KR100902995B1 - Method for making face image of golden ratio, and apparatus applied to the same - Google Patents

Abstract

The present invention discloses a method for forming a face image with an optimized ratio and an apparatus applied thereto. The apparatus according to the present invention includes an image preprocessing unit for converting an image of an object into an image for face recognition execution, a face region detection unit for extracting a corresponding face region by executing a face detection recognition algorithm based on the preprocessed image, and a face Feature point extracting unit for extracting feature points constituting face feature of region, database unit for storing mask template for forming face image with optimized ratio, matching feature points to mast template to form mapping ratio between mutually corresponding feature points An image matching unit and an image converter for converting the face region into an optimized ratio of the face image through the rendering process based on the mapping ratio. Accordingly, the present invention converts the image of a predetermined object into a face image of an optimized ratio, which is the ratio at which humans feel the most beautiful and comfortable, thereby reflecting the desire of a person who wants to be handsome or beautiful in the image. Creation of new services can be achieved.

Facial image, optimization

Description

Method for forming facial image with optimized ratio and apparatus applied thereto {METHOD FOR MAKING FACE IMAGE OF GOLDEN RATIO, AND APPARATUS APPLIED TO THE SAME}

The present invention relates to a technique for forming a face image, and more particularly, after receiving an image of a predetermined object, an optimization ratio for converting the face image into an optimization ratio, which is a ratio at which humans feel the most beautiful and comfortable. A facial image forming method and apparatus applied thereto.

Wireless mobile communication technology has evolved into GSM, GPRS, and WCDMA networks standardized at the 3rd Generation Partnership Projects (3GPP) International Standardization Conference. , CDMA20001x, CDMA1x-EVDO.

Among these wireless mobile communication technologies, in the case of the IMT-2000 network, the synchronous IMT-2000 is called CDMA2000 1x-EVDO, and the asynchronous IMT-2000 is called WCDMA.

On the other hand, for video telephony service, WCDMA network is provided based on wired circuit network, and part of ITU-T H.324 protocol standard is changed to 3GPP standardization organization for wireless circuit network and standardized to 3G-324M protocol standard. On the other hand, in the CDMA2000 1x-EVDO network, the ITU-T H.323 protocol standard used in the wired packet network is used for the wireless packet network without a separate standard specification in the wireless network.

In addition, the recent mobile communication terminal is equipped with a camera as an additional device for implementing the image capture and transmission, real-time video communication. Such a camera has been developed in a variety of ways, from a removable type using a jack to a type built in the terminal body.

As such, the video call service implemented through the video call network and the mobile communication terminal based on the video call network is focused on communicating by delivering a predetermined face image to the counterpart by performing video communication between service users. Accordingly, there is a problem in that it is not possible to provide additional service other than mutual communication to the service user, thereby activating the video telephone service.

In order to overcome such a problem, as the facial image is recognized in the videophone service, researches to create various additional services based on the recognized facial image have been actively conducted.

In general, recent studies on facial feature detection include methods using edge information, luminance, chrominance, geometrical appearance and symmetry of the face, principal component analysis (PCA), There is a method using template matching, an approach using a curvature of a face, a method using a neural network, and the like.

Since edge information is mostly binarized, it is difficult to produce stable results due to loss of edge information when lighting is uneven.

In addition, Principal Component Analysis has the advantage of easy real-time implementation, but it is sensitive to area separation and background image, and the neural network approach can operate stably under limited conditions, but it is difficult to adjust network variables when learning new inputs. The disadvantage is that it takes a lot of time.

Furthermore, the technology for face feature detection is further expanded to recognize face regions included in planar two-dimensional images, and thereafter, researches for converting into various images, including conversion to three-dimensional images, are actively conducted. It's going on.

Among these studies, most of the conversion to 3D images is performed by synthesizing 2D images of various directions in 4 directions, 8 directions, or 16 directions, and using a panoramic face image as a texture image. Mostly, 3D face model images are used.

In the three-dimensional image modeling method from the two-dimensional image, when a plurality of two-dimensional texture images photographing a face of a specific person in at least four directions are input, the three-dimensional general face model is inputted at the same angle as the photographing angle of the texture image. A first step of projecting onto a two-dimensional image plane, a second step of obtaining feature lines connecting feature points in the texture image, and a second step of connecting feature points in the projected general face model to obtain feature lines corresponding to the feature lines of the texture image Step 3, a fourth step of modifying the feature lines of the general face model to match the feature line of the texture image, repeating the first to fourth steps for all the texture image to obtain a face model similar to the face of a specific person Step 5, step 6, projecting all texture images onto the cylinder with radius R, two adjacent A seventh step of obtaining a visual image of an arbitrary angle using a texture image; an eighth step of obtaining an image at an arbitrary angle by interpolating two visual images of an arbitrary angle; all angles (0 degrees to 359 degrees) ) Performs a ninth step to obtain a panoramic image by performing the sixth to eighth step and a tenth step of texture mapping the panoramic image to the face model deformed similarly to the face of a specific person. You will get it.

As described above, a method for converting and modeling a 2D image into a 3D image has been studied in various ways, but studies for applying the image conversion technology to various fields have been insufficient. As a part of this, a method that can reflect the desire of people who want to be beautiful or handsome through a video screen is required.

Accordingly, the present invention was created to solve the above problems, and an object of the present invention is to convert an image with an optimized ratio, which is a ratio at which a human feels the most beautiful and comfortable after being provided with an image of a predetermined object. A method of forming a face image with an optimized ratio and an apparatus applied thereto are provided.

In addition, another object of the present invention is to set the optimization ratio which is the rate at which the human beings feel the most beautiful and comfortable as a mask template in advance to perform the image conversion for the face image input every time, and the method applied to this To provide a device.

Another object of the present invention is to convert a predetermined face image to an optimized ratio of face images based on a mask template, and to convert a face image of an optimized ratio to a three-dimensional face image and to correspond to the application program. The present invention provides a method for forming a face image having an optimized ratio for outputting an image and a device applied thereto.

According to an aspect of the present invention, there is provided an apparatus for forming a face image having an optimization ratio, the image preprocessing unit converting an image of an object into an image for executing face recognition, and detecting a face based on the preprocessed image. A database for storing a face region detector for extracting a corresponding face region by executing a recognition algorithm, a feature point extractor for extracting feature points constituting face feature elements of the face region, and a mask template for forming a face image with an optimized ratio An image matching unit for matching the feature points to the mast template to form a mapping ratio between corresponding feature points, and an image for converting the face area into a face image having an optimized ratio through rendering processing based on the mapping ratio. It characterized in that it comprises a conversion unit.

The apparatus for forming a face image of the optimized ratio may further include an image output unit for outputting the face image of the optimized ratio by executing an application program corresponding thereto.

Preferably, the feature point extracting unit extracts the feature points by performing a matching comparison between a statistical face model (AAM; Active Appearance Model) previously stored in the database unit and the face region.

Preferably, the feature point extractor extracts the feature points based on the MPEG-4 Face Animation Parameter (FAP) standard.

In addition, in order to achieve the above object, the method for forming an image having an optimized ratio according to the second aspect of the present invention comprises the steps of: (a) receiving an image of an object and performing image preprocessing for face recognition; Extracting a corresponding face region by executing a face detection recognition algorithm based on the processed image, (c) extracting feature points forming a face feature element of the face region, and (d) forming a face image with an optimized ratio Matching the feature points to a mask template for forming a mapping ratio between corresponding feature points; and (e) converting the face region into an optimized ratio face image through a rendering process based on the mapping ratio. It is characterized by.

Preferably, the method for forming a face image of the optimized ratio further comprises (F) outputting by executing an application program corresponding to the face image of the optimized ratio.

Therefore, in the present invention, after receiving an image of a predetermined object, the image is converted into a face image of an optimized ratio, which is the ratio at which humans feel the most beautiful and comfortable, and furthermore, a mask template is converted into an optimization ratio, which is a ratio at which the human is most beautiful and relaxed. It not only converts the face image with the optimized ratio to execute the image conversion for the face image that is input every time by presetting it, but also converts the face image of the optimized ratio into the three-dimensional face image and uses the corresponding application program. By being implemented to output, there is an advantage that can reflect the desire of the person who wants to be handsome or beautiful in the video and create a new service through this.

Hereinafter, a preferred embodiment of the face image forming apparatus according to the present invention with reference to the accompanying drawings in detail as follows.

First, the present invention is to extract the image corresponding to the face from the image after receiving the image for a predetermined object, and to convert the extracted face image to the optimization ratio that the human feels the most beautiful and comfortable, for this purpose Based on the mask template consisting of beautiful and comfortable optimization ratio, the corresponding face image is converted to be mapped to the mapping error range set in the mask template, thereby creating various related services as well as creating a service user. There is an effect that can activate the service created by adding a different fun that can convert the face of oneself or others to handsome or beautiful.

1 is a block diagram of a face image forming apparatus 100 according to an embodiment of the present invention. As shown by way of example only in FIG. 1, the face image forming apparatus 100 having an optimization ratio receives an image of an object (for example, a face image of a specific person), converts it to a gray value, and subsequently converts an integer value constituting the image. The image preprocessing unit 110 converts the image to the image for face recognition, which includes converting the data into a floating value for face recognition, and extracts a corresponding face region by executing a face detection recognition algorithm based on the preprocessed image. Through the feature region extracting unit 130, a statistical method, and the like, to extract the feature points forming the facial feature elements (for example, eyes, nose, mouth, and facial contour) included in the face region. Database unit 140 for storing the mask template, which is the basis for forming a face image of the optimized ratio, which is the ratio at which humans feel the most beautiful and comfortable, Matching the feature points extracted from the scrabble template, the mapping ratio between the corresponding feature points (i.e., the ratio of the width and length of the face area, the deformation of the facial contour, the ratio for adjusting the position of the eyes, nose and mouth, and the width of the nose An image matching unit 150 to form a ratio for adjusting the width, a ratio for adjusting the mouth length and the thickness of the lips, and the like, and an optimized ratio of the face region through rendering processing based on the mapping ratio. And an image converter 160 for converting the image to a face.

Furthermore, the face image forming apparatus 100 executes an application program (eg, a program for displaying an image on a terminal screen) corresponding to the face image of the optimized ratio, and thereby performs the face image of the optimized ratio. It further includes an image output unit 170 for outputting.

The feature point extractor 130 extracts feature points of the face region corresponding to the statistical face model by performing matching comparison between the previously stored statistical face model (AAM) and the face region.

In addition, the statistical face model (AAM; Active Appearance Model) is modeled by dividing the face into the shape (Shape) and texture (Texture), basically composed of the average face (Average Face) using a face test set Defining the face area in the image coming from the outside refers to a modeling technique designed to change the shape of the average face and texture parameters until they converge similarly to the face.

In addition, the feature point extractor 130 extracts at least 81 feature points or extracts at least 76 feature points based on the MPEG-4 Face Animation Parameter (FAP) standard.

Here, MPEG-4 is an ISO / IEC standard established by the Moving Picture Expert Group (MPEG) known as MPEG-1 and MPEG-2. These standards (MPEG-1, MPEG-2, and MPEG-4) enable interactive video services on CD-ROMs, DVDs, and digital TVs, while MPEG-4 provides digital TV, interactive graphics applications, and interactive multimedia. The main focus is on the world wide web, content distribution and access to content.

MPEG-4 also defines and encodes face and body animation parameters, the location of feature points and the direction of movement of the face and body definition mesh, the representation of a visme or visual lip corresponding to phonetic phonology, and the face and body definition parameters. It includes the definition and coding of the feature, the location of three-dimensional feature points, coordinate meshes for three-dimensional face animation, and individual feature points and face texture encoding.

MPEG-4's FAP (Face Animation Parameter) standard in this standardization is data defining facial animations among face and body animations, and uses bit streams to express motion information of each part of the face in applications such as games and communications using avatars. As a result of the present invention, the facial feature points constituting the facial feature element are extracted from a predetermined face image based on the information.

Furthermore, after forming an optimized proportion of the face image based on the mask template, using a three-dimensional face modeling system using AAM to reconvert the gray value of the optimized proportion of the face image based on the mask template And detecting the face region through an image preprocessing process including converting an integer value constituting the optimized ratio face image into a floating value for subsequent face recognition, and detecting the shape of the face region. And a facial feature element including a facial feature point, a brightness value, and the like, by dividing into a texture and a texture.

Subsequently, the 3D face modeling system using the AAM converts a statistical face model that is stored in advance based on the extracted facial feature elements to model a synthesized face image that matches the image forming the face region, and then synthesizes the synthesized face image. It is also possible to convert the optimized proportion of the face image to a three-dimensional face image through a rendering process of combining the mapping source to the face image.

Here, according to an embodiment of the present invention, a process of converting a face image having an optimized ratio into a 3D face image in detail will be described. If the classifier is not below the previous level, face detection is combined with Adaboost algorithm for the formation of the corresponding classifier and SVM (Support-Vector Machine) algorithm for classifying the data of the preprocessed image into the optimal feature pattern. A recognition algorithm is executed to extract a corresponding face region.

After extracting facial feature elements by dividing the shape and texture from the extracted face region, converting the statistical face model based on the extracted facial feature elements to form a synthetic face image of the previous step to form a three-dimensional face image through AAM modeling The ratio of the face is optimized by the rendering process that combines the mapping elements to represent the three-dimensional effect in consideration of the shadow, color, and density appearing differently according to external information such as shape, position, and lighting in the formed synthetic face image. The image can be converted into a three-dimensional face image.

FIG. 2 is a conceptual diagram illustrating a process of forming a face image having an optimization ratio through the face image forming apparatus 100 shown in FIG. 1. As shown only as an example in FIG. 2, after extracting a face region from an image A of a predetermined object and extracting feature points of the extracted face region, the extracted mask template B is extracted. By matching the feature points, it is determined whether the feature points of the mask template are mapped with the feature points of the extracted face region.

The rendering process for image conversion is performed based on the mapping ratio formed by determining whether the mapping is performed, and through this, the face region is converted into the face image C having an optimized ratio.

Here, the mask template may be formed through various methods including statistical methods for handsome or beautiful women, and the ratio of the most beautiful and comfortable human beings is 1: 1.618 with the optimized golden ratio. There is a result. The ratio of the results of this study includes 1: 1 face up and down and 1: 1 1: 1 eyes and glaucoma centered on the center of the eye, and a representative template is Dr Stephen's RF Mask.

Hereinafter, an operation process of the face image forming apparatus 100 according to the present invention will be described in detail with reference to the accompanying drawings.

3 is a flowchart illustrating an operation process of the face image forming apparatus 100 illustrated in FIG. 1. As shown by way of example only in FIG. 3, the method of forming a face image with an optimization ratio proceeds to performing an image preprocessing for executing face recognition by receiving an image of an object (S100 and S102).

The face detection recognition algorithm is executed based on the preprocessed image provided as a result of the execution, thereby extracting a face region of a specific person (S102).

Subsequently, a process for extracting feature points of the face region based on a statistical face model is performed (S106), and a mask template which is basic data for forming an optimized face image is loaded (S108).

As the matching between the feature points of the mast template loaded in the step S108 and the feature points forming the face region is performed (S110), a mapping ratio is formed based on the difference formed (S112). After the face region is formed into an optimized ratio face image through a rendering process based on the formed mapping ratio (S114 and S116), a corresponding application program is started to output an image (S118).

In addition, the method of forming the face image of the optimization ratio is formed as a two-dimensional image and outputs the image through a corresponding application program, and is implemented to convert the face image of the optimization ratio which is a two-dimensional image into a three-dimensional image and output the image. It is also possible.

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art will be variously modified and changed within the scope of the invention without departing from the spirit and scope of the invention described in the claims below. I can understand that you can.

In addition, the present invention is intended to convert to a face image of the optimized ratio, which is the ratio at which humans feel the most beautiful and comfortable after being provided with an image of a predetermined object, not only the possibility of commercialization or sales, but also realistically implemented. It is an invention with industrial applicability since it can be done.

1 is a block diagram of a face image forming apparatus according to an embodiment of the present invention,

FIG. 2 is a conceptual diagram illustrating a process of forming a face image of an optimization ratio using the face image forming apparatus shown in FIG. 1;

FIG. 3 is a flowchart illustrating an operation process of the face image forming apparatus shown in FIG. 1.

<Description of Symbols for Main Parts of Drawings>

100: face image forming apparatus 110: image preprocessing unit

120: face area detection unit 130: feature point extraction unit

140: database unit 150: image matching unit

160: image conversion unit 170: image output unit

Claims (6)

An image preprocessing unit converting an image of the object into an image for performing face recognition; A face region detector for extracting a corresponding face region by executing a face detection recognition algorithm based on the preprocessed image; A feature point extracting unit for extracting feature points forming a face feature of the face region; A database unit which stores a mask template which is pre-formed to form a face image having an optimized ratio; An image matching unit configured to map the feature points to the mask template to form a mapping ratio between corresponding feature points; And And converting the facial feature elements of the face region to match the facial feature elements of the mask template to a predetermined level or more based on the mapping ratio. A face image forming apparatus having an optimization ratio. The apparatus of claim 1, wherein the face image forming apparatus having the optimized ratio is And an image output unit for outputting the optimized ratio face image through execution of a corresponding application program. The method of claim 1 or 2, wherein the feature point extraction unit And extracting the feature points by performing a matching comparison between a statistical face model (AAM) previously stored in the database unit and the face area. The method of claim 3, wherein the feature point extraction unit And extracting the feature points based on the FAP (Face Animation Parameter) standard of MPEG-4. (A) receiving an image of an object and performing image preprocessing for face recognition; (B) extracting a corresponding face region by executing a face detection recognition algorithm based on the preprocessed image; (C) extracting feature points forming a facial feature element of the face region; (D) forming a mapping ratio between corresponding feature points by matching the feature points to a pre-stored mask template to form an optimized ratio face image; And (E) forming a face image of the optimized ratio by converting the facial feature elements of the face region to match the facial feature elements of the mask template to a predetermined level or more based on the mapping ratio. A facial image forming method having an optimized ratio. The method of claim 5, wherein the method for forming a face image with an optimization ratio (F) The method for forming a face image of the optimized ratio, characterized in that it further comprises the step of outputting a face image of the optimized ratio by executing an application program.

Images