System and method for automated table game activity recognition

The system uses image and depth sensors with computing devices for real-time analysis to accurately monitor table games, addressing installation and accuracy issues, and providing automated data on game events and player behavior.

JP2024156762AInactive Publication Date: 2024-11-06ANGEL GRP CO LTD
View PDF 12 Cites 0 Cited by

Patent Information

Application Number
JP2024121001
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Priority Date
2016-05-16
Filing Date
2024-07-26
Publication Date
2024-11-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing systems for monitoring table games in gambling establishments are expensive to install and maintain, and they lack the accuracy needed to effectively track game events and player behavior due to environmental noise, cluttered tables, and varying lighting conditions.

Method used

A system utilizing image and depth sensors to capture game events, combined with computing devices for real-time analysis, enabling automatic recognition of game states, chip detection, and wager prediction through image processing techniques.

Benefits of technology

Enables accurate, automated monitoring of table games without human intervention, providing real-time data on game events, chip values, and player behavior, enhancing operational efficiency and strategic planning for casinos.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024156762000001_ABST
    Figure 2024156762000001_ABST
Patent Text Reader

Abstract

To provide a system, a method and a computer-readable medium for automated gaming recognition.SOLUTION: A system according to some embodiments comprises: at least one image sensor configured to capture image frames of a field of view including a table game; at least one depth sensor configured to capture depth of field images of the field of view; and a computing device configured to receive the image frames and the depth of field images, and configured to process the received image frames and depth of field images in order to produce an automated recognition of at least one gaming state appearing in the field of view. Further embodiments relate to methods and systems for monitoring game play and / or gaming events on a gaming table.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] FIELD OF THE DISCLOSURE

[0001] The described embodiments relate generally to monitoring table games. In particular, the embodiments relate to systems and methods for monitoring events within table games at a gaming establishment. [Background technology]

[0002] Casinos and other similar venues currently use monitoring technology and other management software to monitor players and plan their business strategies. These venues aim to deploy real-time behavioral analysis, algorithms (or processes), and player tracking technology to maximize player revenue, optimize staffing, and optimize venue floor space allocation to game types that maximize venue revenue. Most casino patrons participate in loyalty programs that require them to use player cards instead of coins, bills, or tickets. This has given casinos the opportunity to record and analyze individual gambling behavior, create player profiles, and record things like the amount each bettor bets, the bettor's wins and losses, and the rate at which bettors press slot machine buttons. However, table games are not as easy to monitor as either slot machines or button-operated gaming machines.

[0003]

[0003] Systems for monitoring and managing table games have typically proven to be expensive to install and maintain, and fail to achieve the level of accuracy required to be truly beneficial. Other options include sensors in casino chips and other offline revenue management solutions, but these have been shown to be ineffective. The operating environment of a gaming venue is fast-paced, with many visual and auditory noises and disturbances, cards and betting chips may be cluttered on the table, and lighting may vary significantly.

[0004]

[0004] Casinos or other such gambling establishments conduct a number of table-based games, such as baccarat, blackjack, roulette, etc., that involve players placing bets on the occurrence or non-occurrence of certain events. Each game has its own set of defined events that start the game, determine the outcome of bets placed during the game, or end the game. Most games are conducted by a designated dealer who undertakes specific actions for each game that can start the game, trigger events that determine the outcome of bets placed, or end the game.

[0005] Casinos and other such gambling establishments are interested in reviewing transaction data associated with events occurring on a gaming table or playing surface. This information can assist in planning the casino's business strategy and in monitoring player behavior. Information regarding the events and outcomes of games on the table can form the basis for the casino to review optimal staffing, floor space allocation for specific games, and other such revenue-enhancing or patron experience-enhancing decisions. One method of reviewing transaction data associated with events occurring on a gaming table used by casinos is random sampling, where individuals visually inspect events occurring at a portion of the table and report their observed information. The reported information can be extrapolated to predict the overall level of activity on the tables in the casino. However, because such visual inspections occur at intervals of an hour or more and rely on human judgment, inefficiencies may exist in such methods.

[0006]

[0006] Systems for monitoring and managing table games have typically proven to be expensive to install and maintain, and fail to achieve the level of accuracy required to be truly beneficial. Other options include sensors in casino chips and other offline revenue management solutions, but these have been shown to be ineffective. The operating environment of a gaming venue is fast-paced, with many visual and auditory noises and disturbances, cards and betting chips may be cluttered on the table, and lighting may vary significantly. Summary of the Invention [Problem to be solved by the invention]

[0007] It would be desirable to overcome or ameliorate one or more shortcomings or drawbacks associated with the prior art for monitoring events within table games in gaming establishments, or at least provide a useful alternative.

[0008]

[0008] Throughout this specification, the term "comprise", or variations such as "comprises" or "comprising", shall be understood to imply the inclusion of stated elements, integers, or steps, or groups of elements, integers, or steps, but not the exclusion of any other elements, integers, or steps, or groups of elements, integers, or steps.

[0009]

[0009] In this specification, a statement that an element may be "at least one of" a list of options is understood to mean that the element may be any one of the listed options or any combination of two or more of the listed options.

[0010]

[0010] Any discussion of documents, statutes, materials, devices, articles or the like contained in this specification shall not be construed as an admission that any or all of such matters form part of the prior art base or that they are common general knowledge in the field relevant to this disclosure by virtue of their existence prior to the priority date of any claim in this application. [Means for solving the problem]

[0011]

[0011] According to a first aspect, some embodiments provide a system for automatic game recognition comprising at least one image sensor configured to capture image frames of a field of view including a table game, at least one depth sensor configured to capture a depth of a field of view image of the field of view, and a computing device configured to receive the image frames and the depth of the field of view image, and configured to process the received image frames and the depth of the field of view image to generate automatic recognition of at least one game state appearing within the field of view.

[0012]

[0012] According to a second aspect, some embodiments provide a method for automatic game recognition, the method including acquiring image frames of a field of view including a table game, acquiring a field of view image depth of the field of view, and processing the received image frames and the field of view image depth to generate automatic recognition of at least one game state appearing within the field of view.

[0013]

[0013] According to a further aspect, some embodiments provide a non-transitory computer-readable medium for automatic game recognition that includes instructions that, when executed by one or more processors, cause the following to be performed: acquiring image frames of a field of view including a table game; acquiring a field of view image depth of the field of view; and processing the received image frames and the field of view image depth to generate automatic recognition of at least one game state appearing within the field of view.

[0014]

[0014] The image frame may include an image in or comprising the visible spectrum, or may include an infrared or ultraviolet image. The field of view image depth may include time-of-flight data points of the field of view and / or phase information data points reflecting the depth of the field. The at least one game state appearing in the field of view may include one or more or all of a game start, chip detection, chip value prediction, chip stack height prediction, and game end. The game start and / or game end may be effected by card detection or dolly detection. The table game may be a card game such as poker, blackjack, or baccarat, or a non-card based game such as roulette.

[0015]

[0015] Some embodiments relate to a method of monitoring game play on a table surface of a game table, the method comprising: analyzing captured images of the table surface in real time to identify the presence of a game object in any one of a first plurality of desired predefined areas on the table surface; recording a timestamp of a game event in response to the game object being identified as being present in any one of the first plurality of desired predefined areas; and transmitting game event data to a server, the game event data including the timestamp, an indication of the game event, and an identifier of the any one of the first plurality of desired predefined areas.

[0016]

[0016] The game object may be a game card or a position marker. The analyzing may include identifying a presence of at least one betting object on the table surface. The at least one betting object may be different from the game object. The presence of the at least one betting object may be identified in one or more of the plurality of second predefined regions of interest. The analyzing may include identifying one or more groups of betting objects within the one or more second predefined regions of interest.

[0017]

[0017] The analyzing may further include predicting a height of each of the one or more groups of betting objects relative to the table surface and predicting a number of betting objects present in each group of betting objects. The analyzing may further include identifying a color of the top betting object in each group of betting objects.

[0018]

[0018] The method may further include automatically predicting a bet amount associated with each second predefined area of ​​interest in which the presence of at least one bet object is identified, the prediction being based on the identified color of the top bet object in each group of bet objects and the predicted number of bet objects in each group of bet objects within each second area of ​​interest.

[0019]

[0019] The captured image may include a multispectral image, and the analyzing may further include multi-frame processing of the multispectral image to identify the presence of a game object in any one of a plurality of first predefined areas of interest or second predefined areas of interest on the table surface.

[0020]

[0020] Some embodiments relate to a system for monitoring game play on a table surface of a game table, the system comprising at least one camera configured to capture images of the table surface and a computing device in communication with the camera and configured to analyze the captured images of the table surface in real time to automatically identify the presence of a game object in any one of a plurality of first predefined areas of interest on the table surface.

[0021]

[0021] The game object may be, for example, a game card or a position marker. The computing device may be configured to identify the presence of at least one of the betting objects on the table surface. At least one of the betting objects may be different from the game object. The presence of the at least one betting object is identified by the computing device in one or more of the plurality of second predefined areas of interest. The computing device may be configured to identify one or more groups of the betting objects in the one or more second predefined areas of interest.

[0022]

[0022] The at least one camera may further comprise a depth sensing device for communicating depth data of the gaming objects relative to the table surface to the computing device. The computing device may be further configured to predict a height of each of the one or more groups of betting objects relative to the table surface. The computing device may be further configured to predict a number of betting objects present in each group of betting objects. The computing device may be further configured to identify a color of the top betting object of each group of betting objects.

[0023]

[0023] The computing device may be configured to automatically predict a bet amount associated with each second predefined area of ​​interest in which the presence of at least one bet object is identified, the prediction being based on the identified color of the top bet object in each group of bet objects and the predicted number of bet objects in each group of bet objects within the respective second area of ​​interest.

[0024]

[0024] The captured image may include a multispectral image, and the computing device may be configured to perform multi-frame processing of the multispectral image to identify the presence of a game object in any one of a plurality of first predefined areas of interest or second predefined areas of interest on the table surface.

[0025]

[0025] Some embodiments relate to a system for automatically monitoring game events on a game table comprising: a depth imaging device configured to capture depths of a plurality of game objects on a game area of ​​the game table; a plurality of visual imaging cameras configured to capture visual images of the game area; a game configuration module including configuration data associated with a plurality of games, a configuration of the game table, locations of target areas, patterns for recognizing game objects, and definitions of game events as changes in states of game objects on the game table; and a computer system configured to receive data from the depth imaging device and the plurality of visual imaging cameras and access the configuration data in the game configuration module to automatically recognize objects on the game area and game events occurring during the games.

[0026]

[0026] The methods described herein may be fully automated, so that gaming behavior monitoring may occur without any need for human judgment or intervention. However, some human interaction may occur within the system configuration process, such as establishing areas of interest for betting and for locating gaming objects such as cards or dollies. [Brief description of the drawings]

[0027] [Figure 1]

[0027] FIG. 1 is a block diagram of a game monitoring system. [Diagram 2]

[0028] FIG. 2 is a schematic diagram of a system for automatic table game recognition forming part of the game monitoring system of FIG. 1. [Diagram 3]

[0029] 2 is an image of a game table surface that may form part of the gaming environment of the system of FIG. 1. [Figure 4]

[0030] 2 is an image of another game table surface that may form part of the gaming environment of the system of FIG. 1. [Diagram 5]

[0031] 2 is an image of another game table surface that may form part of the gaming environment of the system of FIG. 1. [Figure 6]

[0032] 2 is an image of another game table surface that may form part of the gaming environment of the system of FIG. 1. [Figure 7]

[0033] FIG. 2 is a block diagram of a depth sensing device and camera for use in the system of FIG. 1. [Figure 8]

[0034] FIG. 2 is a front view of the inside of a housing for a depth sensing device and camera according to some embodiments. [Figure 9]

[0035] FIG. 2 is a block diagram of a computing device of the system of FIG. 1. [Figure 10]

[0036] FIG. 2 is a block diagram of a message broker server of the system of FIG. 1. [Figure 11]

[0037] FIG. 2 is a block diagram of a database server of the system of FIG. 1. [Figure 12]

[0038] FIG. 2 is a block diagram of a web application server of the system of FIG. 1. [Figure 13]

[0039] 2 is a screenshot of a web application showing an interface for managing the configuration of another game table that may form part of the gaming environment of the system of FIG. 1. [Figure 14]

[0040] 2 is another screenshot of a web application showing an interface for managing the configuration of another game table that may form part of the gaming environment of the system of FIG. 1. [Figure 15]

[0041] 2 is another screenshot of a web application showing an interface for managing the configuration of another game table that may form part of the gaming environment of the system of FIG. 1. [Figure 16]

[0042] 1 is a set of images to illustrate the results of several types of thresholding operations on sample images. [Figure 17]

[0043] 11 is a set of images to illustrate the results of further types of thresholding operations on sample images. [Figure 18]

[0044] 1 is a set of images to illustrate the results of erosion and dilation operations on a sample image. [Figure 19]

[0045] 1 is a flow chart of an edge detection process. [Figure 20]

[0046] FIG. 20 is a diagram for illustrating edge detection in part of the process of FIG. 19. [Figure 21]

[0047] 20 is an exemplary graph to illustrate an exemplary criterion applied to the edge detection process of FIG. 19; [Figure 22]

[0048] 1 is a set of images to illustrate the application of contour detection to sample images with different parameters. [Figure 23(a)]

[0049] 1 is a plot of a set of points to which a planar prediction process is applied. [Figure 23(b)]

[0050] 23(b) is a plot of the results of applying a plane prediction process and the orthogonal distance of the predicted plane to the points shown in the plot of FIG. 23(a). [Figure 24]

[0051] 1 is a flow chart of a game monitoring system according to some embodiments. [Diagram 25]

[0052] 11 is a flowchart of a game monitoring system according to a further embodiment. [Figure 26(a)]

[0053] 11 is an image frame illustrating the application of several card detection processes to another gaming table; [Figure 26(b)] 11 is an image frame illustrating the application of several card detection processes to another gaming table; [Figure 27(a)]

[0054] 1 is an image frame of another gaming table to which the card and chip detection process may be applied; [Figure 27(b)]

[0055] 27(b) is an image frame of the game table of FIG. 27(a) obtained by applying a thresholding technique to the image frame of FIG. 27(a); [Figure 28(a)]

[0056] 11A-11C are image frames obtained by an infrared camera illustrating the application of several card and chip detection processes to another gaming table. [Figure 28(b)] 11A-11C are image frames obtained by an infrared camera illustrating the application of several card and chip detection processes to another gaming table. [Figure 29(a)]

[0057] 1 is an image frame that can be input to the chip detection process. [Figure 29(b)]

[0058] 29(b) is an image frame obtained by applying a binary thresholding operation to the image frame of FIG. 29(a). [Figure 29(c)]

[0059] 29(b) is an image frame obtained by applying an erosion operation to the image frame of FIG. 29(b). [Figure 29(d)]

[0060] 29(c) is an image frame obtained by applying a dilation operation to the image frame of FIG. [Figure 29(e)]

[0061] 29(b) is an image frame illustrating the result of applying a tip detection process to the input image frame of FIG. 29(a); [Diagram 30]

[0062] 1 is an image frame illustrating the results of applying a chip detection process to a gaming table where a portion of the field of view of the gaming table is obstructed. [Diagram 31]

[0063] 30(b) is an image frame illustrating the results of application of a chip detection process on the gaming table of FIG. 30(a) taken after the obstruction in FIG. 30(a) has disappeared within the image frame; [Diagram 32]

[0064] 11 is an image frame illustrating the results of applying a chip detection process to a gaming table, where the input image frame is based on an image captured by a visual image camera; [Diagram 33]

[0065] 33 is an image frame illustrating the results of applying a chip detection process to the gaming table of FIG. 32, where the input image frame is based on an image captured by an infrared camera. [Diagram 34]

[0066] 1 is a flowchart of a multi-frame processing technique in accordance with some embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0028]

[0067] FIELD OF THE DISCLOSURE The described embodiments relate generally to monitoring table games. In particular, the embodiments relate to systems and methods for monitoring events within table games at a gaming establishment.

[0029]

[0068] Game monitoring system: FIG. 1 is a block diagram of a game monitoring system 100 according to some embodiments. The system 100 may comprise a number of game monitoring configurations 105, a game monitoring infrastructure 115, an administrator client 170, and a database client 180. The game monitoring configuration 105 includes a game environment 110, a depth-sensing device and camera 120, and a computing device 130. The system 100 is suitable for installation and operation in one or more game rooms of a gaming establishment, such as a casino. Each game room may have one or more game tables located therein, some or each of which may form part of the respective game monitoring configuration 105. Commercially available devices such as Microsoft™ Kinect or Asus™ Xtion or Infineon™ 3D image sensor REAL3™, other similar depth-sensing devices with camera capabilities may be used as the depth-sensing device and camera, for example. The depth sensing device and camera 120 are coupled or connected to a computing device 130 to receive instructions from and transmit recorded data to the computing device 130 using a link 107. For example, a Microsoft™ Kinect device may be connected to the computing device using a USB port on the computing device.

[0030]

[0069] A gaming house may have multiple gaming environments, e.g., areas or rooms where table games are played, and multiple game monitoring configurations 105 may exist to monitor each one of those gaming environments. Multiple game monitoring configurations 105 may be coupled or linked to a common game monitoring infrastructure 115 using network links 187. Network links 187 include network links 117 between computing devices 130 and message broker server 140 and network links 167 between database server 150 and computing devices 130. Game monitoring infrastructure 115 may also be coupled or linked to game monitoring configurations 105 in two or more different gaming houses. In some embodiments where a gaming house may have multiple gaming environments 110, multiple game monitoring infrastructures 115 may be coupled to different portions of game monitoring configurations 105 in the same venue.

[0031]

[0070] The game monitoring infrastructure 115 includes a message broker server 140, a database server 150, and a web application server 160. The message broker server 140 may be connected to a number of computing devices 130 via a bidirectional network link 117. A network link 127 may exist between the message broker server 140 and the database server 150 to enable the transfer of data or instructions. A network link 137 may exist between the web application server 160 and the database server 150 to enable the transfer of data or instructions. Each of the servers 140, 150, and 160 may be implemented as a standalone server, or as an apparent virtual server on one or more physical servers, or within a cloud computing service. Each of the servers 140, 150, and 160 may also be implemented through a network of two or more servers configured to handle improved performance or high availability requirements.

[0032]

[0071] Administrator client 170 may be an end-user computing device such as a computer or tablet and may be connected, for example, to web application server 160 via network link 147. Database client 180 may be an end-user computing device or interface for relaying data to other end-user computing devices or other databases and may be connected to database server 150 via network link 157.

[0033]

[0072] Game Environment: The configuration of the game environment 110 may vary depending on the particular game being played, but most games monitored by any one of the present embodiments have common elements. Figure 2 illustrates a system 200 for automatic table beam recognition according to some embodiments. The main functions of the system of the presently described embodiment of the invention are to detect when a game starts and ends, to detect the location of placed chips, and to predict the value and height of the chip stack (how many chips there are). The system is based on a combination of image processing techniques and sensing device capabilities.

[0034]

[0073] The gaming environment 110 includes a playing surface or game table 210 on which a game is played. The playing surface 210 is generally a substantially horizontal, planar surface on which various game objects, such as cards 211 or chips 213 or other objects may be placed that may be detected by the game monitoring system 100. The depth sensing device and camera 120 may be mounted on a post or pole 220 at a height that positions the depth sensing device and camera 120 above any obstructions in the field of view of the depth sensing device, and may be angled to point the field of view of the depth sensing device and camera 120 somewhat downward toward the game table 210. The obstructions may be, for example, a dealer conducting a game at the table, or a participant of the game, or a temporary obstruction such as a passerby. In some embodiments, the depth sensing device or camera 120 may be positioned behind and above the dealer's shoulder to obtain a view of the playing surface of the game table 210. The position of the depth sensing device or camera 120 and the computing device 130 may be above or adjacent to other display screens on a pillar or pole located on the game table 210.

[0035]

[0074] FIG. 3 is an image 300 illustrating a portion of a playing surface of a game table configured for the game of blackjack. The playing surface or game table includes a number of predefined regions of interest and may have specific orientations and functions with respect to the operation of the game depending on the characteristics of the game. The predefined regions of interest may be designated for the detection of specific game objects such as game cards or betting objects. For example, in FIG. 3, a first predefined region of interest 305 may be designated for locating cards 211 dealt to a player, and a second predefined region of interest 308 may be designated for locating chips or betting objects 213 on which a player may bet in a game. In some embodiments, one or more of the predefined regions of interest may overlap with one or more of the other predefined regions of interest. In some embodiments, one predefined region of interest may form a portion of another predefined region of interest.

[0036]

[0075] Participants in a game include players who may place bets and dealers who conduct the game. To place bets or conduct a game, objects described as game objects are used by players or dealers. Game objects may include cards 211 in a particular shape with specific indicia for identification, chips or betting objects 213, or other such objects that may specify the amount a player may bet in a game, or may include position markers or other objects with obvious shapes that may specify the outcome of a game, such as dollies used in a game of roulette. A game is conducted through a series of game events including the start of a game, the placing of bets by players during the game, intermediate results during the game, and the end of the game, which determines the final outcome of the game. During a game, players may place bets by placing their betting objects (i.e., betting tokens or chips) within a designated target area for placing bets. For example, in a game of blackjack as shown in FIG. 3, a player may place bets during a game by placing one or more betting objects, such as chips 213, in a designated area 308 for placing bets. Chips or betting objects may be placed in groups or stacks within a desired area. Often, groups or stacks of betting objects include betting objects of a common color.

[0037]

[0076] In a particular game, a player may select a target area that is associated with a winning and payout probability associated with a bet placed. For example, the playing surface 210 of FIG. 6 is a playing surface having markings for betting areas for a game of roulette. Different target areas 308 on the playing surface 600 may have different winning and payout probability for a player's bet. The playing surface may have several different configurations with respect to the location and structure of the various target areas depending on the different rules and / or betting conventions associated with the game. FIG. 4 is an image 400 of a game table designated for a game of baccarat. FIG. 5 is an image 500 of a game table designated for another game according to some embodiments.

[0038]

[0077] Depth sensing device and camera: The depth sensing device and camera 120 performs the function of capturing visual images and depth information of the field of view in front of the device. The device 120 is placed in front of a playing surface or a game table to capture all the designated areas of interest identified on the game table. The device 120 includes an infrared projector 710, an infrared sensor 720, a camera 730, a processor 740, a communication port 750, and an internal data link 705 that connects the infrared sensor 720 and the camera 730 with the processor 740. The internal data link 706 connects the processor 740 with the communication port 750. The depth sensing device and camera 120 can capture images from multiple spectrums of light visible to the human eye and / or invisible to the human eye. For example, the depth sensing device and camera 120 can capture images from the visible light spectrum through the camera 730 and from the infrared spectrum through the infrared sensor 720, and can consequently operate as a multispectral camera.

[0039]

[0078] The depth sensing device and camera 120 may rely on time-of-flight technology to sense the depth or scene of the field of view in front of it. The infrared projector 710 may project pulsed light or dimmed light in a continuous wave, which may be a sine wave or a square wave. To improve the accuracy of the depth information, multiple phases of the projected light may be projected and sensed. The measured phase shift between the light pulse emitted by the infrared projector 710 and the reflected pulse sensed by the infrared sensor 720 is relied upon by the processor 740 to calculate the depth of the field in front of the device. In some embodiments, the depth sensing device and camera 120 may include a structured light 3D scanner that relies on the principle of using a projected light pattern and a camera for depth sensing. In some embodiments, the depth sensing device and camera 120 may include a stereo camera that performs the function of depth sensing by using images from two or more lenses and corresponding image sensors. The depth sensing device and camera 120 may rely on other alternative means of depth or range sensing, or a combination of two or more techniques, to obtain depth information of the game environment, and specifically the table playing surface 210.

[0040]

[0079] The sensed depth information is combined with pixel grid information determined by the processor and presented as output to the communication port 750. The pixel grid information is also combined with the visual image captured by the camera 730 and combined with the depth information and presented as output through the port 750. Apart from the depth information, the infrared sensor 720 also senses the intensity of infrared light reflected by the field of view, and this information is combined with the pixel grid information by the processor 740 and passed to the port 750. The port 750 may be in the form of a physical port, such as a USB port, or a wireless transmitter, such as a wireless network adapter. The depth sensing device and camera 120 return sensed depth, color, and infrared imaging data in different coordinate spaces that may be mapped with each other to obtain integrated depth, color, and infrared data associated with a particular area or point in the field of view of the device. FIG. 8 is a front view of the inside of a depth sensing device and camera housing according to one embodiment 800, illustrating some of its components including an infrared projector 710, an infrared sensor 720, and a camera 730.

[0041]

[0080] Computing Device: Data generated by the depth sensing device and camera 120 is received by the computing device 130 via a communication port 990. The port 990 may be in the form of a USB port or a wireless adapter that interfaces with the communication port 750 to receive sensor data or send instructions to the depth sensing device and camera 120. The hardware components 910 of the computing device 130 include a memory 914, a processor 912, and other components necessary for the operation of the computing device. The memory 914 stores necessary software modules 920, which include an image processing library 922, a depth sensing device and camera API 924, a runtime environment driver 926, a game monitoring module 928, a batch script 930, scheduled jobs 932, and a message generator module 934.

[0042]

[0081] The image processing library 922 is a set of programs for performing basic image processing operations, such as performing thresholding operations, morphological operations on images, and other programs necessary for the image processing steps performed by the game monitoring module 928. OpenCV is an example of an image processing library that may be used. The depth sensing device and camera API 924 is a set of programs that allow the computing device 930 to establish a communication channel with one or more depth sensing devices and cameras 920. For example, if a Microsoft™ Kinect™ device is used as the depth sensing device and camera, the Kinect for Windows™ SDK is used as the depth sensing device and camera API 924. This API 924 allows the computing device 130 to query the depth sensing device and camera 120 in the appropriate protocol and understand the format of the returned results. This API 924 allows the data generated by the depth sensing device and camera 120 to be received and processed by the game monitoring module 928. The computing device 130 also includes the necessary runtime environment drivers 926 to provide the necessary dependencies for the execution of the game monitoring module 928. The game monitoring module 928 includes programs that monitor game events that occur during the course of a game.

[0043]

[0082] The software module 920 also includes batch scripts, which may be in the form of Windows PowerShell scripts or scripts in other scripting languages, to perform necessary system administration and maintenance tasks for the game monitoring module 928. The batch scripts 930 may be executed on a scheduled basis through scheduled jobs 932, which may be in the form of Windows Scheduler jobs or other similar job scheduling services. A message generator module 934, based on instructions from the game monitoring module 928, generates messages that are passed to the message broker server 140. The message generator module may be based on standard messaging systems, such as, for example, RabbitMQ or Kafka. Based on a stored message broker configuration 942 in the configuration module 940, the message generator module 934 may communicate messages to the message broker server 140 via the communication port 990 and the network link 117. The configuration module 940 also includes a table configuration 942 and a game start and end trigger configuration 944. The components of the configuration module 940 are stored in the memory 914 in the form of one or more configuration files. The configuration files may be stored, for example, in an XML format.

[0044]

[0083] Message Broker Server: The message broker server 140 implements a message broker service and waits for messages from the multiple computing devices 130 via the network link 117. The message broker server 140 may be located on the same premises as the computing devices 130 in a common local network, or it may be located outside the premises (remotely), but still in communication via the network link 117 established between the two premises to enable the transfer of messages and data. The message broker server 140 may be centralized and connected to the computing devices 130 in multiple gaming venues to provide a centralized message broker service. The message broker server 140 has a hardware component 1010 with a memory 1014, a processor 1012, and other necessary hardware components for the operation of the server. The message queue module 1020 implements queues for receiving, interpreting, and processing messages from the multiple constituent devices 130. Messages are received via a communication port 1090, which may be in the form of a network adapter or other similar port that may enable the two-way transfer of data and instructions to or from the message broker server 140. The message queue module 1020 may be implemented through a message broker package such as RabbitMQ or Kafka. When the message queue module 1020 receives a message containing transaction information regarding a game event occurring on a game table, it initiates the database parsing module 1030. The database parsing module 1030 parses the message received by the message queue module 1020 into a database query that is subsequently executed on the database server 150 via the network link 127.

[0045]

[0084] Database Server: The database server 150 serves the purpose of receiving game event data from the message broker server 140, storing table configuration data managed via the web application server 160, and acting as a repository for the database clients 180 to provide access to the game event data captured by the game monitoring system 100. The database server 150 has a hardware component 1110 with a memory 1114, a processor 1112, and other necessary hardware components for the operation of the server. The communication port 1190 may be in the form of a network adapter or other similar port that can enable bidirectional transfer of data and instructions to or from the database server 150 over one or more network links. The database module 1120 may be implemented through a database management system such as MySQL™, Postgres, or Microsoft™ SQL Server.

[0046]

[0085] The database module 1120 maintains data including table configuration data 1122 and game event data 1124. The game event data 1124 includes transaction data that represents game events occurring on a game table or playing surface. Records forming the game event data may include a timestamp of the time the game event was recognized, a unique identifier for the game table at which the game event occurred, an identifier for characteristics of the game event such as placing a bet, an intermediate outcome of the game, a final outcome of the game, an identifier of a destination area associated with the game event, a prediction of the bet value associated with the destination area, and other related attributes that represent the game event.

[0047]

[0086] The table configuration data 1122 includes a unique identifier for the game table and associated computing device 130, a location of a target area 308 on the game table in the form of a polygon and coordinates of pixels associated with the depth sensing device and camera 120 forming the endpoints of the polygon, characteristics of the target area 308, where the target area 308 is an area for placing cards or chips or a specific game object to be detected, characteristics of game start and end trigger events, where the start of a game is detected by placing a card in the target area or a specific game object in a specific target area, model contours of game objects such as cards or chips to enable detection by, for example, the game monitoring module 928, and other associated data necessary to represent the parameters relied upon by the game monitoring system 100. In some embodiments, the table configuration data 1122 and the game event data 1124 may be maintained in separate database servers to allow for greater scalability and manageability of the game monitoring system 100.

[0048]

[0087] The database server 150 also includes a table configuration transfer module 1140 that performs the function of transferring the table configuration data 1122 to the respective computing devices 130. The table configuration transfer module may be implemented through a combination of database scripts and command line scripts that first generate the table configurations 942, the game start and end trigger configurations 944, and the message broker configurations 946 in the form of configuration files, such as, for example, XML. The generated configuration files may be transferred to the respective computing devices 130 via a communication port 1190 that responds over a network link 167. The network link 167 may be a local network link, if the database server 150 and the computing devices 130 are in the same local network, or may be a network link that spans multiple computer networks, if the database server 150 and the computing devices 130 are located in separate networks. The transfer of the configuration files may be accomplished through an appropriate network protocol, such as, for example, the File Transfer Protocol or the SSH Transfer Protocol.

[0049]

[0088] Web Application Server: The web application server 160 hosts web applications that facilitate the configuration and management of the table configuration data 1122 on the database server 150. The web application server 160 has a hardware component 1210 with a memory 1214, a processor 1212, and other necessary hardware components for the operation of the server. The communication port 1290 can be in the form of a network adapter or other similar port that can enable bidirectional transfer of data and instructions to or from the web application server 160 over one or more network links. The web application server includes a web application module 1220 with a web interface that allows a user to create and update the table configuration data 1122 on the database server 150. The web application can be implemented through a web application framework, such as, for example, Django in python or ASP.NET or other similar web frameworks. The web application server 160 also includes a database parsing module 1230 that translates instructions received by the web application module 1220 via the web interface into specific database queries or instructions that create or update table configuration data 1122 to reflect operations performed by the administrator client 170. The database queries or instructions are executed on the database server 150 over a network link 137. The network link 137 may be a local area network link if the database server 150 and the web application server 150 are in a common network, or it may span multiple networks if the database server 150 and the web application server 150 are located in separate networks.

[0050]

[0089] Web Interface: Figure 13 is a screenshot 1300 of a web application showing an interface for managing the configuration of an embodiment of a game table that may form part of the gaming environment 110. Parameters and parameters that may be required to configure a game table may include a unique identifier for the table, such as the IP address of the associated computing device located in screen area 1310. A portion of an XML configuration file that may be communicated to the computing device 130 to codify the table configuration 942 is shown in screen area 1320. Button 1330 may be used to create records for additional game tables, and register button 1340 allows a user to register a new configuration.

[0051]

[0090] FIG. 14 is another screenshot 1400 of the web application showing another interface for managing configurations of an embodiment of a game table that may form part of the game environment 110. A deploy button 1410 may be clicked to deploy a set of saved configurations to the computing device 130 over the network link 167. A delete button 1424 may be clicked to delete any saved configurations. 1420 is a sample portion of another XML file that may be used to save and communicate configuration information to the computing device 130. Screen areas 1412, 1414, and 1416 represent depth, color, and infrared image streams from the depth sensing device and camera 130. Configuration details associated with the individual streams may be viewed by clicking button 1422. Configuration details may be deleted by clicking button 1418. Screen area 1440 allows the user to set a default configuration for all tables that may be saved by clicking save button 1430.

[0052]

[0091] Button 1430 may be used to save changes to the game table configuration prior to deployment.

[0053]

[0092] FIG. 15 is another screenshot 1500 of a web application showing another interface for managing the configuration of an embodiment of a game table that may form part of the game environment 110. The interface shown in screenshot 1500 allows the type, location, and boundaries of an area of ​​interest to be defined using user interface tools. Such defined areas of interest then become "predefined areas of interest" as referred to herein once saved in the game configuration data. A button 1510 can be clicked to obtain a refreshed image of the game table if the position of the game table relative to the depth sensing device and camera 120 has changed. The image frame 1515 shown in screenshot 1500 includes one or more bounding areas of interest 1520, each bounding area of ​​interest being defined by a polygon 1525. Custom polygons can be drawn using selectable handles 1560 and added to a list of polygons 1530, for example. A save button 1540 allows the user to save changes made to the polygon, and a remap button 1550 allows the user to remap an existing polygon to a different location.

[0054]

[0093] In order to develop a system for automatic recognition of games, it is necessary to understand the behavior of the game. Two kinds of game tables are addressed here. One is card-based games or card games, which are any games that use playing cards as the main device by which the game is played, whether they are traditional or game-specific. Examples of this kind are blackjack, baccarat, and poker. Other kinds of table games are not card-based, such as roulette. In this game, players can choose to place a bet on either a single number or a range of numbers, red or black, or whether the number is odd or even. To determine the winning number and color, the dealer spins the wheel in one direction and then spins the ball in the opposite direction around a circular track around the circumference of the wheel.

[0055]

[0094] The behavior of the card-based game is as follows: · Players are free to make bets for a period of time. The dealer says "No more bets" and starts dealing cards to the players. This is when the game begins. After the game result is decided, the dealer collects the losing players' chips and distributes the winning chips. After that, the dealer puts away all the cards, which is the end of the game.

[0056]

[0095] The behavior of non-card based games, especially roulette. · Players will be placing bets for some time. The dealer spins the wheel, waits a moment until the ball is about to come to a stop, then announces "No more bets." This is when the game begins. After the ball stops, the dealer places the dolly on the table on the winning number. Winning / losing chips are allocated / collected by the dealer. After that, the game ends.

[0057]

[0096] Overall Monitoring Process: The game monitoring system 100 supports two fundamental aspects in its operation: contour detection and plane prediction. Contour detection includes a set or process or techniques that can be implemented in real time or near real time to recognize shapes in the images captured by the depth sensing device and camera 120. Near real time processing can include processing with a latency of a few seconds from the occurrence of an event, for example, 2-3 seconds or less. Plane prediction includes a set of processes or techniques that can be implemented in real time or near real time to predict the position of a plane representing a game table or playing surface based on the depth data and images captured by the depth sensing device and camera 120. When the process of plane prediction is implemented, the obtained plane position information can be combined with additional depth data captured by the depth sensing device and camera 120 to predict the height of a stack of game objects on a game table and make inferences about values ​​associated with the stack of game objects, such as a stack of chips.

[0058]

[0097] Image Pre-Processing Steps: Before contour detection or plane prediction techniques can be applied, several image pre-processing steps are applied to the images captured by the depth sensing device and camera 120. These image pre-processing steps improve the performance and accuracy of the process of implementing contour detection and plane prediction techniques.

[0059]

[0098] Thresholding: One image pre-processing technique that may be used is thresholding. The depth sensing device and camera 120 return data in color and infrared image frames of the game environment 110. Thresholding techniques may be applied to both the color and infrared data streams. Each pixel in a particular frame of either the color or infrared stream is represented by a numerical value that indicates the color or intensity of the pixel. The color image frames may be converted to grayscale image frames before performing the thresholding operation. Global thresholding is one way of performing thresholding. In global thresholding, each pixel value is compared to an arbitrary threshold, and if the pixel value is greater than this threshold, it is assigned a value corresponding to white, e.g., 255 in an 8-bit scale, otherwise it is assigned a value corresponding to black, e.g., 0 in an 8-bit scale. Through a series of images 1600, FIG. 16 illustrates an example of the results of thresholding on a sample image 1601. Image 1602 is the result of applying global thresholding to image 1601 using a value of 127, image 1601 being represented in 8-bit format. Global thresholding may not be sufficient in various practical applications: an image may have different lighting conditions in various parts of the image, and application of a global thresholding technique may reduce parts of the image that have low lighting conditions.

[0060]

[0099] Adaptive thresholding is an alternative to address the limitations of global thresholding. In adaptive thresholding, there are thresholds for different small regions of an image and are applied to those regions for thresholding purposes. The threshold for adaptive thresholding may be calculated by using the average value of the pixels in a pixel's neighborhood or a weighted sum of nearby pixels where the weights may be obtained from a Gaussian distribution. In FIG. 16, image 1603 is an example of the output of the application of an adaptive mean thresholding technique and image 1604 is an example of the output of the application of an adaptive Gaussian thresholding technique to the original image 1601. Another alternative to thresholding is Otsu's binarization. In this method, thresholding is performed based on the image histogram. Through a series of images 1700, FIG. 17 illustrates an example of the application of Otsu's binarization technique to a set of sample images 1701 with a representative histogram 1702. One or more of the alternative thresholding techniques may be applied to color or infrared image frames at a pre-processing stage. The image processing library 922 may provide a reusable library implementing the proposed thresholding techniques that may be invoked by the game monitoring module 928 during the image pre-processing stage.

[0061]

[0100] Morphological Transformations: Morphological transformations are operations based on image geometry. They are usually performed on binary images. They require two inputs, the first being the original image, and the second, called the structuring element or kernel, which determines the characteristics of the operation. The two basic morphological operators are erosion and dilation.

[0062]

[0101] Morphological transformations are performed on the images captured by the depth sensing device and camera 120 to enhance features to be detected in the image, as well as to improve the performance and accuracy of the contour detection process. Erosion and dilation are examples of morphological transformations that may be applied during the image pre-processing stage. Both the erosion and dilation processes require two inputs: image data in the form of a matrix captured by the depth sensing device and camera 120, and a structuring element or kernel that determines the characteristics of the morphological operation to be performed on the input image. The kernel may be square or circular in shape, has a defined center, and is applied as an operation by traversing through the input image.

[0063]

[0102] Erosion: The erosion morphological transformation involves the sharpening of foreground objects in an image by using a kernel such that as the kernel traverses through the image, the pixel's value is left at a value of 1 or a value corresponding to white only if all values ​​corresponding to the kernel are 1 or a value corresponding to white. Kernels of size 3×3 or 5×5 or other sizes may be used for the erosion operation. The erosion operation erodes the boundary of the foreground object. Through a series of images 1800, FIG. 18 illustrates an example of the application of the erosion and dilation operators. An example of the effect of the erosion operation on the input image 1801 may be seen in the erosion output image 1802. The erosion operation may be implemented by a predefined library in the image processing library 922. For example, if the OpenCV library is used, the function "erode" may be called by the game monitoring module 928 to work on the images captured by the depth sensing device and the camera 120.

[0064]

[0103] To achieve erosion, a kernel slides through the image (as in 2D convolution). A pixel in the original image (which is either a 1 or a 0) is considered to be 1 only if all the pixels under the kernel are 1, otherwise it is eroded (zeroed out).

[0065]

[0104] Dilation: A dilation operation is the inverse of erosion. For example, in a dilation operation using a 3×3 square matrix kernel, the pixel at the center of the kernel may be left with a value of 1 or a value corresponding to white if any one of the values ​​in the corresponding kernel is a value corresponding to 1 or white. An example of the effect of this operation on the input image 1802 may be seen in the eroded output image 1803. As a result of dilation, features in the image become more continuous and larger. The dilation operation may be implemented by a predefined library in the image processing library 922. For example, if the OpenCV library is used, the function "dilate" may be called by the game monitoring module 928 to operate on images captured by the depth sensing device and camera 120.

[0066]

[0105] Dilation is the exact opposite of erosion. Here, a pixel element is 1 if at least one pixel under the kernel is 1. So it increases the white area in the image or increases the size of the foreground object. Usually, erosion is followed by dilation, such as in denoising, because erosion removes white noise, but it also shrinks the object here. So we dilate it. Noise elements are removed by erosion, so they are not reintroduced by dilation, but the object area increases. It is also useful for joining corrupted parts of the object.

[0067]

[0106] The application of thresholding techniques to an image produces a binary image. To further enhance the features present in the image, erosion and dilation morphological transformations are applied. Advantageously, morphological transformations help reduce noise from the image, separate individual elements, and combine disparate elements within the image.

[0068]

[0107] An image contour comprises a curve that connects all the consecutive points along the boundary of an object represented in an image. Contours are useful tools for shape analysis and object detection and reconstruction. Contour approximation is used to approximate the likeness of a particular shape to that of a desired shape in an application. The desired shape may be in the form of, for example, a polygon or a circle or an ellipse. For better accuracy and performance, a contour detection operation may be performed on the binary image after an edge detection operation has been performed.

[0069]

[0108] Edge detection: Edge detection is an image processing technique for finding the boundaries of objects in an image. It works by detecting discontinuities in brightness. Among them, Canny is a popular multi-stage edge detection algorithm (or process) that can be described in the following steps:

[0070]

[0109] Edge detection may be performed by detecting brightness discontinuities between neighboring pixels and pixel clusters. Several image processing techniques may be used to perform this operation. Some embodiments implement a Canny edge detection operator or process to detect edges in the images captured by the depth sensing device and camera 120. FIG. 19 is a flow chart 1900 depicting a series of steps involved in implementing a Canny edge detection operator. Step 1910 involves preparing an image for use as input to the operator. This step may include applying a suitable thresholding technique to the image, and applying erosion or dilation to improve the performance of the remaining edge detection process. Step 1920 involves reducing undesired noise from the image. This may be achieved, for example, by application of a 5×5 Gaussian filtering kernel. This step smooths out features in the image and improves the performance of the remaining process. Examples of Gaussian filtering kernels that may be used are as follows:

[0071]

number

[0072]

[0110] Step 1930 involves the prediction of the luminance gradient of the image. To perform this operation, the input image is filtered by two Sobel kernels: the operation of the kernel Gx returns the first derivative of the image in the horizontal direction, and the kernel Gy returns the first derivative of the input image in the vertical direction. The kernels Gx and Gy that can be used are:

[0073]

[0111]

number

[0074]

[0112] Based on the first horizontal and vertical derivatives of the input image, the edge gradient G and direction θ for each pixel can be calculated as follows:

[0075]

[0113]

number

[0076]

[0114] The gradient direction is approximately perpendicular to the edge and may be rounded to one of four angles representing vertical, horizontal, and two diagonal directions.

[0077]

[0115] In step 1940, a complete scan of the edge strength image may be performed to remove any undesired pixels that may not constitute an edge or a desired edge. This is accomplished by checking whether each pixel is a local maximum within its neighborhood in the direction of its gradient. As illustrated in graph 2000 of FIG. 20, point A is perpendicular to an edge, and the gradient direction is perpendicular to the edge. Points B and C are located in the gradient direction, and therefore point A may be compared to points B and C to see if it forms a local maximum. If so, it is considered for the next stage 1950 in the process, otherwise it may be suppressed by assigning a pixel value of 0 to point A. The result is a binary image with pixels of value 1 corresponding to a thin edge and 0 corresponding to no edge.

[0078]

[0116] In step 1950, it is predicted which of the edges detected in the previous step are true positives, meaning those that are not false positives and are most likely to represent edges in the real world represented by the input image. To perform this operation, two thresholds may be defined: minVal and maxVal. Edges with intensity gradients greater than maxVal are considered to be definite edges, and those below minVal may be considered non-edges and discarded. Edges that fall within the two thresholds may be further classified as edges or non-edges by their connectivity properties. If they are connected to definite edge pixels, they are considered to form part of an edge. If not, they may be discarded as false positives. For example, in the graph 2100 of FIG. 21, edge A is above maxVal, so it may be considered a true positive. Edge C is below maxVal, but it is connected to edge A, so it may also be treated as a true positive edge and the entire curve may be considered valid. Edge B is above minVal and is in the same region as that of edge C, but is not connected to any true positive edges, so it may be treated as a false positive. The values ​​of minVal and maxVal are selected to achieve optimal results: for example, minVal may be set to a value between 20 and 60, and maxVal may be set to a value between 60 and 180. This stage may also remove noise in the form of small pixel clusters.

[0079]

[0117] 19 may be implemented through programs available in the image processing library 922. For example, if the OpenCV library is used, the "Canny" edge detection function call may be used. Other alternative methods of edge detection may also be utilized as alternatives to Canny edge detection to achieve the same result of identifying edges in the input image.

[0080]

[0118] Contour Detection: After an edge detection operator is applied to the input image to identify edges, a contour detection process may be applied to the result of the edge detection operation to approximate the similarity of shapes in the image to a particular model shape, such as a polygon or circle.

[0081]

[0119] A contour may be described as a curve that joins all consecutive points (along a boundary) that have the same color or intensity. Contours are useful tools for shape analysis and object detection and reconstruction. For better accuracy, it is suggested that a binary image is used as input to the contour detection algorithm (or process). Therefore, it is suggested that thresholding or Canny edge detection should be applied before finding the contour. In the application of the present invention, boundary tracing algorithms (or processes) are used for topological analysis of digitized binary images. These algorithms (or processes) determine the envelopment relationship between the boundaries of the binary image. Since the outer boundary and the hole boundary have a one-to-one correspondence to the connected components of pixels and holes, respectively, the present algorithms (or processes) result in a representation of a binary image from which some features can be extracted without reconstructing the image. The second boundary tracing algorithm (or process), which is a modified version of the first boundary tracing algorithm, tracks only the outermost boundaries (i.e., the outer boundaries that are not enclosed by holes).

[0082]

[0120] Depending on the accuracy specified in the present invention, contour approximation is also performed to approximate the contour shape to another shape (polygon) with a smaller number of vertices, which can be implemented through the Douglas-Peucker algorithm as follows: function DouglasPeucker(PointList[],epsilon) / / Find the point with the maximum distance dmax=0 index=0 end=length(PointList) for i=2 to(end-1){ d=perpendicularDistance(PointList[i], Line(PointList[1],PointList[end])) if(d>dmax){ index=i dmax=d } } / / If max distance is greater than epsilon, / / recursively simplify if(dmax>epsilon){ / / Recursive call recResults1[]=DouglasPeucker(PointList[1...index ], epsilon) recResults2[]=DouglasPeucker(PointList[index...end], epsilon) / / Build the result list ResultList[]={recResults1[1...length(recResults1)-1], recResults2[1...length(recResults2)]} }else{ ResultList[]={PointList[1],PointList[end]} } / / Return the result return ResultList[] end

[0083]

[0121] In Fig. 22, a series of images 2200 represent various stages of application of the Douglas-Peucker algorithm. Image 2210 is an original image that may be used as an input image. Line 2205 in image 2220 represents the fitted curve for a value of epsilon equal to 10% of the arc length. Line 2215 in image 2230 represents the fitted curve for a value of epsilon equal to 1% of the arc length.

[0084]

[0122] The contour prediction operation may be implemented by using pre-prepared functions in the image processing library 922 and calling them in the game monitoring module 928. For example, when OpenCV is used to implement the contour prediction process, for example, the functions "findContours" or "drawContours" or "approxPolyDP" may be called to implement this process.

[0085]

[0123] Plane Detection Process: As a precursor to predicting the height of a stack of chips on a game table or playing surface, the game monitoring system 100 predicts the location of the plane that includes the game table or playing surface. One way to predict the equation of the plane is by Principal Component Analysis (PCA). PCA minimizes the perpendicular distance from a set of data to a fitted model. This is a linear case of what is known as orthogonal regression or total least squares. For example, given two data vectors x and y, a line in the form of a linear equation with two variables can be predicted that minimizes the perpendicular distance from each of the points (xi, yi) to the line. More generally, an r-dimensional hyperplane can be fitted in a p-dimensional space, where r is less than p. The choice of r is equivalent to the choice of the number of components to keep in the PCA.

[0086]

[0124] The basic mathematical model of a plane can be formulated as follows: ax+by+cz+d=0

[0087]

[0125] The values ​​a, b, c, and d need to be predicted to minimize the distance from a selected point on the game table or playing surface. Based on the binary version of the image acquired during the chip detection stage, 100 or more points that are not detected as chips may be selected as input points for PCA. Since these selected points are in two-dimensional space, the depth information associated with the points is utilized to obtain coordinates in the two-dimensional space. The principle of the pinhole camera model is relied upon to convert the points in the two-dimensional space to points in three-dimensional space. Given a point in the two-dimensional space with coordinates (x,y), a depth value Z, and (Cx,Cy) as the x,y coordinates of the optical center of the depth sensing device and camera 120, the coordinates (X,Y,Z) of the same point in the three-dimensional space may be determined using the following equations:

[0088]

[0126]

number

[0089]

[0127] 23(a) and 23(b) illustrate the application of PCA to a set of points 2310 in a three-dimensional coordinate graph 2300. In the coordinate graph 2301 in FIG. 23(b), the application of PCA allows the identification of a place 2350 that minimizes the orthogonal distance 2320 between the point 2310 and a plane 2350. The plane prediction operation can be implemented by using pre-prepared functions in the image processing library 922 and calling them in the game monitoring module 928. For example, if OpenCV is used to implement the contour prediction process, the functions implemented by the class "cv::PCA" can be called, for example, to implement the plane prediction process.

[0090]

[0128] To predict the value of the chip, the conventional Euclidean distance of the image can be used to classify the chip. Chip template images are collected in advance for comparison purposes. Then, a k-nearest neighbor algorithm is used to assign the value of the chip. However, there are some chip types that have similar colors. To distinguish these types of chips, the present invention further utilizes the reflectance of chip types from the infrared image to classify them using the k-nearest neighbor algorithm.

[0091]

[0129] After the table plane is indicated, the distance from the center of each chip to the plane can be predicted to obtain the height of the chip stack. The number of chips can also be predicted based on this height by linear or non-linear mapping. Calculating the distance from a point to a plane is derived as follows: Given a plane in three-dimensional space, ax+by+cz+d=0 and the point x0 = (x0;y0;z0) and the normal vector to the plane is

[0092]

number

[0093]

number

[0094]

[0130] An algorithm (or process) for card detection may be implemented. There may be two types of games, for example card-based and non-card-based games. For card-based games, an algorithm (or process) in 1 may be used to detect cards and trigger a game start event. For roulette, dolly detection may be used to trigger game start and end events. When the game monitoring system detects a dolly (or position marker), it may infer that the game may have started a few seconds, for example 20-30 seconds, before the detection of the dolly. The dolly may stay in that position until it is removed by the dealer. Removal of the dolly after the initial detection may trigger a game end event. Due to the reflective characteristics of the dolly, infrared images may be used to detect it rather than color images.

[0095]

[0131] Algorithm 1 Card detection algorithm (process): 1: Procedure CardDection(img). Input grayscale image img. 2: Apply Canny edge detection to the image 3: Dilate the Canny output to remove possible holes between edge segments 4: Apply Canny edge detection to the image 5: Finding and approximating contours 6: Accept contours that meet the following criteria: The outline area must be within the area size of the card, e.g. between 40-70cm2 or between 50-60cm2 The contour must have four vertices after approximation The cosine of the angle between the bond edges should be small, for example between -0.2 and 0.2 or between -0.1 and 0.1.

[0096]

[0132] Algorithm 2 Dolly detection algorithm (process): 1: Procedure DollyDection(img). Input infrared image img. 2: Apply global thresholding to img 3: Erode the output to remove small objects. 4: If there is an object remaining that meets the size criteria, it is a detected dolly. 5. Chip Detection to Detect Chips,We use adaptive thresholding to segment chips from the background. The output binary is eroded and dilated to remove small objects. Then, all blobs that meet the size criteria are detected as chips. This algorithm (or process) can be applied to both color and infrared images. The following subsections are a review of the techniques used, and the layout details of the algorithm (or process) are explained in Algorithm 3.

[0097]

[0133] Algorithm 3 Chip detection algorithm (or process): 1: Procedure CardDection(img). Input grayscale image img. 2: Apply adaptive thresholding to img 3: Erode and dilate the output to remove small objects. 4: If an object remains that meets the size criteria, it is a detected chip.

[0098]

[0134] A particular criterion, such as a size criterion or a cosine of an angle criterion, may be stored in the configuration module 940 as part of a table configuration 942 .

[0099]

[0135] The tip height prediction algorithm can be seen in Algorithm 4:

[0100]

[0136] Algorithm 4 Chip stack height prediction algorithm (process): 1: Procedure ChipHeightEstimate(img). Input the depth image img. 2: Convert selected table points that are not chips to 3D coordinates 3: Fit these points to a plane called the table plane 4: Find the distance from each center of the chip stack to this plane to get the chip stack height. 5: Divide the height by the height of a single chip to predict the number of chips stacked

[0101]

[0137] 26(a) and 26(b) are image frames 2600 and 2650, respectively, of an embodiment of a game table in which card detection and chip detection are implemented. In FIG. 26(a), boundary 2620 represents the area of ​​the image frame in which cards are detected. Boundary 2630 represents the area of ​​the image frame in which chips are detected. Card 2610 is not detected by the card detection algorithm (or process) because it is not fully presented to the depth sensing device and camera 120 and is in the process of being dealt on the game table 210. In FIG. 26(b), a different card 211 than the card presented in FIG. 26(a) has been detected and is surrounded by boundary 2620. Chip 213 in FIG. 26(b) is in a different position than the chip in FIG. 26(a) and is detected using boundary 2630.

[0102]

[0138] FIG. 27(a) and FIG. 27(b) represent image frames 2700 and 2701, respectively, of an embodiment of a game table on which card detection and chip detection have been implemented. FIG. 27(b) is the result of application of any one of the binary thresholding techniques to FIG. 27(a). Shape 2710 in FIG. 27(b) may be a potential false positive detection of a card 211. These false positives may be eliminated by step 6 of algorithm 1, which compares the area of ​​a shape to the expected area of ​​a card. Similar shape 2720 may be eliminated as a false positive detection of a chip 213 by comparing the area of ​​shape 2720 to the expected area of ​​a chip 213. FIG. 28(a) represents an infrared image of an embodiment of a game table 2800 with game objects such as cards 211 and chips 213. In FIG. 28(b), chip 211 has been identified by chip detection algorithm 3 and is greyed out to indicate detection. The results of the application of the game object detection process to the color image frames may be combined with the infrared image frames to improve the accuracy of detection and chip stack height prediction.

[0103]

[0139] 29(a)-29(e) illustrate the application of a chip detection process to an image frame 2900 according to some embodiments. The image frame 2900 may be captured by a depth sensing device and camera 120. FIG. 29(b) is an image frame 2901 obtained by application of a binary thresholding operation to the input image frame 2900 of FIG. 29(a). Image frame 2902 of FIG. 29(c) is obtained by application of an erosion operation to the image frame 2901 of FIG. 29(b). FIG. 29(d) is an image frame 2903 obtained by application of a dilation operation to the input image frame 2902 of FIG. 29(c). The combination of the erosion and dilation operations helps to reduce noise from the input image frame and makes observable features such as chips more noticeable. FIG. 29(e) is an image frame 2904 illustrating the result of the chip detection process. The detected chip 213 is identified by its outline 2920. Shapes 2910 in Figure 29(d) are not identified as chips because they fail to meet the criteria as part of the process of chip detection, which may be the size range of shapes to be detected as chips, or other similar distinguishing criteria.

[0104]

[0140] Some embodiments follow the steps in the flowchart 2400 of FIG. 24 to monitor game events at a game table. A frame grabber 2410 collects three streams of frames, including color video frames, depth frames, and infrared frames. The raw color images have a resolution of 1920x1080. The depth stream provides depth values ​​for all points within the visible area. The depth value is the distance of the observed point from the sensor. The infrared stream allows darker parts of the field of view to be viewed more clearly.

[0105]

[0141] Game start detection occurs at 2420. To detect when a game starts, some embodiments use cards as a trigger. For cardless games such as roulette, a dolly can be detected when the game ends and a game start event is triggered retroactively to the nth frame.

[0106]

[0142] Chip detection and prediction occurs at 2430. This is the process of detecting the location of the chips and predicting the chip monetary value and how many chips there are in the chip stack. This algorithm (or process) is based on adaptive thresholding and morphological operations. It can be applied to both color and infrared images.

[0107]

[0143] Chip value prediction is performed at 2460 by using template matching in the color image and the reflectance features of each chip type in the infrared image. A chip height prediction is then calculated based on the distance between the top surface of the chip and the table plane.

[0108]

[0144] Game end detection 2460 is the opposite process to game start detection. When the dealer puts the cards away, it triggers the game end event. In the case of roulette, the detection of the removal of the dolly triggers this event.

[0109]

[0145] Game Start and End Detection: FIG. 25 depicts a flow chart of a game monitoring process 2500 that some embodiments may implement to monitor events on a game table. Two key steps in this process are game start detection 2520 and game end detection 2560. The specific characteristics of the events that define the game start and end triggers may be stored in a game start and end trigger configuration 944 and may be referenced by the game monitoring module 928 to predict that a game has started or ended on the table. For example, for a table designated for a game of blackjack, the presence of one or more cards in the image frame extracted in step 2510 may be treated as the start of the game. Similarly, the absence of any cards in the image frame after the start of the game may be treated as the end of the game for the purposes of game end detection 2560. For non-card-based games such as roulette, the presence of other game objects such as dollies may be used as start and end triggers for the game. The specific shapes and characteristics of the game objects that activate the game start or end triggers may be stored in the game start and end trigger configuration 944 of the configuration module 940 of the computing device 130. These configurations may be managed through the web application server 160 on the database server 150 and then transferred to the computing device 130 via the table configuration transfer module 1140 .

[0110]

[0146] Overall monitoring process: The game monitoring process 2500 starts with a step of extraction of an image frame 2510. This step includes capturing an image of the game environment 110 by the depth sensing device and camera 120, and transmitting the captured image via link 107 to the computing device 130 available for analysis by the game monitoring module 928. The image may be captured in color or infrared format, or both, or any other format in which the depth sensing device and camera are capable of capturing images. The next step 2520 includes detection of the start of the game, for example, by application of Algorithm 1: Card Detection Algorithm or Algorithm 2: Dolly Detection Algorithm. Once the start of the game is detected, a step of automatically predicting the plane of the game table or playing surface 2530 is performed. Step 2530 may be performed using the PCA discussed above, and the results may be stored in the memory of the computing device 130.

[0111]

[0147] The next step 2540 involves detecting chips. This step may be performed through an Algorithm 3: Chip Detection algorithm. If chips are detected, the next step 2550 may be predicting the value of the stack of chips in the image. This step is performed by performing an Algorithm 4: Chip Stack Height Prediction algorithm or process, which may be performed in real time or near real time. Part of this process may also include automatically predicting the color of the chip or betting object at the top of the stack and retrieving a value associated with the color from the table configuration module 942. The retrieved value may be multiplied by the predicted number of betting objects or chips in the stack to automatically predict the value of the entire stack. The results of the application of this algorithm or process may be stored in the memory of the computing device 130 and recorded as an event or transaction. The detected stack of chips is also associated with an area of ​​interest on the table to distinguish it from other stacks of chips on the same gaming table.

[0112]

[0148] Multiple stacks of chips or betting objects on a single target area may be detected separately and the predicted value of each stack may be recorded separately along with the unique identifier of the target area. Alternatively, the values ​​of multiple stacks of chips on a single target area may be summed and recorded along with the unique identifier of the target area. The height of multiple stacks of chips may be predicted in one iteration or several iterations of performing algorithm 4. After predicting the value of the stacks of chips, another image frame is grabbed via step 2514 and game end detection step 2560 is performed. Game end detection may be performed, for example, by looking for the absence of any cards or the absence of dollies on the game table in the image frame grabbed at step 2514. If the game end is not detected, another image frame is grabbed at step 2512, followed by yet another iteration of chip detection at step 2540.

[0113]

[0149] Once the end of the game is detected, in step 2570, the event data captured during the monitoring process is sent to the database server 150 via the message broker server 140 to be stored as game event data 1124. In other embodiments, reporting the transaction to the database server 150 may occur as the event is being detected in a real-time or near real-time manner. This event data may include a timestamp where the event occurred, the target area where the chip was detected, the estimated value of the chip stack, the game start and end times, and other relevant parameters captured by the game monitoring module 928. After reporting the transaction to the database server, the game monitoring system may continue to monitor the game table and wait for the start of the next game in steps 2510 and 2520. Examples of some records of event data captured during a game are as follows:

[0114]

[0150] Game Record 1<Table ID: R.2745; Game ID: 17.0.20170421150202; Game Start Time: 2017-04-21 15:02:02; Game End Time: 2017-04-21 15:04:02>

[0115]

[0151] Game Object Record 1<Table ID: R.2745; Game ID: 17.0.20170421150202; Target Domain ID: Player-1; Bet Object Value: 5; Number of Bet Objects: 5>

[0116]

[0152] Game Object Record 2<Table ID: R.2745; Game ID: 17.0.20170421150202; Target Domain ID: Player-2; Betting Object Value: 10; Number of Betting Objects: 2>

[0117]

[0153] In the above example of game event data, game record 1 having a unique identifier "game ID: 17.0.20170421150202" is associated with a unique game table or playing surface identified by a unique identifier "table ID: R.2745". The game record also includes game start and end timestamp values. Associated with game record 1 are two game object records: game object record 1 and game object record 2. Game object record 1 represents predictions of five betting objects of value 5 detected within a destination area with a unique identifier "player-1". Game object record 2 represents predictions of two betting objects of value 10 detected within a destination area with a unique identifier "player-2".

[0118]

[0154] Multi-frame processing: The approach to game object detection described above considers a single frame or snapshot to perform any processing. The results provided by this approach can be further improved to avoid false positives and achieve higher accuracy by multi-frame processing techniques. This technique involves processing all frames for game object detection and chip height prediction, and logging the results obtained after all processing iterations. To improve the game object detection and chip stack height prediction process, certain processing rules can be applied to the logged results. For example, averaging of predicted chip stack values ​​over several frames can be used to compensate for a few frames with erroneous observations. Also, in games where the first card detection may be delayed, the game monitoring module 928 may go back in time across the captured frames to perform card detection again and retroactively identify the beginning of a new game. Moving objects that trigger a game can be reduced by detecting cards within multiple regions of interest and treating the game as started only if a certain percentage of the regions of interest, for example, 3 out of 8 regions, or 2 out of 12 regions, are detected with cards.

[0119]

[0155] Multi-frame processing techniques may also be used to detect gaming objects that may be temporarily obstructed from the field of view of the depth sensing device and camera 120. FIG. 31 is an image frame 3000 representing a field of view of a gaming table with an obstruction in the form of a dealer's head 3010. FIG. 31 is an image frame 3100 representing a field of view of the gaming table of FIG. 30 taken a few seconds before and after image frame 3000. Image frame 3100 does not have the obstruction 3010 and application of the chip detection algorithm results in the identification of a chip 3113 not previously detected in image frame 3000. A combination of results obtained by application of the gaming object detection process to image frames that may have been acquired within a short time period, such as 2-4 seconds or 1-2 seconds or within 1 second, may be used to improve the overall accuracy of the gaming object detection process.

[0120]

[0156] The multi-frame processing techniques described above may also be extended to image frames obtained from different sources, such as images obtained from a visual image camera and an infrared camera, or images obtained from two or more depth sensing devices and cameras 120. FIG. 32 is an image frame 3200 obtained from a visual image camera that may be part of the depth sensing device and camera 120. In this image 3200, application of a chip detection process may generate a false positive result by identifying a chip 3210. FIG. 33 is an image frame 3300 obtained from an infrared camera after an image pre-processing step and edge and contour detection techniques have been applied to the raw image frame. Application of the chip detection process in the image frame 3300 does not generate the false positive result generated in the image frame 3200. Thus, a combination of results obtained by application of a game object detection process to multiple different (types of) image frames may be effective in some embodiments to improve the accuracy of the game object detection results.

[0121]

[0157] The flowchart of Figure 34 is an example of an implementation of a multi-frame processing technique 3400 that some embodiments may implement. Step 3410 includes acquiring two image frames A and B through a depth sensing device and camera 120. Image frames A and B may be from a visual image camera or an infrared camera and may be separated in time of capture by, for example, a few milliseconds. Alternatively, image frames A and B have been acquired from a visual image camera and an infrared camera, respectively, for example, at the same point in time. Both image frames A and B are taken from substantially the same viewpoint and represent the same game table or playing surface.

[0122]

[0158] Step 3420 includes applying a game object detection process to each of the target areas in image frames A and B. The game object detection process may be, for example, a card detection process, or a chip or betting object detection process. Results obtained from the game object detection process for a given target area may include, for example, one or more of the following: identification of the presence of game objects in the target area, the number of game objects in the target area, the number of different groups of game objects in the target area, the number of game objects in each identified group, and / or the color of the game objects. If the game object detection process does not identify game objects in a particular target area, the result of that process may be a NULL result (no result reported) or a result indicating that no game objects are detected. Particular target predefined areas may be designated for identification of the presence of particular game objects. For example, a target predefined area may be designated for identification of game cards as part of the game object detection process. Other target predefined areas may be designated for identification of betting objects as part of the game object detection process.

[0123]

[0159] Step 3430 involves a comparison of the results obtained by application of the game object detection process to image frames A and B in step 3420. If the results obtained from image frames A and B match, e.g., the location, number and color of the game objects detected in image frames A and B are identical, then in step 3440 the results obtained are accepted and reported as a game event by computing device 130.

[0124]

[0160] If the results from image frames A and B do not match with respect to either the location or number or color of the detected game objects, then in step 3450 image frame C is acquired from the depth sensing device and camera 120. Image frame C may be captured from a different source, such as a visual image camera or an infrared camera, or may have been acquired at a point in time separated by, for example, a few milliseconds from the time image frames A or B were acquired. All images A, B, and C are acquired from substantially the same viewpoint and represent the same game table or playing surface.

[0125]

[0161] In step 3460, the game object detection process applied in step 3420 is also applied to image frame C. In step 3470, the results obtained from the application of the game object detection process to image frames A, B, and C in steps 3420 and 3460 are compared. If the result obtained from image frame A matches the result obtained from image frame C, the result obtained from image frame B is discarded as an exception and the result obtained from image frame A or C is accepted in step 3440. If the result obtained from image frame B matches the result obtained from image frame C, the result obtained from image frame A is discarded as an exception and the result obtained from image frame B or C is accepted in step 3440. If none of the results obtained from image frames A or B or C match, a new set of images may be acquired in step 3410 to obtain better game object detection results.

[0126]

[0162] In this description of some embodiments, reference is made to diagrams that are specific abstractions of software and hardware systems that combine to form the embodiments. These abstractions should be understood as aiding in the understanding of the embodiments and enabling their reproduction and implementation, and are therefore presented herein. The division of the system into blocks is according to the functionality of the system, and as such is presented with the understanding that in implementation, it is not necessary to have the software or hardware components logically or physically divided in such a manner. Similarly, physical installation is one possibility, as the functionality of the system may be distributed differently without departing from this disclosure. For simplicity, the system and specific examples are described in terms of the games of blackjack, baccarat, and roulette. It should be understood that some or all of the components of the system may also be applied to other table games.

[0127]

[0163] Some embodiments may further use an initial background image of the camera field of view. This background image may be of an empty table. This image may be warped into a bird's-eye view based on the main feature in the image (the table layout). This new image may be called the normalized background image and may be used from now on for background abstraction.

[0128]

[0164] It will be appreciated by those skilled in the art that numerous variations and / or modifications may be made to the above-described embodiments without departing from the broad general scope of the present disclosure. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive.

Claims

1. 1. A method for monitoring game play on a table surface of a game table, comprising: receiving, by a computing device in communication with at least one camera, an image of the table surface captured by the at least one camera in real time; analyzing at least one captured image of the table surface in real time to identify the presence of gaming objects on the table surface, where the gaming objects are cards or chips; determining, based on identifying a presence of a game object on the table surface, whether the identified game object is a game start trigger game object by comparing at least one visual characteristic of the identified game object with stored configuration data for predetermined game start trigger game objects; responsive to determining that the identified game object is the game start trigger activated game object, determining that a game start event has occurred and recording a timestamp for the game event; transmitting, by the computing device, game event data to a server, the game event data including an indication of the game start event and the timestamp of the game event; A method comprising:

2. analyzing at least one captured image of the table surface in real time to identify that no game initiation trigger activating game object is present on the table surface; determining that a game end event has occurred in response to a game start trigger activated game object being identified as not being present on the table surface and recording a time stamp for the game event; transmitting, by the computing device, game event data to a server, the game event data including an indication of the game end event and the timestamp of the game event; The method of claim 1 further comprising:

3. Determining whether the identified game object is a game initiation trigger activation game object comprises: pre-processing the at least one captured image to obtain a pre-processed image; identifying at least one contour in the pre-processed image; comparing said at least one contour to at least one contour criterion to determine whether said contour is associated with a game initiation trigger activated game object; The method of claim 1 or 2, comprising:

4. Pre-processing the at least one captured image to obtain a pre-processed image includes: performing a Canny edge detection process on said captured image; The method of claim 3 , comprising:

5. Pre-processing the at least one captured image to obtain a pre-processed image includes: Dilating the output of the Canny edge detection process to remove any holes between edge segments. The method of claim 4 further comprising:

6. Comparing the at least one contour to at least one contour criterion comprises: comparing the area defined by the contour with a predetermined area corresponding to an expected size of the game initiation trigger activated game object; The method according to any one of claims 3 to 5, comprising:

7. Comparing the at least one contour to at least one contour criterion comprises: comparing the number of vertices defined by said contour with a predetermined number of vertices corresponding to the expected number of vertices of said game start trigger activated game object; The method according to any one of claims 3 to 6, comprising:

8. Comparing the at least one contour to at least one contour criterion comprises: comparing the cosine of the angle between the joining edges defined by said contours to a predetermined value; The method according to any one of claims 3 to 7, comprising:

9. Pre-processing the at least one captured image to obtain a pre-processed image includes: performing a global thresholding operation on said captured image; The method according to any one of claims 3 to 8, comprising:

10. Pre-processing the at least one captured image to obtain a pre-processed image includes: The method of claim 9 , comprising eroding the captured image to remove objects smaller than a predetermined size.

11. A method according to any one of claims 1 to 10, wherein the game start triggering game object is a card.

12. The method according to any one of the preceding claims, wherein the image of the table surface is either a colour image, a greyscale image or an infrared image.

13. A non-transitory computer readable storage medium comprising instructions which, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 12.

14. 1. A system for monitoring game play on a table surface of a game table, comprising: at least one camera configured to capture an image of the table surface; a computing device in communication with the at least one camera, the computing device comprising: receiving an image of the table surface captured by the at least one camera in real time; analyzing at least one captured image of the table surface in real time to identify the presence of gaming objects on the table surface, wherein the gaming objects are cards or chips; determining, based on identifying a presence of a game object on the table surface, whether the identified game object is a game start trigger game object by comparing at least one visual characteristic of the identified game object with stored configuration data for predetermined game start trigger game objects; responsive to determining that the identified game object is the game start trigger activated game object, determining that a game start event has occurred and recording a time stamp for the game event; transmitting game event data to a server, the game event data including an indicator of the game start event and the timestamp of the game event; A computing device configured to execute A system comprising:

15. the computing device further comprising: analyzing at least one captured image of the table surface in real time to identify that no game initiation trigger activating game object is present on the table surface; determining that a game end event has occurred in response to a game start trigger activated game object being identified as not being present on the table surface and recording a time stamp for the game event; transmitting game event data to a server, the game event data including an indicator of the game end event and the timestamp of the game event; The system of claim 14 configured to execute:

16. Determining whether the identified game object is a game initiation trigger that initializes game objects includes: pre-processing the at least one captured image to obtain a pre-processed image; identifying at least one contour in the pre-processed image; comparing said at least one contour to at least one contour criterion to determine whether said contour is associated with a game initiation trigger activation game object; 16. The system according to claim 14 or 15, comprising:

17. Comparing the at least one contour to at least one contour criterion comprises: comparing the area defined by the contour with a predetermined area corresponding to an expected size of the game initiation trigger activated game object; The system of claim 16 , comprising:

18. Comparing the at least one contour to at least one contour criterion comprises: comparing the number of vertices defined by said contour with a predetermined number of vertices corresponding to the expected number of vertices of said game start trigger activated game object; 18. The system according to claim 16 or 17, comprising:

19. Comparing the at least one contour to at least one contour criterion comprises: comparing the cosine of the angle between the joining edges defined by said contours to a predetermined value; The system according to any one of claims 16 to 18, comprising:

Citation Information

Patent Citations

  • Game machine

    JP2006006912A

  • Computerized method for tracking object in frame sequence

    JP2007213560A

  • Casino table

    JP2011115266A

  • Device and method for displaying image of vehicle surroundings

    JP2011155393A

  • Game machine using game chip and game system

    JP2012157785A