JP2002216161A - Image generator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily generate an image or animation according to direction, size or expression, etc., of a face of characters. SOLUTION: This image generator 1 generates a two-dimensional synthetic image, is provided with a two-dimensional synthesizing part 26 to superpose a head image GR2 on a base picture GR1, a three-dimensional data storage part 22 to store three-dimensional shape data KDT of the head of the user, a head image generating part 25 to generate the head image GR2, a base image storage part 13 to store image control information NM of the head image GR2 to be superposed on the base image GR1 by every scene and a scene setting part 23 to successively select scenes, the head image generating part 25 generates the head image based on size information NM2 and direction information NM3 and the two-dimensional synthesizing part 26 superposes the head image on a position based on positional information NM1.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image generating apparatus and a moving image generating apparatus for generating a two-dimensional image by transforming a face or the like of a character in a video game or the like into a direction, size, or expression suitable for a scene. .

[0002]

2. Description of the Related Art Hitherto, there has been known a video game in which a user's face is used as an image of a character's face. For example, a method of generating an image of a character in a video game using a two-dimensional image of the user's face photographed in advance can be considered.

[0003]

However, if the number of two-dimensional images of the user's face photographed in advance is small, changes in the expression of the characters or the orientation of the face are poor and awkward. Also, many images of characters whose faces are not familiar with the scene are generated.

Therefore, a method of using a large number of two-dimensional images obtained by photographing a user's face with various expressions and from various directions is conceivable. However, photographing takes a lot of time and burdens the user, which is not practical.

SUMMARY OF THE INVENTION In view of the above problems, it is an object of the present invention to provide an image generating apparatus and a moving image generating apparatus which can easily generate an image according to the orientation, size, or expression of a character's face. I do.

[0006]

An image generating apparatus according to the present invention is an image generating apparatus for synthesizing a plurality of two-dimensional images to generate one two-dimensional synthesized image. Superimposing means for superimposing the second two-dimensional image;
Three-dimensional data storage means for storing three-dimensional data of an object; two-dimensional image generation means for generating a second two-dimensional image from the three-dimensional data; and the second two-dimensional image superimposed on the first two-dimensional image Information storage means for storing, for each scene, position information, size information, and direction information of the object relating to the two-dimensional image, and scene selection means for sequentially selecting scenes, The means generates the second two-dimensional image based on the size information and the direction information corresponding to the currently selected scene, and the superimposing means generates the second two-dimensional image at a position based on the position information. Are superimposed.

Preferably, the three-dimensional data storage means stores three-dimensional data of a human head, and the information storage means stores the second two-dimensional image superimposed on the first two-dimensional image. The two-dimensional image generating means stores the second two-dimensional image data of the human head from the three-dimensional data of the human head so that the facial expression is based on the facial expression information. Generate a two-dimensional image.

[0008] Alternatively, standard model storage means for storing a standard model of a human head, deformation means for deforming the standard model based on three-dimensional measurement data obtained by photographing a human face from the front, And the three-dimensional data storage means stores the standard model deformed by the deformation means as three-dimensional data.

Alternatively, the size information and the direction information for superimposing the second two-dimensional image in one scene and the second two-dimensional image in the next scene
A comparing unit for comparing the size information and the direction information for superimposing the two-dimensional image with each other, and as a result of the comparison by the comparing unit, when it is determined that there is any change, The two-dimensional image generating means generates the second two-dimensional image of the next scene, and when it is determined that there is no change in any of the two scenes, the two-dimensional image generating means generates the second two-dimensional image. Is not performed, and the second two-dimensional image of one scene is replaced with the second 2D image of the next scene.
Used as a two-dimensional image.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS [First Embodiment] FIG. 1 is a block diagram showing an image generating apparatus 1 according to the present invention, FIG. 2 is a diagram showing examples of a standard model DS and a three-dimensional shape model DSH,
FIG. 3 is a diagram illustrating an example of the base image DS1.

In the present embodiment, for each scene such as a video game, an image using the face of the user HM as the face of the appearing character ML is generated to generate a moving image. In FIG. 1, an image generation device 1 includes a processing device 11, a standard model storage unit 12 that stores a standard model DS of a human head, a base image GR1 of each scene, and corresponding image control information NM.
, A three-dimensional measurement data storage unit 14 and the like.

The standard model DS is three-dimensional data having a standard face size and shape and structured around the entire circumference of the head. As a standard model DS, FIGS.
The standard model DSa with no expression, the standard model DSb with facial expression that utters the sound of "A", the standard model DSc with surprise expression, and the like are prepared. Besides this,
There are various expressions that express emotions such as emotions and emotions, expressions when various sounds are uttered, and various expressions when performing an operation such as lifting a luggage or exhaling.

The three-dimensional measurement data KDT is three-dimensional data of the face of the user HM composed of point groups. The three-dimensional measurement data KDT is stored in the three-dimensional measurement data storage unit 14 together with the two-dimensional image of the face of the user HM and the reliability data DR indicating the reliability of each point of the three-dimensional measurement data KDT.

In the fitting process of the standard model DS, which will be described later, 3 of the expressionless face photographed from the front is used.
The dimension measurement data KDT is used. Then, user H
Such three-dimensional measurement data KDT relating to M is prepared in advance by photographing with a three-dimensional measurement device.

The two-dimensional image of the face of the user HM and the reliability data DR are used in the fitting process of the standard model DS. As a method of obtaining the reliability DR,
For example, a method disclosed in JP-A-61-125686,
Other known methods are used.

The base image GR1 is a two-dimensional image of the whole body of the character ML as shown in FIG. Base image GR
As shown in FIG. 6, a head image GR2 of the user HM is superimposed on the head GR1a in 1 based on the image control information NM. The image control information NM includes position information NM1, size information NM2, direction information NM3, and expression information NM4 relating to the position, size, direction, and expression of the head of the character ML. The superposition process will be described later.

Returning to FIG. 1, the processing device 11 includes a CPU,
A three-dimensional model generation unit 21, a three-dimensional model storage unit 22, a scene setting unit 23, a three-dimensional model deformation unit 24, a head image Functions such as the generation unit 25 and the two-dimensional synthesis unit 26 are realized.

FIG. 4 is a flowchart for explaining the flow of processing for generating a three-dimensional shape model DSH. FIG. 5 is a diagram showing an example of a standard model DS. FIG. 6 is a diagram showing an example of a two-dimensional composite image GRZ.

The three-dimensional model generation unit 21 uses the standard model D
By fitting S to the three-dimensional measurement data KDT, a three-dimensional shape model DS of the head of the character ML is obtained.
Generate H.

As a method of fitting the standard model DS, for example, a method of performing fitting based on the reliability data DR of each point of the three-dimensional measurement data KDT is used. The processing by such a method is performed according to the flow shown in FIG.

First, rough alignment between the standard model DS and the three-dimensional measurement data KDT shown in FIG. 5 is performed (# 5).
1). That is, the standard model DS and the three-dimensional measurement data K
The direction, size, and position of the standard model DS are changed so that the distance from the DT is minimized. Generally, an expressionless state is used as the standard model DS and the three-dimensional measurement data KDT.

A contour and a feature point are extracted (# 52).
A contour and a feature point to be arranged at the same position as the contour RK and the feature point TT for the standard model DS are arranged on the three-dimensional measurement data KDT or a corresponding two-dimensional image.

As a characteristic point, for example, there is a characteristic part such as an end of an eye or a mouth, a top of a nose, a lower end of a chin, or a part having no characteristic in itself such as an intermediate part thereof. A part that is easily specified in a particular way is selected. As a contour,
A chin line, lip line, or eyelid line is selected.

In order to reduce the calculation amount and the error, the data of the three-dimensional measurement data KDT is reduced (# 5).
3). The standard model DS is modified (# 54). That is, an energy function defined in relation to the distance between each point of the three-dimensional measurement data KDT and the surface of the standard model DS,
Alternatively, the surface of the standard model DS is deformed so as to minimize them by using an energy function or the like defined to avoid excessive deformation.

Then, the target energy function and control point are changed, and the same processing for change as in step # 54 is repeated (# 55). Next, the transformation process in step # 54 will be described.

FIG. 7 is a flowchart for explaining the flow of the deformation process, FIG. 8 is a diagram schematically showing a plane S of the standard model DS and a point P of the three-dimensional measurement data KDT, and FIG. 9 is an abnormal deformation of the standard model DS. FIG. 10 is a view for explaining a virtual spring for preventing the above, and FIG. 10 is a view showing various fitting examples of the surface S with respect to the point P.

In FIG. 8, one of the points forming the three-dimensional measurement data KDT is indicated by a point Pk. On the surface S of the standard model DS, the point closest to the point Pk is indicated by Qk. The point Qk is an intersection when a perpendicular is drawn from the point Pk to the surface S.

The method for fitting the surface S to the point group is as follows. Here, first, a general fitting will be described, and then a fitting in which weighting based on reliability is performed will be described.

One point Pk in the point group, the corresponding point Qk, and the corresponding point group T = ｛(Pk, Qk), k = 1
... for n 、, fitting energy
Energy) The function Ef (U) is set as in the following equation (1).

[0030]

(Equation 1) Here, Qk (U) indicates that Qk is a function of U.

In order to prevent the surface S from being excessively deformed, a virtual spring (elastic bar) KB shown in FIG. 9 is introduced. A stabilizing energy function for stabilizing the shape of the surface S is derived based on the constraint of the virtual spring KB.

That is, FIG. 9 shows a part of a surface (curved surface) S of the standard model DS to be fitted. The surface S has a control point group U = | ui, i = 1... N |
It is formed with. A virtual spring K is provided between adjacent control points.
B is arranged. The virtual spring KB gives a constraint between the control points by a tensile force, and functions to prevent abnormal deformation of the surface S.

That is, when the interval between adjacent control points u increases, the pulling force by the virtual spring KB increases accordingly. For example, when the point Qk approaches the point Pk and the movement of the point Qk increases the interval between the control points u, the tensile force of the virtual spring KB increases. Point Q
If the interval between the control points u does not change even if k moves, that is, if the relative positional relationship between the control points u does not change, the pulling force by the virtual spring KB does not change. The value obtained by averaging the tensile force by the virtual spring KB over the entire surface S is defined as stabilizing energy. Therefore, when a part of the surface S protrudes and deforms, the stabilizing energy increases. If the entire surface S moves on average, the stabilization energy is zero.

The stabilizing energy function Es (U) is expressed by the following equation (2).

[0035]

(Equation 2) here,

[0036]

(Equation 3) Are the initial end points of the virtual spring KB and the end points of the virtual spring KB after deformation, respectively. c is a spring coefficient, and M is the number of virtual springs KB. Further, the following relationship is established.

[0037]

(Equation 4) Therefore, if the spring coefficient c is increased, the virtual spring KB
Are hard and difficult to deform.

Such a stabilizing energy function Es
By introducing (U), a certain constraint is provided on the shape change of the surface S, and excessive deformation of the surface S can be prevented.

Using the fitting energy function Ef (U) and the stabilizing energy function Es (U) described above, a fitting evaluation function E (U) is defined as in the following equation (3).

E (U) = WfEf (U) + WsEs (U) (3) where Wf and Ws are weighting coefficients for normalization, respectively. The deformation of the surface S and the search for the corresponding points are repeated so that the evaluation function E (U) of the equation (3) becomes sufficiently small.
Perform surface fitting. For example, fitting is performed in a direction in which the derivative of E (U) with respect to U approaches zero.

In the above description, the reliability of each point Pk is equal, and the evaluation function E (U) which treats the reliability of the point group equally is derived. An evaluation function E (U) weighted based on gender will be described.

That is, since the data of all the points Pk are not always reliable, when the fitting energy function shown in the equation (1) is used, the data is affected by the low reliability part. Undesirable fitting results may be introduced. However, if a point Pk with low reliability is deleted, the original shape of the point group may not be maintained at the position where the deleted point Pk is located, which is also undesirable.

Therefore, the reliability factor of each point Pk in the point group is introduced as the weight of each correspondence, and the equation (1) is changed to the following equation (1 ').

[0044]

(Equation 5) Here, ρk indicates the reliability of the point Pk. The reliability data DR described above can be used as the reliability ρk. W (ρk) is a function for obtaining a weight from the reliability ρk.

Generally, when reliability is x, W
(X) indicates that a high reliability x (small x) has a large weight, and a low reliability x (large x).
Is a function that gives a small weight to each. Such a function is given, for example, by the following equation (4).

W (x) = 1 / (1 + x) n (4) where n is a weighting factor for reliability. Also, x
> 0. For example, when the above-described number of pixels, which is the reliability data DR, is used as the reliability x, if there is no shift, x = 0 and the weight W (x) is 1.
When x = 1, the weight W (x) becomes 1 / 2n, and x =
In the case of 2, the weight W (x) is 1 / 3n. Thus, when reliability decreases (x increases), the weight W
(X) approaches zero. Therefore, the surface S is deformed according to the reliability of each point Pk.

The above equation (1 ′) is used as the fitting energy function Ef (U) in equation (3). As a result, more accurate fitting is performed for the highly reliable point Pk, and fitting that does not significantly affect the highly reliable part is performed for the less reliable point Pk.

For example, as shown in FIGS. 10C and 10D, out of the six points P1 to P6, when the reliability of the points P1 and P2 is high and the reliability of the points P3 to P6 is low, S is each point P
Deformed according to the reliability of 1 to 6. The surfaces Sc and Sd deformed in this way are shown in FIG. 10 without considering reliability.
Compared with the surface Sa shown in (A) and the surface Sb shown in FIG. 10B in which points with low reliability are deleted, the shape of the point group is reflected according to the degree of reliability, and the shape of the point group is It can be seen that the deformation is kept and the protruding deformation is suppressed.

In FIG. 7, in the transformation process, first, the point P
A point Qk corresponding to k is obtained by calculation, and a set of points Pk and Qk is created (# 61). The surface S is deformed (# 62), and the deformed evaluation function E (U) is calculated (# 63). The process is repeated until the evaluation function E (U) converges (Yes in # 64).

As a method of judging the convergence of the evaluation function E (U), a method in which the convergence is performed when the evaluation function E (U) becomes smaller than a predetermined value, and a method in which the rate of change compared with the previous calculation is a predetermined value. A known method such as a method of converging when the value becomes equal to or less than the value can be used.

By such a process, the standard model DS shown in FIG. 2A is transformed to generate a three-dimensional shape model DSH having the shape of the face of the user HM as shown in FIG. 2D. it can.

Returning to FIG. 1, the three-dimensional model storage unit 22
Stores the three-dimensional shape model DSH generated by the three-dimensional model generation unit 21. The scene setting unit 23
A base image GR1 used for one scene and image control information NM corresponding to the base image GR1 are selected from the base image storage unit 13, a three-dimensional model deformation unit 24, a head image generation unit 2
5, or position information NM1, size information NM2, direction information NM3, or expression information NM4 for processing in the two-dimensional synthesizing unit 26 is set. For example, as shown in FIG. 3, position information NM1, size information NM2, direction information NM
3 and the expression information NM4, the coordinates “(40
0, 50) ”, size“ 150 dots × 150 dots ”, corner“ (θ, φ, ψ) = (75 °, 15 °, 15
°) ", and" a facial expression that utters a sound ". Then, the base image GR1 of each scene and the corresponding image control information NM are sequentially selected, and the same processing is repeated. The “coordinates” are two-dimensional coordinates in the base image GR1 indicating the position where the head image GR2 of the user HM is superimposed. The angles (θ, φ, ψ) are Euler angles with respect to the head facing straight ahead.

The three-dimensional model transforming unit 24 includes the facial expression information NM.
The three-dimensional model generating unit 21
The three-dimensional shape model DSH generated by the above is deformed. For example, in FIG. 2, when “surprise expression” is set in the expression information NM4, the standard model DSb of “surprise expression” shown in FIG. 2E and the “non-expression” shown in FIG. Is calculated from the standard model DSa of “.”, And based on the difference, the three-dimensional shape model DSHa of “expressionless” is transformed to generate the three-dimensional shape model DSH of “surprising expression”.

That is, the amount of movement from each control point of the standard model DSb to each corresponding control point of the standard model DSa is calculated. Then, based on the calculated movement amount of each control point, the corresponding control point of the three-dimensional shape model DSHa is moved to generate the “surprising expression” three-dimensional shape model DSHc.

Returning to FIG. 1, the head image generation unit 25 rotates the three-dimensional shape model DSH so as to be in the direction indicated by the direction information NM3, and projects the three-dimensional shape model DSH onto the two-dimensional plane. A partial image GR2 is generated.

The two-dimensional synthesizing unit 26 adjusts the head image GR2 to a size based on the size information NM2,
By superimposing the head image GR2 on the position on the base image GR1 shown in FIG. 1, the two-dimensional composite image G shown in FIG.
Generate RZ. Two-dimensional composite image GRZ generated sequentially
By outputting, a moving image can be obtained.

Next, the flow of processing up to the generation of the two-dimensional composite image GRZ will be described with reference to a flowchart.
FIG. 11 is a flowchart illustrating the flow of the process of the image generation device 1.

In FIG. 11, three-dimensional measurement data KDT obtained by photographing the expressionless face of the user HM from the front is read (# 11). Model D
SHA is obtained (# 12).

In parallel with or before or after steps # 11 and # 12, position information NM1, size information NM2, direction information NM3, and expression information NM4 to be superimposed on the base image GR1 are set (# 13).

The three-dimensional shape model DSHa is deformed based on the facial expression information NM4 to adjust the facial expression (# 14), and the direction information NM3 is rotated based on the direction information NM3 to adjust the head direction (# 15).

The head image GR2 is generated by projecting the rotated three-dimensional shape model DSH onto two dimensions (# 16), and the size of the head image GR2 is adjusted based on the size information NM2 (# 17). .

Then, the head image GR2 is superimposed on the position on the base image GR1 indicated by the position information NM1, to obtain a two-dimensional composite image GRZ (# 18). If there is a next scene (Yes in # 19), the process returns to step # 11 to repeat the two-dimensional composite image GRZ of the next scene.

According to the present embodiment, the face direction, size,
Alternatively, it is possible to generate the head image GR2 of the user HM according to the facial expression or the like, and easily generate the two-dimensional composite image GRZ of the character ML. Even if the direction, size, expression, etc. of the face are changed, it is not necessary to newly photograph the user HM. Therefore, even when a plurality of two-dimensional composite images GRZ are generated, the burden on the user is reduced. Can be.
[Second Embodiment] FIG. 12 is a block diagram showing an image generating apparatus 3 according to a second embodiment. FIG. 13 is a diagram illustrating an example of a base image GR1 of a continuous scene, and FIG. 14 is a diagram illustrating an example of a moving image MV of a certain continuous scene.

In the first embodiment, the head image GR2 is generated for each scene, and the two-dimensional composite image GRZ is generated.
In the present embodiment, the image control information NM of the preceding and following scenes is compared, a head image GR2 is generated as necessary, and a two-dimensional composite image GRZ is generated. Then, a moving image of a continuous scene is generated.

In FIG. 12, the image generating device 3 includes a processing device 31, a standard model storage 32, and a base image storage 3
3 and a three-dimensional measurement data storage unit 34.
Comparing FIG. 12 with FIG. 1, it can be seen that the configuration of the image generation device 3 is almost the same as the configuration of the image generation device 1. Therefore, a portion different from the function of the image generation device 1 will be described below.

The standard model storage unit 32 and the three-dimensional measurement data storage unit 34 have the same functions as the standard model storage unit 12 and the three-dimensional measurement data storage unit 14, respectively.
The base image storage unit 33 stores the base image GR1 of each scene and the corresponding image control information NM, similarly to the base image storage unit 13. However, position information NM1, size information NM2, direction information NM3, or expression information NM4 relating to a scene between a certain scene and another scene.
Is represented by a predetermined function or the like.

For example, the base image GR11 shown in FIG.
And the base image GR1n are the first and last base images GR1 of a certain continuous scene, respectively. Between the base image GR11 and the base image GR1n, 1
One or a plurality of base images GR1k are stored. Coordinates “(400, 50)”, size “150 dots × 150 dots”, and corners “(θ, φ, ψ) as position information NM1 to expression information NM4 corresponding to base image GR11.
= (75 °, 15 °, 15 °) ”and“ expression expressing a sound ”are stored, and coordinates“ (45) are used as position information NM1 to expression information NM4 corresponding to the base image GR1n.
0, 150) ”, size“ 250 dots × 250 dots ”, corner“ (θ, φ, ψ) = (60 °, 5 °, 0
°) "and" surprising expression ". Then, the position information NM1 to NM1 corresponding to each base image GR1k
A predetermined function is stored as the facial expression information NM4. The function corresponds to each base image G corresponding to a change from the scene of the base image GR11 to the scene of the base image GR1n.
The head image GR2 superimposed on R1k is determined to change as desired.

Returning to FIG. 12, the processing device 31 includes a three-dimensional model generation part 41, a three-dimensional model storage part 42, a scene setting part 43, a three-dimensional model deformation part 44, and a head image generation part 4.
5. The functions of the two-dimensional synthesizing unit 46, the scene comparing unit 47, and the moving image generating unit 48 are realized.

The three-dimensional model generation unit 41 and the three-dimensional model storage unit 42 perform the same processing as the three-dimensional model generation unit 21 and the three-dimensional model storage unit 22, respectively. The scene setting unit 43 includes a base image GR used for each scene.
1 and the corresponding image control information NM are selected from the base image storage unit 33, and the position information NM1, the size information N
M2, direction information NM3, or expression information NM4 is set. In the example illustrated in FIG. 13, in the base image GR11 and the base image GR1n, position information NM1 to expression information NM4 are set based on the corresponding image control information NM. Each base image GR1k between them is set based on the above-described predetermined function and the image control information NM corresponding to the base image GR11 and the base image GR1n.

The scene comparison section 47 compares the position information NM1, the size information NM2, the direction information NM3, or the expression information NM4 of the adjacent base images GR1. The three-dimensional model deformation unit 44 shown in FIG.
The same processing as in step 4 is performed. However, when it is determined that the expression information NM4 of the processing target scene has the same value as the expression information NM4 of the previous scene as a result of the comparison by the scene comparison unit 47, the three-dimensional shape model DSH is not modified. In this case, in the subsequent processing, the three-dimensional shape model DSH of the previous scene is used as it is.

The head image generator 45 performs the same processing as the head image generator 25 shown in FIG. However, when the three-dimensional shape model DSH acquired from the three-dimensional model deformation unit 44 is the same as that of the previous scene and the direction information NM3 has the same value, the generation of the head image GR2 is not performed. In this case, in the subsequent processing, the three-dimensional shape model DSH of the previous scene is used as it is.

The two-dimensional synthesis unit 46 performs the same processing as the two-dimensional synthesis unit 26, and performs a two-dimensional synthesis image G for each scene.
Generate RZ. However, when the three-dimensional shape model DSH acquired from the head image generation unit 45 is the same as that of the previous scene and the size information NM2 has the same value, the head image GR2 generated for the previous scene is used. To generate a two-dimensional composite image GRZ.

The moving image generator 48 controls the output timing of the generated two-dimensional composite image GRZ, and generates a moving image MV as shown in FIG. According to the present embodiment, using the three-dimensional measurement data KDT of one user HM, a character M having various face orientations, face sizes, or facial expressions is used.
The L moving image MV can be easily generated.

Further, in a predetermined case, the same 3
Since the dimensional shape data KDT or the two-dimensional composite image GRZ can be used, the process of generating and storing the head image GR2 for some scenes can be omitted.
Therefore, the overall processing speed can be increased, and the storage capacity can be reduced.

In the two embodiments described above, the shape and the like of the human head are deformed using a standard model.
Although the two-dimensional image has been generated, a two-dimensional image may be generated for other various objects by the same method.

As a method of fitting the standard model to the three-dimensional measurement data, a method of locally fitting only a finely shaped portion after the entire fitting may be used. Further, a method other than the method described above may be used.

The configuration, processing contents, processing order, and the like of the entire image generating device 1 or the image generating device 3 can be changed as appropriate in accordance with the gist of the present invention.

[0078]

According to the present invention, the orientation of a character's face,
It is possible to easily generate a two-dimensional image or a moving image according to the size or the expression.

[Brief description of the drawings]

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

FIG. 2 is a diagram illustrating an example of a standard model and a three-dimensional shape model.

FIG. 3 is a diagram illustrating an example of a base image.

FIG. 4 is a flowchart illustrating a flow of a process of generating a three-dimensional shape model.

FIG. 5 is a diagram showing an example of a standard model.

FIG. 6 is a diagram illustrating an example of a two-dimensional composite image.

FIG. 7 is a flowchart illustrating a flow of a deformation process.

FIG. 8 is a diagram schematically showing a plane S of a standard model and a point P of three-dimensional measurement data.

FIG. 9 is a diagram for explaining a virtual spring for preventing abnormal deformation of the standard model.

FIG. 10 is a diagram showing examples of various fittings of a surface S to a point P.

FIG. 11 is a flowchart illustrating a flow of a process performed by the image generation apparatus.

FIG. 12 is a block diagram illustrating a moving image generation device according to a second embodiment.

FIG. 13 is a diagram illustrating an example of a base image of a continuous scene.

FIG. 14 is a diagram illustrating an example of a moving image of a certain continuous scene.

[Explanation of symbols]

1, 3 image generation device 12 standard model storage unit (standard model storage unit) 13, 33 base image storage unit (information storage unit) 21 three-dimensional model generation unit (deformation unit) 22, 42 three-dimensional model storage unit (three-dimensional Data storage unit) 23, 43 Scene setting unit (scene selection unit) 25, 45 Head image generation unit (2D image generation unit) 26 2D synthesis unit (overlaying unit) 46 2D synthesis unit (synthesis unit) 47 Scene comparison unit (comparison means) DS standard model (three-dimensional data) GR1 base image (first two-dimensional image) GR2 head image (second two-dimensional image) GRZ two-dimensional composite image NM1 position information NM2 size information ( Size information) NM3 Direction information NM4 Expression information

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Shigeaki Tochimoto 2-3-13-1 Azuchicho, Chuo-ku, Osaka-shi, Osaka Inside Osaka International Building Minolta Co., Ltd. 2-3-1-3 Osaka International Building Minolta Co., Ltd. F-term (reference) 2C001 BA00 BA06 BC00 BC05 BC10 CB01 CC02 CC08 5B050 AA08 BA08 BA11 BA12 CA07 EA05 EA06 EA12 EA13 EA19 EA24 EA27 FA02 FA10 5B057 AA20 CA12 CB12 CA16 CB CC03 CD11 CD14 CE08 DA07 DA16 DB03 DC05 DC16 5C082 AA06 CA32 CA42 CA52 CA56 CB05 DA86 MM02 MM08 MM09

Claims (4)

[Claims] 1. An image generating apparatus for generating a two-dimensional composite image by synthesizing a plurality of two-dimensional images, comprising: superimposing means for superimposing a second two-dimensional image on a first two-dimensional image; A three-dimensional data storage unit that stores three-dimensional data of the object; a two-dimensional image generation unit that generates a second two-dimensional image from the three-dimensional data; and the second one that is superimposed on the first two-dimensional image. Information storage means for storing, for each scene, position information, size information, and direction information of the object relating to the two two-dimensional images; and scene selection means for sequentially selecting the scenes. A generating unit that generates the second two-dimensional image based on the size information and the direction information corresponding to a currently selected scene; and the superimposing unit generates the second two-dimensional image at a position based on the position information. Two two An image generating apparatus, comprising: superimposing a two-dimensional image. 2. The three-dimensional data storage means stores three-dimensional data of a human head, and the information storage means stores the second two-dimensional image superimposed on the first two-dimensional image. The facial expression information of the human face is stored for each scene, and the two-dimensional image generating means converts the two-dimensional image from the three-dimensional data of the human head into a facial expression based on the facial expression information. The image generation device according to claim 1, wherein the image generation device generates an image. 3. A standard model storing means for storing a standard model of a human head; a deforming means for deforming the standard model based on three-dimensional measurement data obtained by photographing a human face from the front; The image generation device according to claim 2, further comprising: the three-dimensional data storage unit stores the standard model deformed by the deformation unit as three-dimensional data. 4. The size information and the direction information for superimposing the second two-dimensional image in one scene and the size information for superimposing the second two-dimensional image in the next scene. And the direction information is compared with each other. If it is determined that there is a change in any of the results of the comparison by the comparing means, the two-dimensional image generating means The second two-dimensional image is generated. If it is determined that there is no change in any of the two-dimensional images, the second two-dimensional image is not generated by the two-dimensional image generation unit, and the one- The image generation device according to any one of claims 1 to 3, wherein a second two-dimensional image is used as the second two-dimensional image of a next scene.