JP2010207519A - Related human retrieval system, related human retrieval method, and related human retrieval program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a system or the like for searching a human related to a player, when the player commits a fraudulent act in a game. <P>SOLUTION: In a related human retrieval server, information which combines a human ID that is an ID for identifying a human and face image data of a human, and a management table for managing game time and a position of a game machine are stored. Image information imaged by an image imaging device is obtained. It checks whether a face of the human appears in the obtained image, and when the face appears, a face area in the image is detected. It identifies what ID is assigned to the human in the face image by searching the face image similar to the detected face image. It determines which game machine the identified human in the face image plays in. The game time is calculated based on the time when the face is authenticated. A related score between players is searched from the obtained human ID, the position of the game machine, and the game time. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a person search system, a search method, and a search program, and more particularly to a system, method, and program for searching for a person related to a certain person.

  In recent years, in gaming machines such as pachinko machines and pachislot machines, there has been an endless crime of intentionally applying some mechanism to gaming machines and exploiting the balls. In order to solve this problem, a method of installing a fraud prevention device in the gaming machine itself has been proposed (see, for example, Patent Document 1). However, even if such a fraud prevention device is installed, a fraudulent game method for avoiding it is newly developed. Therefore, in addition to the method of installing an anti-fraud device on the gaming machine itself, measures such as early detection of an illegal player are also required.

  In view of this point, as a method to prevent playing against a person who has been confirmed as an unauthorized player, a surveillance camera is arranged at the entrance and exit of a game store, and the player's face image captured by the surveillance camera, There has been proposed a method of searching for a pre-registered fraudulent player's face image using face recognition technology.

  As such a method, for example, in the technique described in Patent Document 2, when an unauthorized player visits a store, a warning is issued to prevent the unauthorized player from playing a game.

  Similarly, in the technique described in Patent Document 3, a method of adaptively changing the threshold value in face matching according to the detection frequency in a video monitoring system using face authentication has been proposed. According to the proposed method, a mechanism is introduced that makes it easier to identify an individual with respect to a person with a high detection frequency by lowering the matching threshold when the detection frequency is high.

JP 2006-110311 A JP 2001-178958 A JP 2007-061394 A

H.Rowley, S.Baluja, andT.Kanade, "HumanFaceDetectioninVisualScenes," AdvancesinNeuralInformationProcessingSystems8,1996, pp.875-881. M. Turkand A. Pentland, “Facerecognitionusingeigenfaces”, ProceedingsofIEEE, CVPR, pp. 586-591, Hawaii, June, 1991.

  The invention described in each of the above-mentioned patent documents mainly assumes an independent action of an unauthorized player. However, in general, when a fraudulent player commits a fraudulent game, he / she often acts in a plurality. However, the method based on the behavior analysis of a single person as described above cannot grasp the relationship between players.

  Therefore, the present invention relates to a related person search system, a related person search method, and a related person search that can search information related to a person related to the person when a certain illegal player invades. The purpose is to provide a program.

  According to the first aspect of the present invention, in a related person search server for extracting a person highly relevant to a certain person, a person ID that is an ID for identifying the person and the face image data of the person are obtained. A registration information database for storing the combined information, a management table for managing the game time and the game machine position, face image acquisition means for acquiring image information captured by the image capturing device, and a person in the acquired image A face area detecting means for detecting a face area in the image, and a face image similar to the detected face image from the registered information database. By searching for a face image collating means for identifying which ID of the person of the face image is assigned, and on which gaming machine the person of the identified face image is playing Size Based on the game position detection means for storing the determination result in association with the ID in the registration information database and the time when the face is authenticated, the game time is calculated, and the calculation result is stored in the registration information database. And a game time recording means for storing the game in association with the person, and a person relevance score calculating means for obtaining a relevance score between players from the obtained person ID, gaming machine position and game time. A related person search server is provided.

  According to a second aspect of the present invention, a search result for displaying a search result of the related person search server and a person ID input means for inputting an ID of a person who wants to search for a related person in the related person search server. In the search result display system comprising display means, the related person search server searches for a person highly relevant to the ID input from the person ID input means using the relevance score, and the search result The search result display system is further provided with related person search means for notifying the search result display means.

  According to a third aspect of the present invention, in a related person search method for extracting a person highly related to a certain person, a person ID that is an ID for identifying the person and the face image data of the person are obtained. A step of preparing a registration information database for storing the combined information, a management table for managing game time and game machine position, a face image acquiring step for acquiring image information captured by the image capturing device, and the acquired It is checked whether or not a person's face is shown in the image, and if the face is shown, a face area detecting step for detecting a face area in the image; and the detected face image from the registration information database; By searching for similar face images, a face image matching step for specifying which ID of the person of the face image is assigned, and which person of the specified face image is Based on the gaming position detection step of determining whether or not a game is being played on the gaming machine and storing the determination result in association with the ID in the registration information database, and calculating the gaming time, and calculating the gaming time A game relevance score calculation for obtaining a relevance score between players based on a gaming time recording step of storing a result in association with the ID in the registration information database, and the obtained person ID, gaming machine position, and gaming time. And a related person search method characterized by comprising the steps of:

  According to the fourth aspect of the present invention, in a related person search program for extracting a person highly relevant to a certain person, a person ID that is an ID for identifying the person and the face image data of the person are obtained. A registration information database for storing the combined information, a management table for managing the game time and the game machine position, face image acquisition means for acquiring image information captured by the image capturing device, and a person in the acquired image A face area detecting means for detecting a face area in the image, and a face image similar to the detected face image from the registered information database. By searching for a face image matching means for identifying which ID of the person of the face image is assigned, and on which gaming machine the person of the identified face image is playing Based on the gaming position detection means for determining and storing the determination result in association with the ID in the registration information database, and the time when the face is authenticated, the game time is calculated, and the calculation result is stored in the registration information database. Related person search comprising: a game time recording means for storing in association with an ID; and a person relevance score calculating means for obtaining a relevance score between players from the obtained person ID, gaming machine position and game time. A related person search program characterized by causing a computer to function as a server is provided.

  According to the present invention, by obtaining the relationship between players, when a certain player is playing an illegal game, it becomes possible to search for a person related to that player. It is possible to improve the detection rate of fraudulent players.

It is a block diagram showing the basic composition of the embodiment of the present invention. It is the flowchart (1/2) which showed the flow of the process for calculating a person relevance score. It is a figure shown about an example of a game time and a position management table. It is the flowchart (2/2) which showed the flow of the process for calculating a person relevance score. It is the flowchart which showed the flow of the process for searching related person ID.

  Here, the basic idea of the present invention will be described. In the embodiment of the present invention, a score related to the relationship between persons is calculated so that the value increases as the game position is closer and the game time overlap is larger between two players who are playing.

  Then, using the obtained score relating to the relationship between the persons, a person having a close relationship with the unauthorized player is searched. The greater the score value, the closer the game tables are, and the greater the overlap of the game time. Therefore, it can be said that the higher the score, the higher the relevance of the unauthorized player. This is the basic idea of the present invention.

  Next, the best mode for carrying out the invention will be described in detail with reference to the functional block diagram of FIG.

  Referring to FIG. 1, the embodiment of the present invention includes an image capturing unit 100, a related person search server 200, a person ID number input unit 310, and a search result display unit 320. Further, the related person search server 200 includes a face image acquisition unit 210, a face area detection unit 220, a face image collation unit 230, a game position detection unit 240, a game time recording unit 250, a person relevance score calculation unit 260, a related person search. Section 270 and registration information database 280.

  The image photographing unit 100 is an image photographing device represented by a camera, and is a part that obtains an image. In the present embodiment, a specific image acquisition method is not particularly limited, and an arbitrary type of camera or the like can be used.

  The face image acquisition unit 210 is a part for inputting an image obtained by the image photographing unit 100 to the related person search server 200.

  The face area detection unit 220 is a part that checks whether or not a face is reflected in the image obtained by the image photographing unit 100 and identifies the position of the face in the image when the face is reflected.

  The face image collation unit 230 specifies which ID is assigned among the registered persons by searching for similar face images from a database in which information is registered in advance. Part.

  The gaming position detection unit 240 is a part that determines which gaming machine the person of the identified face image is playing.

  The game time recording unit 250 is a part that records the game time based on the time when the face is authenticated.

  The person relevance score calculation unit 260 obtains the relevance between the players based on the person ID, the gaming machine position, and the gaming time obtained from the face image matching unit 230, the gaming position detection unit 240, and the gaming time recording unit 250. Part.

  The related person search unit 270 is a part that performs processing for searching for related persons.

  The person ID number input unit 310 is a part for inputting the ID number of a person to be searched to the related person search server 200. The search result display unit 320 displays search results in the related person search server 200.

  As shown in FIG. 1, the registration information database 280 is a database in which a person ID, face image data, a game time / table position management table, and a person relevance score management table are accumulated. Information other than that exemplified here may be added to the registration information database 280 and stored. Specifically, when there are different types of game machines such as pachinko and pachislot machines, the distance between the game machines may be held in advance as a table. A method of using this table will be described later.

  Each part described above is used for two operations. First, the image photographing unit 100, the face image obtaining unit 210, the face image region extracting unit 220, the face image collating unit 230, the game position detecting unit 240, the game time recording unit 250, and the person relevance score calculating unit 260 are: The image obtained by the face image photographing unit 100 is analyzed, and the information for calculating the relevance score of the person and the calculated relevance score of the person are used to accumulate in the registration information database 280.

  Second, the related person search unit 270, the person ID number input unit 310, and the search result display unit 320 search for a person related to the unauthorized player from the person relevance score accumulated in the registration information database 280. To use.

  Next, processing for accumulating information in the registration information database 280 will be described in detail with reference to the flowchart of FIG.

  First, the image capturing unit 100 acquires an image with an image capturing device represented by a camera or the like (step S101). In this embodiment, since it is necessary to record the game time, it is most desirable to always acquire data with a video camera or the like. However, for example, data may be acquired with a frequency of about once per minute. .

  Next, the face image acquisition unit 210 acquires the image data acquired by the face image photographing unit 100, and inputs the acquired image data to the face area detection unit 220 (step S102). When the related person search server 200 acquires the image data, the following person relevance score can be calculated.

  The face area detection unit 220 extracts a face area from the image input from the face image acquisition unit 210 (step S103). As a method for extracting a face region from an image, for example, a method described in Non-Patent Document 1 is known. In the present embodiment, the face position is designated as a rectangular area in the image. If no face is detected in the image, it is assumed that there is no player in the captured range, and the processing for the main image ends. Note that the method described in Non-Patent Document 1 is merely an example, and any method can be used as a method for extracting a face region from an image in the present embodiment.

  After that, the face image matching unit 230 identifies a person from the image shown in the face area. As a premise, it is assumed that facial images of people who have visited the store in the past are stored in the registration information database 280. In addition, it is assumed that an ID number is assigned to each of the accumulated persons. All persons are managed by ID numbers. The face image collation unit 230 collates the face image held in the registration information database 280 with the face image obtained by the face area detection unit 220, and the ID of the person shown in the image obtained by the image photographing unit 100 Identify the number. In face image matching, since the similarity between persons is usually calculated as a score, the score of the person with the highest similarity among the registered persons is obtained (step S104).

  Then, it is confirmed whether the score of the person with the highest degree of similarity exceeds the threshold value. If the score exceeds the threshold value (Yes in step S105), the ID number of the person with the highest degree of similarity is assigned (step S106).

  On the other hand, if it is equal to or less than the threshold (No in step S105), the person of the current face image is assigned as a new person, and a new ID is assigned and registered in the database 280 (step S107).

  As one of face image collation methods, for example, the method described in Non-Patent Document 2 is known. Note that the method described in Non-Patent Document 2 is merely an example, and any method can be used in the present embodiment, as in the above-described face region extraction.

  Subsequently, the game position detection unit 240 detects the position where the person specified by the face image matching unit 230 is playing (step S108). In the present embodiment, the game position is equal to the position of the gaming table, but the game position may be determined using another method.

  It should be noted that the simplest method for specifying the position of the gaming machine position from the image is a case where cameras are installed in the vicinity of each machine, and only the person playing the game on that machine is shown on each camera. In this case, the game table can be specified by the number of the camera that the player is playing. If a plurality of cameras are used, the position of the game table can be specified using triangulation or the like. As long as the position of the play stand can be specified, a method using a sensor other than the means for acquiring the face image is also conceivable.

  Next, the game time recording unit 250 measures the time that the person specified by the face image matching unit 230 is playing (step S109). The acquisition time of the first image in which the person can be identified by the face image matching unit 230 is set as the game start time, and the time when the image is acquired last for the same person is set as the game end time. If you are away for a short time even in the middle of a game or if your face is not facing the camera, the verification process may fail. If the person is not detected even after shooting for a minute, it is necessary to process the start / end time so as not to deviate when the verification fails, such as the end of the game time.

  The game machine position and game time data obtained by the game position detection unit 240 and the game time recording unit 250 are stored in the registration information database 280 and managed. FIG. 3 shows an example of the game time / position management table. As shown in FIG. 3, a person ID number, a game machine number, a game start time, and a game end time are managed in a table format.

  Next, a process in which the person relevance score calculation unit 260 calculates the relevance of a person playing a game as a score will be described. Here, it is assumed that the score representing the relevance is a value determined from the game position and game time between two players who play. The process of calculating the person relevance score will be described with reference to the flowchart of FIG.

  Referring to FIG. 4, the process of calculating the person relevance score includes step S <b> 201 for calculating the overlap of game times, step S <b> 202 for calculating the distance between game tables, and step S <b> 203 for calculating the person relevance score. Is done.

  In this process, the ID numbers of two persons who are targets for score calculation are input. Here, the ID numbers of the two persons to be targeted are A and B, respectively.

  First, for the persons with ID numbers A and B, the total time over which the game time overlaps in a certain survey period is calculated from the information registered in the game time / position management table in the registration information database. (Step S201). Since the value of the overlap period of the game tends to increase as the survey period becomes longer, it is more likely that an accurate result will be obtained if the survey period is limited to a certain time, such as the last three months.

  Next, the similarity regarding the positional relationship of the player is calculated from the distance of the game table (step S202). As the simplest calculation method, the game machine position is measured in advance, and the distance between the machines is set as the distance between game companies. As described above, when a plurality of the same machines are delivered at the gaming facility or there are different types of gaming machines such as pachinko and pachislot machines, the distance between the gaming machines is held in advance as a table due to the nature of the gaming machines. Alternatively, the similarity may be calculated by referring to the table. For example, when playing on the same kind of game stand, the degree of similarity may be increased.

  Subsequently, a score representing the relationship between persons is calculated (step S203). A score z representing the relationship between persons is obtained by the following formula 1, where the variable x is the time over which the game overlaps and the variable y is the distance between the game tables.

但し、
ｚ：人物間の関係性を表すスコア
ｘ：遊技が重なる時間
ｙ：遊技台間の距離
ａ：遊技が重なる時間の重みを決定するための係数
ｂ：遊技台間の距離の重みを決定するための係数
であるものとする。

However,
z: score representing the relationship between persons x: time for game overlap y: distance between game tables a: coefficient for determining weight of time for game overlap b: for determining weight of distance between game tables It is assumed that the coefficient is.

  Here, the variables a and b are coefficients for determining the distance between the game tables and the weight of the time when the games overlap, and take positive values. In general, the score z becomes longer as games overlap and the distance between game tables becomes shorter. Therefore, the greater the value of z, the more relevant the person can be determined to be.

  A score representing the correlation between the two parties obtained by the above processing is calculated for all combinations among all registrants and stored in the registration information database 280. The table holding the score has N × N elements when there are N registered persons.

  Next, a method for searching for a person related to a specific player based on the score related to the person's relevance obtained in the above process will be described. For the description, reference is made to the flowchart of FIG.

  First, the ID of a player who wants to specify a related person is input from the person ID number input unit 310 (step S301). The ID to be entered is a person ID number described in the registration information database 280.

  Next, the related person search unit 270 searches for related persons based on the score obtained by the person relevance score calculation unit 260 (step S302). The related person search unit 270 inputs the ID number acquired by the person ID number input unit 310. In step S302, the score of the person related to the input ID number is extracted from the person relevance score table of the registration information database 280. Here, it is assumed that the extracted score group is Z1, Z2, Z3,.

  Thereafter, the extracted score group is sorted in descending order (step S303). This is because the higher the score, the more relevant the person.

  Next, the IDs of the top m persons of the sorted scores are extracted (step S304). The processing of this step may be a method of extracting the IDs of persons exceeding the threshold by dividing the score by a threshold instead of extracting the IDs of the top m persons. The ID numbers for m names extracted in this step (or for persons exceeding the threshold) are output to the related person search unit 270 and delivered to the search result display unit.

  The search result display unit 320 displays the ID number obtained by the related person search unit 270. With the above processing, it is possible to retrieve the ID number of a person whose game time / game position is close from the ID of a person who seems to be an unauthorized player.

  Each unit and the related person search server 200 according to the embodiment of the present invention can be realized by hardware, software, or a combination thereof.

  Moreover, although the above-described embodiment is a preferred embodiment of the present invention, the scope of the present invention is not limited only to the above-described embodiment, and various modifications are made without departing from the gist of the present invention. Implementation in the form is possible.

  According to the present invention, a person highly relevant to a certain person can be extracted. In particular, the present invention is suitable for applications such as detecting unauthorized players in gaming facilities such as pachinko stores and pachislot stores.

DESCRIPTION OF SYMBOLS 100 Image photographing part 200 Related person search server 210 Face image acquisition part 220 Face area detection part 230 Face image collation part 240 Game position detection part 250 Game time recording part 260 Person relevance score calculation part 270 Related person search part 310 Person ID number Input unit 320 Search result display unit 280 Registration information database

Claims (15)

In a related person search server for extracting a person highly relevant to a certain person,
A registration information database that stores information that combines a person ID, which is an ID for identifying a person, and face image data of the person, and a management table that manages game times and game table positions;
Facial image acquisition means for acquiring image information captured by the image capturing device;
A face area detecting means for examining whether or not a person's face is reflected in the acquired image and detecting the face area in the image when the face is reflected;
A face image collating means for identifying a face image similar to the detected face image from the registered information database to identify which ID of the person of the face image is assigned;
Game position detection means for determining which gaming machine the person of the identified face image is playing, and storing the determination result in association with the ID in the registration information database;
Game time recording means for calculating a game time based on the time when the face is authenticated, and storing the calculation result in association with the ID in the registration information database;
A person relevance score calculating means for obtaining a relevance score between players from the obtained person ID, gaming machine position and game time;
The related person search server characterized by comprising. The related person search server according to claim 1, wherein the person relevance score calculating means includes:
The related person search server characterized in that the person relevance score is calculated for each combination of IDs and recorded in the registration information database as a person relevance score management table. The related person search server according to claim 1 or 2, wherein the person relevance score calculation means includes:
The following formula 1,

However,
z: score representing the relationship between persons x: time for game overlap y: distance between game tables a: coefficient for determining weight of time for game overlap b: for determining weight of distance between game tables The coefficient of
The related person search server characterized by calculating | requiring the said relevance score using. The related person search server according to any one of claims 1 to 3,
A person ID input means for inputting an ID of a person who wants to search for a related person to the related person search server;
In a search result display system comprising search result display means for displaying a search result of the related person search server,
The related person search server searches for a person highly relevant to the ID input from the person ID input means using the relevance score, and notifies the search result display means of the search result display means. A search result display system further comprising means.   5. The search result display system according to claim 4, wherein the related person search means extracts and extracts relevance scores between the input ID and all IDs registered in the registration information database. A search result display system that sorts score groups in descending order and notifies the search result display unit of IDs of the top m (m is an arbitrary natural number) names of the sorted scores. In a related person search method for extracting a person highly relevant to a certain person,
Providing a registration information database for storing information combining a person ID, which is an ID for identifying a person, and face image data of the person, and a management table for managing game time and game table position;
A face image acquisition step of acquiring image information captured by the image capturing device;
A face area detection step of checking whether or not a person's face is reflected in the acquired image and detecting the face area in the image when the face is reflected;
A face image collation step for identifying a face image similar to the detected face image from the registered information database to identify which ID of the person of the face image is assigned;
A game position detecting step of determining in which gaming machine the person of the identified face image is playing, and storing the determination result in association with the ID in the registration information database;
A game time recording step of calculating a game time based on the time when the face is authenticated, and storing the calculation result in association with the ID in the registration information database;
A person relevance score calculating step for obtaining a relevance score between players from the obtained person ID, gaming machine position and game time;
The related person search method characterized by comprising. The related person search method according to claim 6, wherein in the person relevance score calculation step,
The related person search method characterized in that the person relevance score is calculated for each combination of IDs and recorded in the registration information database as a person relevance score management table. In the related person search method according to claim 6 or 7, in the person relevance score calculation step,
The following formula 1,

However,
z: score representing the relationship between persons x: time for game overlap y: distance between game tables a: coefficient for determining weight of time for game overlap b: for determining weight of distance between game tables The coefficient of
The related person search method characterized by calculating | requiring the said relevance score using. The related person search method according to any one of claims 6 to 8,
A person ID input step for inputting an ID of a person for which a related person is to be searched;
A related person search step of searching for a person highly relevant to the ID input from the person ID input step using the relevance score;
A search result display step for displaying a search result in the related person search step;
A related person search method comprising: The related person search method according to claim 9,
In the related person search step, relevance scores between the input ID and all IDs registered in the registration information database are extracted, the extracted score group is sorted in descending order, A related person search method, wherein the search is performed using an ID of a person having a top m (m is an arbitrary natural number) name. In a related person search program for extracting a person highly related to a certain person,
A registration information database that stores information that combines a person ID, which is an ID for identifying a person, and face image data of the person, and a management table that manages game times and game machine positions;
Face image acquisition means for acquiring image information captured by the image capturing device;
A face area detecting means for checking whether or not a person's face is reflected in the acquired image, and detecting the face area in the image when the face is reflected;
A face image collating means for identifying a face image similar to the detected face image from the registered information database to identify which ID of the person of the face image is assigned;
Game position detection means for determining which gaming machine the person of the identified face image is playing, and storing the determination result in association with the ID in the registration information database;
Game time recording means for calculating a game time based on the time when the face is authenticated, and storing the calculation result in association with the ID in the registration information database;
A person relevance score calculating means for obtaining a relevance score between players from the obtained person ID, gaming machine position and game time;
A related person search program characterized by causing a computer to function as a related person search server. The related person search program according to claim 11, wherein the person relevance score calculating means includes:
The related person search program characterized in that the person relevance score is calculated for each combination of IDs and recorded in the registration information database as a person relevance score management table. The related person search program according to claim 11 or 12, wherein the person relevance score calculating means includes:
The following formula 1,

However,
z: score representing the relationship between persons x: time for game overlap y: distance between game tables a: coefficient for determining weight of time for game overlap b: for determining weight of distance between game tables The coefficient of
The related person search program characterized by calculating | requiring the said relevance score using. In the related person search program according to any one of claims 11 to 13,
A person ID input means for inputting the ID of a person who wants to search for related persons;
Search result display means for displaying the search results;
Related person search further comprising related person search means for searching for a person highly relevant to the ID input from the person ID input means using the relevance score and notifying the search result display section of the search result. A related person search program characterized by causing a computer to function as a server.   15. The related person search program according to claim 14, wherein the related person search means extracts and extracts relevance scores between the input ID and all IDs registered in the registration information database. A related person search program characterized by sorting score groups in descending order and notifying the search result display section of IDs of the top m (m is an arbitrary natural number) names of the sorted scores.

Images