KR100716422B1 - System and method for matching service using pattern recognition - Google Patents

Abstract

A matching service system using pattern recognition that provides a matching service of a face or a hand to subscribers by utilizing a pattern recognition technique and a method thereof are described. To this end, at least one terminal interlocked with a wired or wireless network; A pattern recognition server for searching for a matching face using a pattern recognition technique based on an image file of a photograph transmitted from the terminal and transmitting the result to the terminal; And a face information DB storing face pictures and face pattern information of each layered person, the terminal generating an image file and transmitting the image file to the pattern recognition server; Recognizing a pattern of the image file by the pattern recognition server and extracting similar photo and similarity information by comparing the pattern with the image stored in the face information DB; And transmitting the similar photo and / or photo-similarity information to the terminal via the wired / wireless network, thereby providing the service provider with new revenue generation and providing the user with joy .

Pattern recognition, pattern recognition server, standard face extractor, similarity, matching service

Description

[0001] SYSTEM AND METHOD FOR MATCHING SERVICE USING PATTERN RECOGNITION USING PATTERN RECOGNITION [0002]

1 is a block diagram illustrating a matching service system using pattern recognition according to the present invention;

2 is a configuration diagram showing a pattern recognition server according to the present invention;

3A to 3E are views for explaining a face recognition method according to the present invention.

4 is a flowchart showing a matching service method using pattern recognition according to the present invention.

Description of the Related Art [0002]

100: terminal

200: Face Recognition Server

210: Face recognition engine

220: Standard face extractor

230: Similarity Analyzer

300: face DB

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a matching service system and a method thereof, and more particularly, to a matching service system and method using pattern recognition that provides a matching service of a face or a hand to a subscriber using a pattern recognition technique.

The existing pattern recognition technology occupies an important position in the field of image recognition and machine vision, and active research is underway as a next generation technology for human interface. For example, in addition to fingerprint / biometrics, there are many applications for security, surveillance, and access control related to access at offices, airports, and casinos.

Meanwhile, recently, digital cameras and cameras (hereinafter referred to as camera phones) attached to mobile terminals have been widely used, and their use has been greatly increased, and they have become a necessity among the general public. Accordingly, for the mobile communication service provider and the contents provider, it is necessary to develop the additional service based on the camera.

An object of the present invention is to provide a matching service system and method using pattern recognition that provides an additional service using a degree of similarity with a photograph, in cooperation with a wired or wireless network, with respect to a photograph captured by a digital camera and a camera phone have.

In order to achieve the above object, a matching service system using pattern recognition according to the present invention includes at least one terminal having a function of receiving and storing a photograph through a wired or wireless network or photographing; A pattern recognition server for searching for matching person information based on an image file of a photograph transmitted from the terminal and transmitting the result to the terminal; And a face information DB storing a plurality of portrait photographs and face pattern information.

According to another aspect of the present invention, there is provided a matching service method using pattern recognition, including: (a) transmitting, by a terminal, an image file to a pattern recognition server; (b) the pattern recognition server recognizes a face pattern of the image file and compares the face pattern with a photograph stored in the face information DB, and then extracts a similar photograph and / or similarity information; and (c) transmitting the similar photo and / or photo similarity information to the terminal.

The step (b) may further include: (b-1) extracting a face from the image file; (b-2) extracting a main feature portion from the face extracted in the step (b-1) to generate a standard face; And (b-3) analyzing the similarity with the face pattern stored in the face DB based on the standard face.

According to another aspect of the present invention, there is provided a matching service method using pattern recognition, including: (a) receiving a matching service menu from a service providing server; (b) the terminal receiving a service menu selection from among the service menus; (c) transmitting the user image information and / or the matching object information to the pattern recognition server through the relay server; (d) receiving the image file corresponding to the matching target name from the face DB and matching the matching image with the user image file; And (e) transmitting the image file and / or similarity information according to the matching result to the terminal by the pattern recognition server.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and preferred embodiments. In the drawings, the same or similar components are denoted by the same reference numerals.

1 is a block diagram illustrating a matching service system using pattern recognition according to the present invention.

As shown in FIG. 1, the system includes at least one terminal 100 interlocked with a wired or wireless network; A face recognition server 200 for searching for a matching face using a pattern recognition technique based on an image file of a photograph transmitted from the terminal 100; And a face DB 300 in which photographs and face pattern information of each layered person are stored.

The server 200 further includes a relay server (not shown) that interprets the service address added to the image file and transmits the interpreted service address to the face recognition server 200. For example, the relay server may be a multimedia messaging service (MMS) server or a photo mail server.

The terminal 100 is based on wired or wireless, and does not necessarily include a camera, but includes a memory (not shown) capable of receiving and storing photographs from the outside.

As shown in FIG. 2, the face recognition server 200 includes a face recognition engine 210 for extracting face images from the image file; A standard face extractor 220 for extracting feature portions of the extracted face using various algorithms; And a similarity analyzer 230 for analyzing the similarity between the standard face and the face pattern stored in the face DB 300 based on the standard face.

2, the face recognition engine 210 includes a GFCD converter 211 for emphasizing the face color, a BWCD converter 212 for emphasizing black and white, A preprocessor 213 for enhancing the image of a specific portion of the image file to separate the face and the background image, and a background labeler 214 for separating the face and the background image.

A matching service method using pattern recognition according to the present invention will be described with reference to the accompanying drawings.

First, an image file is generated and transmitted to the face recognition server 200 (S401). That is, the matching service user captures a face through a camera to generate an image file, and then transmits the generated image file to a relay server. Then, the relay server analyzes the service address (for example, # 5454) added to the image file and transmits the image file to the face recognition server 200 if the service address corresponds to the matching service.

Then, the face recognition server 200 recognizes the pattern of the image file and extracts a similar photo from the face DB 300 (S402). The step 402 can be subdivided into a process of 1) identifying a specific position of a face, 2) extracting features of a face, and 3) analyzing the similarity with the photographs stored in the face DB 300. This will be described in more detail.

1) The process of identifying a specific position of a face

As shown in FIGS. 3A to 3D, the face recognition engine 210 identifies the positions of the face, eyes, nose, and mouth in a step prior to face recognition in the image file (S402-A). For this purpose, a facial color distribution process, a monochrome color distribution process, a binarization process using a threshold value, and a background labeling process are sequentially performed.

(a) First, the GFCD converter 211 performs face color distribution processing for extracting a face region. The facial color distribution is called GFCD, and it is sometimes described as GSCD in face recognition or image recognition.

Usually, the captured image has a frequency of color occurrence with respect to the skin color concentrated around a specific region. Therefore, in order to extract only the skin color of the face in the input color image, the color value of a pixel having a value similar to that of the face skin color has a high probability value, and the other color region has a low probability value.

(b) Next, the BWCD converter 212 performs black-and-white color distribution processing for extracting facial region through skin color. The black-and-white color distribution process refers to a process of normalizing an input image to a luminance component in order to prevent sensitivity to changes in illumination or other surrounding conditions in the process of using color information.

More specifically, when each pixel of the color image input to the face search engine 210 is a three-dimensional color model of RGB, each pixel of the input color image is normalized to a luminance component R + G + B . Then, since the normalized color model is b = 1- (r + g), each of the pixels can be represented by only two elements r and g.

Then, in the color model that has undergone the normalization of the luminance component, the portion of interest is the skin color of the human skin and white and black colors of the main constituent components in the face. The skin color and white and black colors of various conditions in which the luminance component is removed are concentrated in a narrow region around a specific value.

Conversion using this distribution will allow you to emphasize specific colors. Accordingly, the face search engine 210 defines the BWCD conversion for emphasizing the GSCD and the white and black colors for emphasizing skin color as shown in the following Equation (1).

Y (x, y) = G (r (x, y), g (x, y)

In this color conversion process, the region corresponding to the desired color (mr, mg) in the color image is displayed as a high contrast value, and the region having a different color from the color is displayed as a low contrast value. That is, by adjusting the values of (mr, mg), it is possible to extract only flesh color and white and black areas. Y (p) is the transformed result and G (p) is the two-dimensional Gaussian function. r (p) and g (p) denote the r and g components of the input image, respectively, expressed as a normalized color model.

(c) Next, a preprocessing process for extracting a specific part of a face in the preprocessor 213, that is, a binarization process using a threshold value is performed. Binary processing, which is one of the most basic techniques of image processing, is a relatively simple algorithm. This binarization technique is used to enhance the image as a preprocessing process for extracting specific information from fingerprint recognition technology and character recognition technology.

(d), the background labeling unit 214 performs a background labeling process for extracting only the face area. That is, background labeling separates the face and the background from the captured image. The link component analysis, or labeling, is a module used to extract the connected components while scanning the input image, and it is scanned from top to bottom and from left to right. Analysis of the linkage component enables analysis of size and location for each lump, which is used as important information for candidate region determination. In other words, it scans from top to bottom, left to right, and finds only the portion of the entire image that corresponds to the face area. This way, the face area is found through the labeling.

2) The process of extracting facial features

As shown in FIG. 3E, the standard face extractor 220 extracts a standard face by geometric method, Eigenfaces, fuzzy algorithm, neural network algorithm, genetic algorithm, etc., and any of them can be used. Accuracy, analysis time, etc.) (S402-B).

3) a process of analyzing the degree of similarity with the photographs of the face DB 300

The similarity analyzer 230 analyzes the facial region information of the extracted facial features by using the facial contour model, the distance between the feature points, the angle, and the weight, The degree of similarity with the reference face is determined from the DB 300 (S402-C).

That is, A = {a1, a2, ... , an}: the shape feature set of the query object and B = {b1, b2, ... , bn}: Define the similarity function d (A, B) with the shape feature set of the object stored in the database, and determine the degree of similarity between A and B through this. At this time, the retrieved photographs, the photograph similarity information, and the like are extracted as the retrieval result values. In this case, the parameters for determining the degree of similarity include δ: regional weight of feature, φ: similarity coefficient (0.5), λ: global weight of shape, and w:

Then, the similar photo and / or photo similarity information is provided to the terminal 100 of the service user through the wired / wireless network (S403)

Preferred embodiments of the matching service method using pattern recognition according to the present invention described so far will be described, respectively. However, the present invention is applicable to a wireless network (e.g., WIBRO (Wireless Broadband), etc.) other than a mobile communication network and a wired network (e.g., TCP / IP, PSTN, etc.) And can be serviced through the terminal 100.

(Embodiment 1) The similarity service of a specific person himself /

And accesses the wireless Internet through the terminal 100 to receive a matching service menu from the service providing site. At this time, the terminal 100 receives a service menu selection corresponding to the embodiment among the corresponding service menus from the user.

The terminal 100 then transmits to the face recognition server 200 via the relay server (for example, an MMS server in the mobile communication network) the name of the person himself / herself and the specific person (for example, politician, entertainer, etc.) .

Then, the face recognition server 200 activates the face recognition engine 210 to receive the photograph of the specific person from the face DB 300, and matches the photograph with the user's photograph. The result value at this time, that is, similarity degree, is transmitted to the terminal 100 through the mobile communication network.

(Embodiment 2) A person who has the highest degree of similarity with a specific person of this week

And accesses the wireless Internet through the terminal 100 to receive a matching service menu from the service providing site. At this time, the terminal 100 receives a service menu selection corresponding to the embodiment among the corresponding service menus from the user.

Then, the terminal 100 transmits its own photo stored in advance to the face recognition server 200 through a relay server (e.g., an MMS server in a mobile communication network).

Then, the face recognition server 200 activates the face recognition engine 210 to receive the photograph of the specific person from the face DB 300, and matches the photograph with the user's photograph. At this time, the service provider selects the highest user among the results of similarity between the photographs transmitted from the at least one terminal 100 and the specific person, and transmits the winning information together with the similarity value to the user terminal 100 do. The above embodiment can be applied to a person contest similar to an entertainer.

(Third embodiment) Similarity degree service

And accesses the wireless Internet through the terminal 100 to receive a matching service menu from the service providing site. At this time, the terminal 100 receives a service menu selection corresponding to the embodiment among the corresponding service menus from the user.

Then, the terminal 100 transmits a plurality of pictures to be compared to the face recognition server 200 through a relay server (e.g., an MMS server in a mobile communication network).

Then, the face recognition server 200 activates the face recognition engine 210 to match the two pictures. The result value at this time, that is, similarity degree, is transmitted to the terminal 100 through the mobile communication network.

The above embodiment can be applied to a comparison service of similarity between couples or a comparison service of whether babies resemble mother more or father more.

(Fourth Embodiment) A search service for a missing person or a wanted person

And accesses the wireless Internet through the terminal 100 to receive a matching service menu from the service providing site. At this time, the terminal 100 receives a service menu selection corresponding to the embodiment among the corresponding service menus from the user.

Then, the user picks up a suspect person who is suspected of being a child or a suspect through the terminal 100 and transmits the image to the face recognition server 200 through a relay server (e.g., an MMS server in a mobile communication network).

Then, the face recognition server 200 activates the face recognition engine 210 to match the pictures transmitted from the terminal 100 with the pictures stored in the police database. And transmits the resultant value, whether it is a missing person or a criminal, to the terminal 100 through the mobile communication network.

The above embodiment can be applied to a missing child search service, a search service for searching for an alien, an alumni search service, a service for finding a parent of an adopted child, and the like.

Meanwhile, although the face is taken as an example in the present invention, a matching service system and method using the pattern recognition can be applied to a part of a living body (for example, a fingerprint, a palm, etc.).

Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the present invention.

As described above, according to the present invention, various additional services are provided to a user by interlocking a pattern recognition technique with widely-used camera phones and communication networks, thereby creating a new revenue source for a service provider and enjoyment for a user .

Claims (13)

At least one terminal having a function of receiving and storing a photograph through a wired or wireless network or taking a photograph; A pattern recognition server for searching for matching person information based on an image file of a photograph transmitted from the terminal and transmitting the result to the terminal; And And a face information DB storing a plurality of portrait photographs and face pattern information, The pattern recognition server includes: A face recognition engine for extracting face information from the image file, the face recognition engine comprising: a GFCD converter for adjusting face color of the image file; a BWCD converter for adjusting black and white of the image file; A preprocessor for adjusting an image of a specific portion of the image file, and a background labeler for separating the face and background image from the image file; A standard face extractor for extracting a feature portion of the extracted face information using at least one algorithm; And And a similarity analyzer for analyzing a similarity between the standard face extracted from the standard face extractor and the face pattern stored in the face information DB based on the standard face extracted by the standard face extractor. The method according to claim 1, Further comprising: a relay server for analyzing a service address added to the image file and transmitting the interpreted service address to the pattern recognition server. delete delete (a) the terminal transmitting an image file to a pattern recognition server; (b) the pattern recognition server recognizes a face pattern of the image file and compares the face pattern with a photograph stored in the face information DB, and then extracts a similar photograph and / or similarity information; And (c) sending the similar photo and / or photo similarity information to the terminal, The step (b) (b-1) extracting face information from the image file, wherein the step (b-1) comprises the steps of: performing color distribution processing of face information for extracting a face region in the image file; Performing a black-and-white color distribution process for extracting a face area through the image file, performing a binarization process using a threshold value to extract a specific region of a face in the image file, Performing a background labeling process; (b-2) extracting a main feature part from the face information extracted in the step (b-1) to generate a standard face; And (b-3) analyzing the similarity with the face pattern stored in the face information DB based on the standard face. delete delete 6. The method of claim 5, wherein step (a) (a-1) the terminal generates an image file, adds an address of a corresponding service server, and transmits the image file to a relay server; And (a-2) the relay server analyzes the service address added to the image file and transmits the image file to the pattern recognition server if the service address corresponds to the matching service. Way. (a) a terminal receiving a matching service menu from a service providing server; (b) the terminal receiving a service menu selection from among the service menus; (c) the terminal transmitting user image information and / or comparison object information to a pattern recognition server through a relay server; (d) receiving the image file corresponding to the comparison object information from the face information DB and matching the image file with the user image information; And (e) transmitting, by the pattern recognition server, similarity information according to the matching result to the terminal and the service providing server, The step (d) (d-1) extracting face information from the image file, wherein the step (d-1) comprises: performing color distribution processing of face information for extracting a face region in the image file; Performing a black-and-white color distribution process for extracting a face area through the image file, performing a binarization process using a threshold value to extract a specific region of a face in the image file, Performing a background labeling process; (d-2) extracting a main feature part from the face information extracted in step (d-1) to generate a standard face; And (d-3) analyzing the degree of similarity with the face pattern stored in the face information DB based on the standard face. 10. The method of claim 9, (f) transmitting, by the service provider, the fact of winning to the terminal corresponding to the highest degree of similarity among at least one or more similarity information transmitted from the pattern recognition server. delete delete 11. The method according to claim 9 or 10, wherein step (c) Wherein the user identification information is user image information and / or comparison information captured by the mobile terminal.

Images