JPH1074271A - Method and device for preparing three-dimensional portrait - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically prepare a portrait observed from an optional angle by using the portraits of manually drawn front and side faces and collecting more features of each person. SOLUTION: Portraits 100A, 100B corresponding to front and side faces are inputted to a personal computer and stored as respective portrait data 104A, 104B. Then the data 104A, 104B are displayed on a portrait display device 103 and the feature points of the likeness pictures such as outlines are inputted from an instruction input device 110. Then portrait feature parameters are extracted based on differences of feature data between the feature point data of the portraits and three-dimensional (3D) reference face data 105 and the data 105 are converted into 3D portraits data 106. Finally the data 104A, 104B are mapped on the data 106 as textures and deformed in optional directions by a portrait generating program 109 and the generated portrait is outputted on the display device 103.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for creating a portrait related to each individual using a computer.

[0002]

2. Description of the Related Art Heretofore, portraits are basically drawn in two dimensions, and have been created by drawing them individually, such as a frontal face, a side face, and a face viewed obliquely. The portraits thus created are basically independent two-dimensional images.

On the other hand, Murakami et al. Proposed an existing technique for generating three-dimensional portraits, in which three-dimensional face data of an individual is used.
A feature that derives features from differences from the dimensional average face data and creates a three-dimensional relief using the median splitting method (Murakami, Kondo,
Kosui, Suzuki, "A Method for Generating 3D Caricatures",
IEICE Technical Report, Vol. 94, no. 51, (PRU946
−15), pp. 63-70, 1994) or a method of inputting a two-dimensional face photograph proposed by Shimada et al. And generating a three-dimensional portrait based on a feature vector obtained from a difference from the two-dimensional average face data (Shimada, Shiono, "A method of caricature generation based on a three-dimensional face model", IEICE D-2, Vol.
No. 76-D-2, no. 12, pp. 2513-252
1, (1993-3)).

[0004]

However, in a conventional caricature generating method which is created by drawing each of the face, such as a frontal face, a side face, and a face viewed obliquely, for example, different directions (directions 1 and 2) are used. In the case where there are two portraits corresponding to the face viewed from the direction 2), in order to create a large number of portraits viewed from the direction between the direction 1 and the direction 2, it is necessary to manually add all of them. There is a problem that it requires a great deal of labor to create a portrait.

[0005] Further, in the above-described existing technology for generating a three-dimensional portrait, there is a problem in that there are few aspects because the input of the parameter extraction of the exaggerated expression is limited to only a photograph.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems. It is possible to automatically create a large number of portraits viewed from an arbitrary direction without manual labor, and to make each person's characteristics unique. It is an object of the present invention to provide a three-dimensional portrait creation method and apparatus capable of generating a portrait with a variety of variations in which the portrait is captured.

[0007]

In order to achieve the above object, a three-dimensional portrait creation method according to the present invention comprises the steps of providing a portrait of a front face and a side face drawn from a front face and a side face of each individual. Extracting a face feature parameter based on a difference between feature point data input from the portraits of the front face and side face and feature data obtained from a pre-stored three-dimensional standard face model; Deforming the three-dimensional standard face model using parameters to finally create a three-dimensional face model of each individual.

In the above-mentioned three-dimensional portrait creation method, a portrait of a frontal face and a portrait of a side face are mapped as textures to a three-dimensional portrait model deformed from a three-dimensional standard model using the facial feature parameters. It is characterized by having a process of creating a portrait.

Further, the present invention provides a three-dimensional portrait creation apparatus comprising: an image input device for inputting a portrait of a front face and a side face drawn from a front face and a side face of each individual; An instruction input device for instructing a feature point from face portrait data; and a difference between feature point data of the designated front face and side face portrait and feature data obtained from a previously stored three-dimensional standard face model. A computer that extracts a facial feature parameter based on the extracted facial feature parameter, transforms the three-dimensional standard face model of the three-dimensional grid data by using the extracted facial feature parameter, and creates a three-dimensional facial model, It is characterized by having.

Further, in the above-mentioned three-dimensional portrait creation apparatus, the computer further adds a portrait of a front face and a portrait of a side face to a three-dimensional portrait model deformed from a three-dimensional standard model using the facial feature parameters. A similar face is created by mapping as a texture.

According to the present invention, in order to be able to project and handle three-dimensional information of the front and side faces, which are originally two-dimensional images, a three-dimensional standard face model is previously stored in processing means such as a computer. The three-dimensional face model is prepared in a processing means such as the computer by deforming this based on the face feature parameters obtained from two portraits of a front face and a side face written by hand. After that, a facial portrait viewed from an arbitrary direction is automatically created by image data processing using a computer or the like. As described above, by using a facial feature parameter obtained by processing a hand-drawn portrait based on a photograph or the like of each individual as an extended parameter of the three-dimensional standard model, each individual is more than ever before. It is possible to generate a portrait that captures the features of

[0012]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a view for explaining an embodiment of the present invention, wherein 100A and 100B are two-dimensional portraits of a front face and a side face, respectively, 101 is an image input device, 1
02 is a personal computer (PC), 103 is an image display device, 104A and 104B are two-dimensional portrait data of the input front and side faces, 105 is three-dimensional standard face data, 106 is three-dimensional portrait data, and 107 is Feature point designation program 108 is feature point designation program 1
07 and the three-dimensional standard face data 10
A conversion program for extracting a facial feature parameter based on a difference from the feature data of No. 5 and transforming the three-dimensional standard face data 105 into three-dimensional facial data 106 by transforming the standard facial data 105 using the facial feature parameter , 10
Reference numeral 9 denotes a portrait generating program for mapping the front and side portrait data 104A and 104B to the three-dimensional portrait data 106 obtained above to generate a portrait viewed from a designated direction, and 110 denotes an instruction input device such as a mouse. ,.

In order to operate this, first, according to the procedure shown in the flowchart of FIG.
5 is stored in the personal computer 102.

Next, for example, a two-dimensional portrait 100A of a front face and a side face drawn so as to correspond one-to-one with an image of a front face and a side face taken of each individual's face.
And 100B to the personal computer 1 using the image input device 101.
02, and stored as two-dimensional portrait data 104A and 104B of the front face and the side face.

Next, the input and stored two-dimensional portrait data 104A and 104B of the front face and the side face are displayed on the image display device 103, and under the control of the feature point designation program 107, a mouse or the like is used. Instruction input device 11
Using 0, the feature points of the portrait, such as the contour, eyes, and nose of the face, are input.

Further, the conversion program 108 extracts a facial feature parameter based on a difference between the feature point data of the portrait obtained above and the feature data of the three-dimensional standard face data 105, and uses the facial feature parameter. The three-dimensional standard face data 105 is converted into three-dimensional face data 106.

Finally, a two-dimensional portrait data 104 of a front face and a side face is added to the three-dimensional portrait data 106 obtained above, for example, by the portrait generation program 109.
A and 104B are transformed in an arbitrary direction by mapping them as textures, and output to the image display device 103.

As described above, a portrait viewed from an arbitrary direction can be automatically generated. At this time, the user may use a command input device 110 such as a mouse to input a direction, or it is also possible to automatically generate a portrait image in which the direction is changed little by little automatically. .

Further, by producing a character whose face is a portrait created in this way, the character can be used as an agent in a virtual space.

FIG. 3 shows a method of obtaining a facial feature parameter according to the present invention.

In the figure, 201A and 201B are portrait data obtained by inputting a portrait of a front face and a side face drawn by hand as electronic data, and 202A and 202B are three-dimensional standard data required for extracting feature points. The two-dimensional mesh (lattice) data 203 for the front face and the side face corresponding to the face model, 203 is the portrait data 201A of the front face, the two-dimensional grid data 202A corresponding to the three-dimensional standard face of the front face, and the portrait 201B of the side face. 3D standard face support for face and side face 2
An image display device 204A and 204B for displaying the two-dimensional lattice data 202B in a superimposed manner is an image display device 203.
(X, y1) and (y2, z) extracted from the front face and side face images displayed in FIG.
Is a conversion program for finally changing the motion vectors (x, y1) and (x, y2) obtained from 204A and 204B into the motion vectors (x, y, z) of each vertex.
Reference numeral 6 denotes a movement vector (x, y, z) of each vertex for converting the three-dimensional standard face model into a three-dimensional portrait model;
Reference numeral 7 denotes an instruction input device necessary for designating a feature point from an image displayed on the image display device 203.

First, using a not-shown image input device, a manual portrait is converted into digital portrait data 201A and 201B of front and side faces. On the other hand, two-dimensional grid data 202A and 202B of a front face and a side face corresponding to a three-dimensional standard face model necessary for inputting a feature point are prepared.

Next, the face data 201A of the front face and the grid data 202 corresponding to the standard face are input to input the feature points.
A, The portrait data 201B of the side face and the grid data 202B corresponding to the standard face are superimposed and displayed on the image display device 203.

Using an instruction input device 207 such as a mouse, feature points corresponding to the front face and the side face are input. The input feature points are compared with the feature points of the standard face model, and based on the difference, the movement vectors (x, y1) (y2, z) of the feature points in the front and side directions are input. The front movement vector and the side movement vector are mixed to finally determine a movement vector (x, y, z) for each vertex for developing a three-dimensional standard face model.

FIG. 4 is a diagram for explaining the difference between the three-dimensional portrait creation method of the present invention and the conventional method.

In the figure, reference numerals 301A and 301B denote face data obtained by digitizing photographs of front and side faces using a scanner or the like, and 302A and 302B depict portraits of front and side faces drawn by hand. The input portrait data 303 is a feature parameter obtained from the face data 301A and 301B using the image processing technique used in the conventional method, and the like 304 is obtained from the portrait data 302A and 302B by the method of the present invention. The personalized facial feature parameter 305 of each individual is a completed three-dimensional facial portrait created by deforming the three-dimensional standard face data using the parameters.

In the conventional method, the feature parameters 30 obtained by performing image processing on the photograph face data of 301A and 301B.
3 is used to transform the three-dimensional standard model to complete a three-dimensional portrait.

In the method of the present invention, as shown in the range of the one-dot chain line, the facial feature parameter 3 obtained from the facial sketch data 302A and 302B drawn and input by hand.
Finally, the three-dimensional face model is completed by deforming the three-dimensional standard model using 04.

[0030]

As described above, the present invention has an advantage that any number of portraits viewed from an arbitrary direction can be automatically created. In addition, since a face is obtained by transforming the three-dimensional standard model using a face feature parameter obtained from portrait data drawn by hand, a face that captures the characteristics of each individual can be generated more than ever. There is an advantage.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a three-dimensional portrait creation system showing an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a three-dimensional portrait creation method according to an embodiment of the present invention.

FIG. 3 is a diagram illustrating a facial feature parameter extraction method according to an embodiment of the present invention.

FIG. 4 is a diagram for explaining a difference between the three-dimensional portrait creation method of the present invention and a conventional method, and includes a drawing substitute photograph.

[Explanation of symbols]

100A, 100B: Two-dimensional portraits of front and side faces 101: Image input device 102: Personal computer 103: Image display device 104A, 104B: Two-dimensional portrait data of input front and side faces 105: Three-dimensional standard Face data 106 ... Three-dimensional portrait data 107 ... Feature point designation program 108 ... Conversion program 109 ... Face portrait generation program 110 ... Instruction input device 201A, 201B ... Face portrait data of front and side faces 202A, 202B ... Front and side face two-dimensional grid data 203: Image display devices 204A, 204B: Movement vectors obtained from frontal and side faces 205: Conversion program 206: Movement vector of each vertex 207: Instruction input device 301A, 301B: Front Face data of face and side face 3 02A, 302B: portrait data corresponding to the front face and side face 303: feature parameters obtained from face data by existing image processing technology 304: portrait feature parameters newly used in the present invention 305: completion of a three-dimensional portrait Figure

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Katsunori Shimohara 3-19-2 Nishishinjuku, Shinjuku-ku, Tokyo Nippon Telegraph and Telephone Corporation

Claims (4)

[Claims] 1. A process of giving a portrait of a front face and a side face drawn from a front face and a side face of each individual, and feature point data input from the portraits of the front face and the side face are stored in advance. Extracting a facial feature parameter based on a difference from feature data obtained from the three-dimensional standard face model, and finally transforming the three-dimensional standard face model using the facial feature parameter to obtain a 3 Creating a three-dimensional face model; and a three-dimensional face creation method. 2. A process of creating a portrait by mapping a portrait of a frontal face and a portrait of a side face as a texture to a three-dimensional facial model deformed from a three-dimensional standard model using the facial facial feature parameter as a texture. The three-dimensional portrait creation method according to claim 1, wherein: 3. An image input device for inputting a portrait of a front face and a side face drawn from a front face and a side face of each individual, and instructing a feature point from the input portrait data of the front face and the side face. An instruction input device; extracting a facial feature parameter based on a difference between the feature point data of the portrait of the designated front face and side face and feature data obtained from a previously stored three-dimensional standard face model; A computer that transforms the three-dimensional standard face model of the three-dimensional grid data by using the extracted facial feature parameters to transform the three-dimensional standard face model to create a three-dimensional facial model; and a three-dimensional facial model, comprising: Creating device. 4. The computer creates a portrait by mapping a portrait of a front face and a portrait of a side face as a texture to a three-dimensional face model deformed from a three-dimensional standard model using the face feature parameters. The three-dimensional portrait creation device according to claim 3, wherein