JP2004288082A - Portrait creation method, portrait creation device, as well as program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a portrait creation method, device and program which can easily create a suitable portrait reflecting features of a model even with few face parts prepared. <P>SOLUTION: For each face image, using multiple pieces of face image data of the same figure, the amounts of features of face parts are extracted for each face image data, and the parts to be used are selected from the parts group of the face parts for portraits set in advance using the amount of the features of the face parts. Then, the same face parts are composed. The portrait of the figure is then created using the parts obtained by such composition, thereby solving the problem. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention belongs to the technical field of caricature creation, and more specifically, a caricature creation method and a caricature creation apparatus capable of creating a suitable caricature that appropriately captures the characteristics of a model person, and a program for realizing the same About.
[0002]
[Prior art]
The creation of a portrait using a computer or the like is usually based on various facial parts such as (facial) contours, eyebrows, eyes, nose, mouth, ears, hairstyle, etc. This is done by preparing parts and selecting parts similar to the facial parts of the person (hereinafter referred to as a model) from whom a portrait is created, and combining them appropriately.
However, this method requires a certain amount of technology and takes time and effort to create a portrait that appropriately captures the features of the model.
[0003]
On the other hand, a technique for automatically creating a portrait is also known. For example, in Patent Document 1, a face image of a model is input and image analysis is performed to extract the feature amount of the shape and position of each face component. Using this feature amount, a face component prepared in advance is extracted. A method of creating a portrait by selecting a part from a group of parts (face part data group), determining an arrangement position of each part, and changing the arrangement position as necessary is disclosed.
According to this caricature creation method, a caricature can be automatically created simply by inputting an image and changing the arrangement position of the facial parts as necessary.
[0004]
[Patent Document 1]
Japanese Patent Laid-Open No. 10-225017
[Problems to be solved by the invention]
However, it is rare that a single image appropriately represents the characteristics of a person, that is, it is difficult to properly extract the characteristics of all the facial parts of a model from a single image. .
In addition, since only parts of facial parts prepared in advance, that is, predetermined parts can be used, it is often impossible to use facial parts that appropriately match the features of the model. On the other hand, in order to cope with a large number of persons suitably, a very large number of parts are required, and it takes time to select a part using the feature amount.
[0006]
An object of the present invention is to solve the above-mentioned problems of the prior art, and it is possible to automatically create a portrait by a simple operation, and even if the number of facial parts to be prepared is small, the model It is an object to provide a portrait creation method capable of creating a suitable portrait that appropriately captures the characteristics of the above, a portrait creation device that implements the portrait creation method, and a program for realizing these.
[0007]
[Means for Solving the Problems]
In order to achieve the above object, the portrait creation method of the present invention uses a plurality of face image data of the same person, extracts feature amounts of each face part for each face image data, and features the face parts. Using the amount, the part to be used is selected for each facial image from the parts group of facial parts for caricature set in advance, and the parts of the same facial parts are synthesized and obtained by synthesis There is provided a portrait creation method characterized by creating a portrait of the person using parts.
[0008]
Further, the caricature creation device of the present invention includes a storage unit that stores a plurality of parts for caricatures for a plurality of facial parts, an acquisition unit for face image data, and a plurality of face image data of the same person acquired by the acquisition unit. For each face image data, the extraction means for extracting the feature amount of each face part and the parts used from the storage means for each face image using the feature amount of the face part extracted by the extraction means Selecting means for selecting the same, combining means for combining the parts of the same facial part selected by the selecting means, and portrait generating means for generating a portrait of the person using the parts combined by the combining means A caricature creation apparatus characterized by the above is provided.
[0009]
Further, the program of the present invention uses a plurality of face image data of the same person to extract feature amounts of each face part for each face image data, and stores a plurality of parts for a portrait for the plurality of face parts. Selecting a part to be used from the storage means using the feature amount of the facial part for each facial image, synthesizing the selected parts for the same facial part, and obtained by the synthesis There is provided a program for executing a step of creating a portrait of the person using the parts.
[0010]
In such a portrait creation method, a portrait creation apparatus, and a program according to the present invention, the composition is performed by performing weighting according to the similarity between the feature amount of the face part and the feature amount of the selected part. preferable.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the portrait creation method, the portrait creation apparatus, and the program of the present invention will be described in detail based on preferred embodiments shown in the accompanying drawings.
[0012]
FIG. 1 shows a conceptual diagram of an example of a portrait creation system according to the portrait creation apparatus of the present invention that implements the portrait creation method of the present invention.
The caricature creation system 10 of the illustrated example creates a caricature using the input face image (face image data). Basically, the input unit 12, the system main body 14, the display 16, and the printer 18 are used. And is configured.
[0013]
In the caricature creation system 10 of the illustrated example, the input means 12 acquires image data of a face image of a person (hereinafter referred to as a model) for which a portrait is created (inputs the face image), and supplies it to the system body 14. It is a part.
The input unit 12 is not particularly limited, and various types can be used as long as image data can be acquired and supplied to the system main body 14. As an example, a reading device for a recording medium such as a smart media ^TM or a CD-R, an image reading device from an imaging medium such as a digital camera, an image reading device for photoelectrically reading a reflection original such as a scanner, and an image (image data) Examples of connection means with a network to communicate are exemplified. A plurality of types may be connected to the system main body 14.
[0014]
In the present invention, the face image may be an image in which the face of a person (model) is present, but it is needless to say that an image of the proportion of the model face such as a face photograph or upper body portrait is preferable. As will be described in detail later, the portrait creation method of the present invention creates one portrait using a plurality of different facial images of the same model. For this reason, it is preferable that the face image to be used has a similar face photographing (observation) angle, model expression, face size, and the like.
In the present invention, there are no particular limitations on the original face image or manuscript, and there are various types of images such as photographs, prints output by various printers, printed materials, images taken with an imaging medium such as a digital camera, and the like. Is available. Further, for example, a face image may be input from different image sources such as a photograph and a CD-R.
[0015]
The system main body 14 (hereinafter referred to as the main body 14) creates a portrait of the model using the image data of the face image supplied from the input means 12, and outputs it to the printer 18 or the like. Is configured to include an extraction unit 30, a part storage unit 32, a part selection unit 34, a part synthesis unit 36, and a portrait creation unit 38. Although not shown, the main body 14 is connected to a mouse and a keyboard for performing various operations.
What is necessary is just to comprise such a main body using a personal computer (PC) or a workstation (WS), for example.
[0016]
The extraction unit 30 analyzes the image data of the face image supplied from the input unit 12 to perform face extraction, and further extracts each face part, extracts feature values of each face part, and positions of each face part Is detected.
Facial parts are each part which comprises a person's face, for example, eyebrows, eyes, nose, mouth, ear, (face) outline, hair, and the like. The feature amount of the facial part is exemplified by information on the size, shape, arrangement, etc. of each facial part, such as the thickness, angle, spacing, and eye spacing of eyebrows.
[0017]
Such face extraction, face part extraction, face part position detection, face part feature quantity extraction are main part extraction, color extraction such as skin color, contour detection, edge detection, extraction using pixel continuity , Pattern matching, template matching, shape recognition, etc. may be used.
[0018]
The present invention creates a single portrait using a plurality of different face images of the same model, and the extraction means 30 extracts each face part and feature quantity for all input face images (face image data). Perform extraction.
Correspondingly, the extraction unit 30 may perform various extractions and the like after electronic scaling of the image and matching the face size as necessary.
[0019]
The parts storage unit 32 (hereinafter referred to as storage unit 32) is a part that stores parts of facial parts for creating a portrait. A part is a two-dimensional picture obtained by deforming the face part to create a portrait. Depending on the face part such as the contour, eyes, and nose, the contour part, the eye part, the nose part, etc. Are created for each facial part, and the storage unit 32 stores them as vector information, for example.
The caricature creation system 10 detects the feature amount of each part created in advance and / or creates each part according to the feature amount set as appropriate, and the storage unit 32 stores each part and its part. The feature amount is stored in association with each other.
[0020]
The part selection unit 34 (hereinafter, referred to as selection unit 34) uses the feature amount of each face part of each face image supplied from the extraction unit 30, and uses the feature amount from the storage unit 32 for each face part. This is a part for selecting similar parts.
As described above, in the present invention, a portrait is created using a plurality of face images. Accordingly, the number of parts selected by the selection unit 34 for a kind of face part is the number of face images input. For example, when three model face images are input, three parts such as contour parts, eye parts, and mouth parts are also selected.
[0021]
The parts synthesizing means 36 (hereinafter, referred to as synthesizing means 36) synthesizes the parts selected by the selecting means 34 for each type of facial part, and forms one part used for creating a portrait. As described above, when there are three input face images, the synthesizing unit 36 synthesizes three parts with one kind of each face part to make one part used for creating a portrait.
[0022]
The synthesis of the parts may be performed by a known method such as a method using morphing. Here, in order to create a portrait that better captures the features of the model, depending on the similarity of the feature quantity with the corresponding facial part for each part, so that parts with similar similarity of the feature quantity are emphasized It is preferable to perform weighting and synthesize the parts for each face part.
Further, the synthesis unit 36 includes a part having a large difference in feature amount from other parts of the same face part (for example, a part in which the difference in feature quantity from the face part exceeds a threshold value determined as appropriate). In this case, the parts may be synthesized for each facial part except for this part.
[0023]
The portrait creation means 38 creates a portrait using the parts synthesized by the synthesis means 36 and the position information of each facial part extracted by the extraction means 30 and outputs the portrait to the display 16 or the printer 18.
The creation of the portrait may be performed by using various methods known in the art using various portrait creation devices, portrait creation software, etc. using the parts for the portrait according to the position information of the facial parts. In addition, the position of each part may be determined according to one appropriately selected face image, or the positions of the face parts in a plurality of face images may be averaged and determined accordingly.
[0024]
The display 16 is a known display used for a PC or the like. For example, a CRT (Cathode Ray Tube) or an LCD (Liquid Crystal Display) may be used. In the caricature creation system 10, various operations are performed by, for example, display on the display 16 and GUI (Graphical User Interface) using the mouse or the like.
The printer 18 is also a known printer such as an ink jet printer, an electrophotographic printer, or a printer that forms a latent image by imagewise exposing a photosensitive material to perform a wet development process.
[0025]
Hereinafter, the present invention will be described in more detail by explaining the operation of the illustrated portrait creation system 10. The program of the present invention is a program for implementing the following operations and functions.
[0026]
When creating a portrait in the portrait creation system 10, first, the input unit 12 acquires model face image data by photoelectrically reading a photograph or reading a CD-R on which image data is recorded. This is sent to the extraction means 30.
As described above, in the portrait creation system 10 that performs the portrait creation method of the present invention, the creation of the portrait is performed using a plurality of different facial images of the same model, so the input means is used to create the portrait. A number of face image data are acquired and sent to the extraction means 30. In this example, it is assumed that three face images are input as an example.
The number of face images to be input may be specified, for example, by using a keyboard or the like prior to inputting the face images. Alternatively, when all the face images have been input, an instruction to end the input is given. May be issued.
[0027]
The extraction means 30 performs image analysis on each of the three face image data supplied as described above, and extracts the face part, the position of the face part, and the feature amount of the face part for each face image. The feature amount of each face part of each face image is sent to the selection means 34 and the position information of the face part is sent to the portrait creation means 38, respectively.
The selection unit 34 supplied with the feature amount of the face part searches the storage unit 32 and selects a part having a feature amount most similar to the feature amount supplied from the extraction unit 30 for each face part. Read out and send to the combining means 36. In this example, since a portrait is created from three face images, the selection means 34 selects three parts for one type of face part.
[0028]
The synthesizing unit 36 synthesizes the parts supplied from the selecting unit 34 for each type of facial part, and sends it to the portrait creating unit 38 as one part used for creating the portrait. In this example, a portrait is created from three face images, so the composing means 36 composes three parts for one type of face part and creates one part for creating a portrait.
[0029]
The caricature creating means 38 creates a caricature of the model to which the face image is input, using each part for caricature creation supplied from the synthesizing means 36 and position information of each facial part supplied from the extracting means 30. The caricature (caricature image data) is output to the printer 18. Upon receiving the portrait, the printer 18 outputs a hard copy that records the portrait.
Before outputting the portrait on the printer 18, the created portrait may be displayed on the display 16 so that the operator can correct the portrait if necessary.
[0030]
As is clear from the above description, the present invention uses a plurality of different face images (face image data) of the same model, extracts feature amounts for each face image, selects parts of each face part, Create a portrait by combining multiple parts for each facial part type. Therefore, the reliability of the extracted feature amount can be improved, and a portrait that more appropriately captures the original feature of the model can be created.
Moreover, since the parts corresponding to a plurality of face images are synthesized, preferably weighted according to the similarity of the feature amount, the parts of the model can be obtained even if the number of parts prepared for each face part is small. By using parts that are close to the original features, it is possible to create a portrait that properly captures the features of the model.
[0031]
In the above example, the part is image data of a two-dimensional image. However, the present invention is not limited to this, and the part is made of three-dimensional image data to create a three-dimensional portrait. Also good.
In order to create a portrait similar to the original face using the parts of the three-dimensional data, it is conceivable that the amount of information of the parts and the like is increased. However, according to the present invention for synthesizing a plurality of parts corresponding to a plurality of face images, a caricature that captures the features of the model can be created even with a small number of prepared parts. Even in the creation of the information, the increase in the total amount of information can be greatly reduced.
[0032]
Further, in the present invention, after extracting the feature amount from the face image as described above, after selecting the part having the closest feature amount, the part is deformed so that the feature amount matches (or approaches) the face image. May be. In this case, a portrait may be created without synthesizing the parts.
[0033]
The caricature creation system 10 using the present invention may output the created caricature (its image data) to a location other than the printer 18 and the display 16, for example, a recording medium such as a CD-R, The caricature may be output to a communication network such as a mobile communication network using a mobile phone.
[0034]
The caricature creation method, the caricature creation apparatus, and the program of the present invention have been described in detail above. However, the present invention is not limited to the above-described embodiments, and various improvements can be made without departing from the gist of the present invention. Of course, changes may be made.
[0035]
【The invention's effect】
As described above in detail, according to the present invention, even if the number of facial part parts to be prepared is small, it is possible to automatically create a suitable caricature that appropriately captures the features of the model with a simple operation. Can do.
[Brief description of the drawings]
FIG. 1 is a conceptual diagram of an example of a portrait creation system according to the present invention that implements the portrait creation method of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 10 Caricature creation system 12 Input means 14 (System) Main body 16 Display 18 Printer 30 Extraction means 32 (Parts) Selection means 34 (Parts) Storage means 36 (Parts) Synthesis means 38 Caricature creation means

Claims (4)

Using multiple face image data of the same person, extract the feature quantity of each face part for each face image data,
Using the feature amount of the face part, selecting a part to be used from a part group of face parts for a caricature set in advance is performed for each face image,
A caricature creation method characterized in that parts of the same facial parts are synthesized and a portrait of the person is created using the parts obtained by the synthesis. The portrait creation method according to claim 1, wherein the composition is performed by performing weighting according to the similarity between the feature amount of the face part and the feature amount of the selected part. Storage means for storing a plurality of parts for a portrait for a plurality of facial parts;
Means for acquiring face image data;
Extracting means for extracting feature amounts of each face part for each face image data for a plurality of face image data of the same person acquired by the acquisition means;
Selection means for selecting a part to be used from the storage means for each face image using the feature amount of the face part extracted by the extraction means;
Synthesizing means for synthesizing parts of the same facial part selected by the selection means;
A caricature creation device comprising caricature creation means for creating a portrait of the person using the parts synthesized by the synthesis means. Extracting a feature amount of each face part for each face image data using a plurality of face image data of the same person;
A step of selecting, for each facial image, selecting a part to be used using a feature amount of the facial part from a storage unit that stores a plurality of parts for a portrait for a plurality of facial parts;
Composing selected parts for each same facial part,
And a program for executing a step of creating a portrait of the person using the parts obtained by the synthesis.

Images