obtained will be described.

Subsequently, as a step S3, the silhouette creating section (not shown) executes the above-mentioned background object images A1 to A36.
From the background image A0 to calculate the difference between the color information of the background object image and the color information of the background image for each pixel, and selectively extract the image of the object from each of the background object images A1 to A36. First silhouette image (object image)
B1 to B36 are created respectively.

Thereafter, as a step S4, the provisional three-dimensional shape creating unit including a voting unit (not shown) performs a first voting process. This first voting process is
A voting process is performed on the cylindrical coordinate system voxel space 35 for each of the first silhouette images B1 to B36, and a portion where the number of obtained votes matches the number of silhouette images (here, 36) is processed as a three-dimensional shape existence region. The provisional three-dimensional shape 36 composed of the existence area is determined as the determined shape part. Here, the temporary three-dimensional shape 36 is an aggregate of voxels in the voxel space 35.

Next, in step S5, the observability determining unit (not shown) determines the observability of each partial region (here, each voxel in the voxel space) of the surrounding region of the temporary solid shape 36, that is, each partial region. The number (X) of directions that can be observed in the region from the plurality of directions regardless of the provisional three-dimensional shape 36 is obtained.

In the same step, the number of possible votes (Y) for each of the above partial areas is obtained by a different part number acquiring unit (not shown). That is, assuming that the provisional three-dimensional shape 36 is projected from the plurality of directions as described above, the provisional three-dimensional shape 36
To obtain the second silhouette images (object images) C1 to C36. Each of the first silhouette images B1 to B36 is compared with each of the second silhouette images C1 to C36 corresponding to each other, and a difference between the first silhouette image and the second silhouette image that do not have a matching relationship is extracted.

Then, for each of the partial areas, the number of the partial areas observed from the different part is determined as the possible vote number (Y). That is, the number of different portions that can be observed (voted) without obstructing each partial region by the provisional three-dimensional shape is defined as the number of possible votes (Y). The number of possible votes (Y) is calculated from the difference from each partial area (each voxel) in the voxel space 35.
It may be obtained by a voting process that sets the number of votes to.

Then, the calculation / control unit 23, which is a fixed shape part judgment unit (judgment processing means), uses the observability X and the possible vote count Y for each partial region to determine the undesired shape for each partial region. The existence estimation rates Y / X, which are missing parts, are obtained.

Next, in step S6, the determined shape portion determination section determines that the existence estimation rate Y / X of the partial area around the temporary three-dimensional shape 36 is a predetermined threshold value (in this embodiment, the predetermined threshold value). Is 0.5) or more, it is determined that the partial area is a three-dimensional shape existing area (determined shape part). In this way, it is determined whether or not to acquire each partial region of the surrounding region of the temporary three-dimensional shape 36 as a three-dimensional shape, and the partial region determined to be acquired and the temporary three-dimensional shape 36 are combined to obtain a three-dimensional shape 37 from the target object. Is determined. Here, the three-dimensional shape 37 is, of course, an aggregate of voxels in the voxel space 35.

Thereafter, in step S7, step S6
A plurality of three-dimensional shape expressing elements (for example, polygons such as triangular patches, hereinafter, the three-dimensional shape expressing elements are referred to as polygons) 38 are formed based on the three-dimensional shape 37 of the target object 33 obtained in Target object 3 determined in S6
The three-dimensional shape 37 is represented by a plurality of polygons 38. The three-dimensional shape 37 represented by the polygon 38 is stored in the shape storage unit 2.
5 is stored.

Finally, in step S8, the texture (color information) corresponding to each polygon 38 created in step S7 and stored in the color information storage unit 27 is stored in the object image B1.
A three-dimensional shape 39 obtained from B36 and mapped to each polygon 38 is obtained.

Hereinafter, the calibration in step S1 will be described.
The voting process in steps S4 and S5 will be described in detail.

(Calibration) As calibration, the internal parameters (perspective ratio) of the camera 13 are obtained. Note that the perspective ratio is a parameter that determines the enlargement / reduction ratio of the size of the object projected on the screen that is enlarged or reduced according to the distance from the viewpoint to the object when projecting the object. FIG. 8 is a diagram for explaining internal parameters of the camera 13.

First, an image is taken by the camera 13 so that the reference block 41 having a height T just enters the screen 43, and the distance L between the camera 13 and the reference block 41 is measured.

The perspective ratio in this case is T / L obtained by dividing the height T of the reference block 41 by the distance L.

(Voting Process) FIG. 9 is a diagram for explaining a cylindrical voxel space in which a voting process similar to that shown in FIG. 7 is performed. The same parts as those in FIG. 7 are denoted by the same reference numerals.

In FIG. 9, the cylindrical voxel space 35 is divided by a plurality of voxels 49, 49,. The plurality of voxels 49, 49,..., When the cylindrical coordinate system voxel space 35 is considered as a cylinder 35 having a central axis 50, the cylinder 35 is equally divided by a plurality of planes parallel to the central axis 50, In addition, the element is divided equally by a plurality of planes perpendicular to the central axis 50 and divided by cylindrical surfaces having different diameters around the central axis 50.

FIG. 10 is a diagram for explaining the voting process. Note that the same parts as those in FIG. 9 are denoted by the same reference numerals, and description thereof is omitted. Here, step S3
In the example shown in FIG. 10, only silhouette images B1 and B2 are shown in FIG. 10, where voting processing is performed on the cylindrical coordinate system voxel space 35 based on the 36 first silhouette images B1 to B36 obtained in step 1. The following description focuses on two silhouette images B1.

First, the projection image 61 of the camera 13 is set as the vertex, and the silhouette image B in the cylindrical voxel space 35 is set.
A conical area having a cross-sectional shape that is a projection locus of the object image 52 (contour of the target object) in 1 is defined as a hypothetical existence area 60. This assumed existence area is B2 other than the silhouette image B1.
To B36.

Here, in the above description, the target object is photographed every 10 ° to obtain 36 object images, and 36 silhouette images B1 to B36 are created. Therefore, in this case, the vertices of the hypothetical existence region (the projection center of the camera 13) are located around the central axis 50 every 10 °.

The positions of the vertices of the hypothetical existence area are determined according to the result of the calibration. That is, the silhouette images B1 to B1 depend on the perspective ratio.
The positional relationship between B36 and the corresponding vertex of the assumed existence area is determined.

Also, depending on the positional relationship between the camera 13 and the rotary table 11, the vertices of the assumed existence area corresponding to the silhouette images B1 to B36 and the voxel space 35 in the cylindrical coordinate system 35
Is determined.

Thereafter, in the voting process, voting is performed for all the voxels 49 existing in the assumed existence area 60. This voting process is performed for all silhouette images B1.
To B36. For example, the number of votes in the voxel 49 in which the hypothetical existence areas corresponding to the 36 silhouette images B1 to B36 overlap is 36.

As a result of the voting process, a distribution with a large number of votes is obtained in a dark portion as shown in FIG. 7, and a distribution with a small number of votes is obtained in a light color portion.

FIG. 11 is a diagram showing a CD-ROM (compact disc read only memory) in which a program for creating a three-dimensional model of a target object is recorded on the computer 15 of FIG. 4 and 6 are denoted by the same reference numerals, and description thereof will be omitted as appropriate.

In FIG. 11, the computer 15 has a CD-R
According to the program 59 stored in the OM 57, a three-dimensional model of the target object is created. This program 59
Includes steps S3 for creating a first silhouette image of the target object to step S8 for performing a texture map.

As described above, in the above description, a three-dimensional shape is obtained by photographing a target object rotated every 10 ° using one camera and a rotary table. However, the present invention is not limited to this. The silhouette image may be obtained by photographing the target object by rotating the camera around the target object.

Although the voxel space is a voxel space in a cylindrical coordinate system, other voxel spaces such as a voxel space in a rectangular coordinate system may be used.

Further, in the above description, the number of possible votes is obtained by using the second silhouette image obtained from the provisional three-dimensional shape. However, a plurality of first silhouette images are arranged so that there is a first silhouette image that is opposed to each other. In the case of preparing the number of possible votes, the number of possible votes may be set as the number of sets of the first silhouette images in the mutually opposite relationship in which different portions can be found in the first silhouette images in the mutually opposite relationship corresponding to the partial areas.

In the above description, the number of possible votes (the number of votes:
An object image related to the target object corresponding to the partial region, and an object image related to the provisional three-dimensional shape corresponding to the object image,
The ratio between the number of pairs of the object image relating to the target object and the object image relating to the provisional three-dimensional shape) which can be found in the above-mentioned three-dimensional shape was determined using a threshold value. You may make it judge by using. in this case,
For example, when XY is equal to or smaller than 1 (threshold), the partial region may be determined as the confirmed shape portion.

[0075]

According to the present invention, it is possible to provide a three-dimensional model creating apparatus, a three-dimensional model creating method, and a medium recording a three-dimensional model creating program capable of creating a more accurate three-dimensional shape.

[Brief description of the drawings]

FIG. 1 is a perspective view illustrating a principle according to an embodiment of the present invention.

FIG. 2 is a top view for explaining the principle according to one embodiment of the present invention.

FIG. 3 is a top view for explaining the principle according to one embodiment of the present invention.

FIG. 4 is a schematic diagram showing an overall configuration of a three-dimensional model creation device according to one embodiment of the present invention.

FIG. 5 is a block diagram schematically showing the three-dimensional model creation device.

FIG. 6 is a flowchart illustrating a flow of a process of creating the three-dimensional model.

FIG. 7 is a diagram for explaining the specific contents of the process of creating the three-dimensional model.

8 is a diagram for explaining a perspective ratio obtained in step S1 of FIG.

FIG. 9 is a diagram for explaining voxels in a voxel space in a cylindrical coordinate system used in step S4 (S5) of FIG. 6;

FIG. 10 is a diagram for explaining voting processing in step S4 and the like in FIG. 6;

11 is a CD-ROM in which a program for causing the computer in FIG. 4 to create a three-dimensional model of a target object is recorded.
FIG.

[Explanation of symbols]

 Reference Signs List 1 target object 2 provisional three-dimensional shape a1 to h1 first silhouette image a2 to h2 second silhouette image 13 camera 15 computer 19 imaging unit 21 image storage unit 23 operation / control unit 25 shape storage unit 27 color information storage unit 33 target object 35 Voxel space in cylindrical coordinate system 37 3D shape 49 Voxel 57 CD-ROM (medium)

Claims (8)

[Claims] 1. A three-dimensional model creating apparatus for creating a three-dimensional model of a target object, comprising: a temporary three-dimensional shape creating means for creating a temporary three-dimensional shape as a fixed shape part based on a plurality of object images related to the target object. A determination process for determining, based on the reliability of the object image, whether or not the partial region is a definite shape portion with respect to the partial region around the temporary three-dimensional shape while considering the existence of the temporary three-dimensional shape And a three-dimensional shape creating means for creating a three-dimensional shape formed by combining the partial region that has been determined as the determined shape portion from the determination process and the temporary three-dimensional shape. 2. A three-dimensional model creating apparatus for creating a three-dimensional model of a target object, comprising: a temporary three-dimensional shape creating means for creating a temporary three-dimensional shape as a fixed shape portion based on a plurality of object images of the target object. And observability determining means for obtaining, for each of the plurality of directions, the number of directions in which the partial region constituting the periphery of the temporary three-dimensional shape can be observed without overlapping the temporary solid shape for each of the partial regions constituting the periphery. And, for each of the partial areas, an object according to the target object in which a different part can be found between the object image of the target object corresponding to the partial area and the object image of the provisional three-dimensional shape corresponding to the object image. Means for obtaining the number of sets of an image and an object image relating to the provisional three-dimensional shape; and obtaining, for each of the partial areas, a value of a ratio or difference between the number of the directions and the number of the sets, and set this value to a predetermined value. With a threshold A determined shape portion determining means for determining whether or not the partial region is a defined shape portion by cutting off; a three-dimensional shape creating means for creating a three-dimensional shape combining the temporary three-dimensional shape and the determined partial shape portion And a three-dimensional model creation device characterized by comprising: 3. The three-dimensional model creating apparatus according to claim 1, wherein the temporary three-dimensional shape is created by using a cone-penetration method. 4. A three-dimensional model creating method for creating a three-dimensional model of a target object, comprising the steps of: creating a temporary three-dimensional shape as a fixed shape portion using a plurality of object images related to the target object; While considering the presence of the shape, for the partial region around the provisional three-dimensional shape, performing a determination process for determining whether the partial region is a definite shape portion based on the reliability of the object image, A step of creating a three-dimensional shape formed by combining the partial region and the temporary three-dimensional shape as the final shape portion from the determination process;
A three-dimensional model creation method characterized by including: 5. A three-dimensional model creating method for creating a three-dimensional model of a target object, comprising: preparing a plurality of object images related to the target object; Creating an object image of the temporary three-dimensional shape corresponding to each of the object images from the temporary three-dimensional shape; and forming a portion around the temporary three-dimensional shape in a plurality of circumferential directions. A step of obtaining, for each of the partial regions constituting the periphery, the number of circumferential directions in which the region can be observed without covering the temporary three-dimensional shape; and for each of the partial regions, an object image of the target object corresponding to the partial region And determining the number of sets of the object image related to the target object and the object image related to the provisional three-dimensional shape in which a difference is found between the object image related to the provisional three-dimensional shape and the object image related to the provisional three-dimensional shape. For each partial area, a value of a ratio or a difference between the number in the circumferential direction and the number of sets is determined, and this value is determined with a predetermined threshold value to determine whether the partial area is a confirmed shape part. And a step of creating a three-dimensional shape by combining the provisional three-dimensional shape and the partial area determined to be the final shape part. 6. The method according to claim 4, wherein the provisional three-dimensional shape is created using a cone-penetration method. 7. A medium storing a program for causing a computer to create a three-dimensional model of a target object,
The program includes: using a plurality of object images related to the target object, creating a temporary three-dimensional shape as a confirmed shape portion; and considering a partial region around the temporary three-dimensional shape while considering the existence of the temporary three-dimensional shape. Performing a determination process for determining whether or not the partial region is a confirmed shape portion based on the reliability of the object image; and determining the partial region and the provisional three-dimensional shape as the confirmed shape portion from the determination process. A medium for recording a three-dimensional model creation program, comprising: a step of creating a three-dimensional shape to be combined. 8. A medium storing a program for causing a computer to create a three-dimensional model of a target object,
The program comprises the steps of: preparing a plurality of object images relating to the target object; creating a provisional three-dimensional shape as a confirmed shape portion from the plurality of object images; corresponding to each of the object images from the provisional three-dimensional shape Creating each of the object images relating to the provisional three-dimensional shape, and, among a plurality of circumferential directions, the number of the peripheral directions that can be observed without a partial region constituting the periphery of the provisional three-dimensional shape overlapping the provisional three-dimensional shape. A step of obtaining for each of the partial regions constituting the periphery, and for each of the partial regions, an object image of the target object corresponding to the partial region, and an object image of the temporary three-dimensional shape corresponding to the object image. Determining the number of sets of the object image relating to the target object and the object image relating to the provisional three-dimensional shape in which a different part can be found; and, for each of the partial regions, the number in the circumferential direction. Determining the value of the ratio or difference of the number of sets, determining whether or not the partial region is a definite shape portion by determining this value with a predetermined threshold value; and A step of creating a three-dimensional shape by combining the determined partial areas; and a medium storing a three-dimensional model creation program.