A Go review method based on video image analysis

Through video image analysis and point-occupying calculation, the problem of inefficiency in the existing Go review methods is solved, and the meticulousness of game analysis and the efficiency of review are improved, helping players to locate the number of errors and pay attention to time points.

CN118762312BActive Publication Date: 2024-11-15YOUTH ZONGHENG (JIANGSU) TECH DEV CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411244893.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-06
Publication Date
2024-11-15
Estimated Expiration
2044-09-06

AI Technical Summary

Technical Problem

The existing Go review methods are difficult to effectively analyze when players suffer major losses and thinking time in the game, resulting in inefficient review, and the analysis of winning and defeated chess games is not detailed enough, making it difficult to locate the number of wrong steps and the opponent's initial layout.

Method used

Get chess game information through video image analysis, filter chess board images without hand time points, count chess pieces occupying points, calculate the reduction coefficient of occupation point and think about hazard coefficient, analyze the winning and failure chess game, and send the attention time points and estimated chess score to Chess player mobile APP, conduct detailed review.

Benefits of technology

It improves the efficiency and accuracy of review, helps players to locate the time points of attention, reduces the difficulty of review, and enhances the meticulousness of game analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118762312B_ABST
    Figure CN118762312B_ABST
Patent Text Reader

Abstract

The invention discloses a Go game review method based on video image analysis, and relates to the technical field of image analysis. The invention comprises: step 1. game acquisition, step 2. preliminary image analysis, step 3. victory game analysis, step 4. failure game analysis and step 5. game review processing. The invention analyzes the chessboard image of the game to facilitate calculation of the number of points occupied by two kinds of chess pieces at each time point in each game, and analyzes the victory game and the failure game separately to facilitate positioning at which step a chess player makes a mistake, and adds the initial layout chess record of the opponent for the failure game, because it is difficult to have exactly the same chess record in the actual chess playing process, so the initial layout chess record is analyzed for similarity, and the game is explained in detail for each estimated chess record, so as to facilitate the chess player to locate the time point of attention, improve the efficiency of the game review, and reduce the difficulty of the chess player's game review.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a Go game review method based on video image analysis. Background Art

[0002] Go is a traditional intellectual game that can improve players' strategic thinking ability. Among the methods to improve Go skills, reviewing the game is one of the most effective methods. Reviewing the game after a game can help players understand their own shortcomings. Traditional review methods are generally recorded manually, which is time-consuming and laborious. With the development of information technology, video images can better record the game, which can make it more convenient for players to summarize. Therefore, it is necessary to study a Go review method through video image analysis.

[0003] Prior art, such as the invention patent application with announcement number: CN112507926B, discloses a Go game review method based on video image analysis, which includes: collecting chessboard video images during the game; identifying chessboard positions based on the collected video images; judging whether the image is blocked by hands, if so, discarding the frame image, and identifying chess piece positions in each frame image based on the identified chessboard position; and completing the game review based on the identified chess piece positions in each frame image. This invention aims to solve the problem that the method of Go game review through video images in the prior art may be interfered by the hands of the chess player and may have misidentification phenomena. At the same time, it also uses rules to judge and correct misidentification phenomena when completing the game review process, and has higher accuracy than the prior art.

[0004] The prior art, such as the invention patent application with announcement number: CN117582651A, discloses a Go review method, device, equipment and storage medium, the method comprising: determining candidate games and element information of candidate games from the historical games that have ended; analyzing the element information of each candidate game based on a preset sorting strategy; and reviewing the target game according to the comprehensive score of each candidate game. The technical solution of this invention can timely screen out higher-quality games for review, avoid wasting server computing resources, and help course instructors better analyze students' chess playing ability and provide more professional suggestions.

[0005] It can be seen from the above scheme that the current Go review method generally only focuses on the order of placing pieces during the game and the selection of the review game. On the one hand, there is a lack of attention to the comprehensive analysis of when major losses occur and the thinking time of each step. The game time of Go is generally relatively long, and it is difficult for players to remember at which step they suffer major losses, and the thinking time will also affect the performance of the players. If a comprehensive analysis is not performed, it will be difficult for the players to locate the time points of attention, thereby reducing the efficiency of the review. On the other hand, there is a lack of attention to the separate analysis of winning games and losing games, a lack of attention to which step the players made mistakes, and a lack of attention to the initial layout of the chess record used by the opponent, which makes it easy for players to find it difficult to locate the steps that need to be focused on during the review, thereby increasing the difficulty of the players' review. Summary of the invention

[0006] The purpose of the present invention is to provide a Go game review method based on video image analysis, which solves the problems existing in the background technology.

[0007] To solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a Go review method based on video image analysis, including: Step 1. Game acquisition: obtaining video recordings, win-loss status and chess pieces held by each player's game, and extracting chessboard images at each time point of each game played by the player.

[0008] Step 2. Preliminary image analysis: Filter the chessboard images of each time point without hand movements of each chess game played by the chess player, count the number of occupied points of the black and white pieces in the chessboard images of each time point without hand movements of each chess game played by the chess player, and filter the chess games played by the chess player in failure and victory.

[0009] Step 3. Analysis of winning chess games: Filter chessboard images of each winning chess game of a player at each time point of interest.

[0010] Step 4. Failed game analysis: Filter the chessboard images of each focused time point of each failed game of the chess player, and analyze the estimated chess records and similarity coefficients of the initial layout of the opponent of each failed game of the chess player.

[0011] Step 5. Replay processing: Get detailed explanations of the games for each estimated chess record, send the estimated chess records of the initial layouts of the opponents of each failed game of the chess player and their similarity coefficients and detailed explanations of the games, chessboard images at each time point of interest, and chessboard images at each time point of interest for each winning game to the chess player's mobile phone APP.

[0012] Preferably, the specific screening method of screening chessboard images of each hand-free time point of each chess game played by the chess player is as follows: based on the chessboard images of each time point of each chess game played by the chess player, and through hand image recognition technology, the chessboard images of each hand time point of each chess game played by the chess player are screened, and the chessboard images of the remaining time points of each chess game played by the chess player are marked as the chessboard images of each chessboard time point; if the chessboard image of a chessboard time point of a chess game played by the chess player is consistent with the chessboard image of another chessboard time point, then the chessboard image of the chessboard time point with an earlier time point in the video recording of the chess game is deleted, so as to obtain the chessboard images of each non-repeated chessboard time point in each chess game played by the chess player, and mark them as the chessboard images of each hand-free time point of each chess game played by the chess player.

[0013] Preferably, the specific statistical method for counting the number of occupied points of black and white pieces in the chessboard image at each no-hand time point of each chess game played by the chess player is as follows: dividing the chessboard image at each no-hand time point of each chess game played by the chess player into xy coordinate axes, with the lower left corner of the chessboard image as the coordinate origin, the direction from the lower left corner to the lower right corner of the chessboard as the positive direction of the x-axis, and the direction from the lower left corner to the upper left corner of the chessboard as the positive direction of the y-axis.

[0014] The ratio of the side length of the small squares of the standard chessboard to the side length of the entire chessboard is obtained from the local database, so as to obtain the coordinates of each point of the chessboard image at each hand-free time point of each chess game played by the chess player.

[0015] Analyze the chess piece placement status of each point in the chessboard image at each time point without a move in each chess game played by the chess player.

[0016] The number of points in the chessboard image at each time point without a hand movement in each chess game played by the chess player is counted, and the number of points in the chessboard image at each time point without a hand movement in each chess game played by the chess player is counted.

[0017] If the placement status of a certain point in the chessboard image at a certain hand-free time point in a chess game to which the chess player belongs is a no-piece state, and the placement status of the points connected to this point in all directions are all white, then this point is used as the white-occupied point in the chessboard image at the hand-free time point in the chess game to which the chess player belongs, thereby screening the white-occupied points in the chessboard images at each hand-free time point in each chess game to which the chess player belongs, and counting the number of white-occupied points in the chessboard images at each hand-free time point in each chess game to which the chess player belongs, plus the number of points in the white state in the chessboard images at each hand-free time point in each chess game to which the chess player belongs, and calculating the number of white-occupied points in the chessboard images at each hand-free time point in each chess game to which the chess player belongs.

[0018] Similarly, the number of black pieces' occupied points in the chessboard image at each time point without a move in each chess game played by the chess player is calculated.

[0019] Preferably, the specific analysis method of analyzing the chess piece placement status of each point of the chessboard image at each no-hand time point of each chess game played by the chess player is: obtaining a predefined chess piece placement determination radius, a white piece gray value threshold, and a black piece gray value threshold from a local database.

[0020] Taking the coordinates of each point of the chessboard image at each hand-free time point of each chess game played by the chess player as the center of the circle, and taking the predefined chess piece placement judgment radius as the radius, the chess piece placement judgment range of each point of the chessboard image at each hand-free time point of each chess game played by the chess player is obtained, and the grayscale value of each pixel point in the chess piece placement judgment range of each point of the chessboard image at each hand-free time point of each chess game played by the chess player is obtained, and the grayscale value of each pixel point in the chess piece placement judgment range is averaged, so as to obtain the grayscale average value of the chess piece placement judgment range of each point of the chessboard image at each hand-free time point of each chess game played by the chess player.

[0021] The chess placement status of each point in the chessboard image at each hand-free time point of each chess game played by the chess player is determined by the grayscale average value of the chess placement determination range of each point.

[0022] Preferably, the specific screening method of screening the chessboard images of each focus time point of each winning game of the chess player is: calculating the occupation point reduction coefficient of each hand-free time point of each winning game of the chess player.

[0023] Analyze the chessboard images at each focus point in time of each winning game of the chess player.

[0024] Preferably, the occupation point reduction coefficient of each hand-free time point in each winning game of the chess player is calculated by: obtaining the winning points a of each winning game of the chess player from the local database x , where x represents the number of each winning game, x=1,2,...,y, and y is a positive integer greater than 2.

[0025] According to the number of occupied points of black and white pieces in the chessboard images at each time point without hand movements in each chess game played by the chess player, the number of occupied points of black and white pieces in the chessboard images at each time point without hand movements in each chess game played by the chess player is extracted.

[0026] According to the chess pieces held by the chess player in each chess game, the chess pieces held by the chess player in each winning chess game are extracted, so as to obtain the number of occupied points b of the chess pieces held by the chess player at each hand-free time point in each winning chess game. xn, where n represents the number of each time point without moves in the winning game, n=1,2,...,m, and m is a positive integer greater than 2.

[0027] Calculate the reduction coefficient of the occupied points at each handless time point in each winning game of the chess player Where e represents a natural constant.

[0028] Preferably, the chessboard images of the failed chess games of the chess player are screened, and the specific screening method is: based on the chess pieces held in each chess game of the chess player, the chess pieces held in each failed chess game of the chess player are extracted, and based on the no-hand time points in each failed chess game of the chess player, the no-hand time points used for the player's own chess moves in each failed chess game of the chess player are extracted, and they are marked as the chess-playing time points of each failed chess game of the chess player.

[0029] Calculate the reduction coefficient δ of the occupied points at each chess time point of each failed chess game of the chess player ip , where i represents the number of each failed game, i=1,2,...,j, j is a positive integer greater than 2, and p represents the number of each no-move time point in the failed game, p=1,2,...,q, q is a positive integer greater than 2.

[0030] Calculate the thinking hazard coefficient ε of each chess player at each chess-playing time point in each failed chess game ip .

[0031] Calculate the error reflection coefficient φ of each chess player at each chess-playing time point in each failed chess game ip =δ ip +ε ip .

[0032] Through the error reflection coefficient, the chessboard images of each focus time point of each failed game of the chess player are analyzed.

[0033] Preferably, the calculation method of the occupation point reduction coefficient at each chess-playing time point of each failed chess game of the chess player is as follows: obtaining the number of failed points c of each failed chess game of the chess player from the local database i .

[0034] According to the chess pieces held by the chess player in each failed game, the number of occupied points d of the chess pieces held by the chess player at each chess-playing time point in each failed game is obtained. ip , and obtain the number of occupied positions f of the chess pieces held by the chess player at the next no-hand time point at each chess-playing time point of each failed chess game ip .

[0035] Calculate the reduction coefficient of the occupied points at each chess time point of each failed chess game of the chess player

[0036] Preferably, the specific calculation method of calculating the thinking hazard coefficient of each chess-playing time point of each failed chess game of the chess player is as follows: according to each chess-playing time point of each failed chess game of the chess player, the time length h from each chess-playing time point of each failed chess game of the chess player to the previous chess-playing time point is calculated. ip .

[0037] Calculate the thinking hazard coefficient of each chess player at each chess-playing time point in each failed chess game Where q represents the number of chess playing time points.

[0038] Preferably, the specific analysis method of the estimated chess records and similarity coefficients of the initial layouts of the opponents of each failed chess game to which the chess player belongs is: obtaining the position of each sorted chess piece of each initial layout chess record, the target value z, the side length value g of the small square of the standard chessboard, and the predefined sorting strategy from the local database.

[0039] According to the chess pieces held by the chess player in each failed game, the chessboard images of each hand-free time point of the opponent in each failed game of the chess player are obtained, and according to the order of time from far to near, the chessboard image of the zth hand-free time point of the opponent in each failed game of the chess player is selected and marked as the target chessboard image of each failed game of the chess player, the position of each chess piece of the opponent in the target chessboard image of each failed game of the chess player is extracted, and the chess pieces are sorted according to a predefined sorting strategy, so as to obtain the position of each sorted chess piece of the opponent in the target chessboard image of each failed game of the chess player.

[0040] According to the positions of the chess pieces in each initial layout chess record, the positions of the chess pieces of the opponent in the target chessboard image of each failed chess game of the chess player and the positions of the nearest chess pieces in each initial layout chess record are screened, and the distance t between the positions of the ranked chess pieces of the opponent in the target chessboard image of each failed chess game of the chess player and the positions of the corresponding ranked chess pieces in each initial layout chess record is calculated. iru , where r represents the number of each sorted chess piece, r=1,2,...,s, s is a positive integer greater than 2, and u represents the number of each initial layout chess record, u=1,2,...,w, w is a positive integer greater than 2.

[0041] Calculate the similarity coefficient between the target chessboard image of each failed game of the chess player and each initial layout chess record

[0042] A similarity coefficient threshold is obtained from a local database, and similarity coefficients of target chessboard images of each failed game of the chess player and each initial layout chess record are compared with the similarity coefficient threshold. If the similarity coefficient between the target chessboard image of a failed game of the chess player and an initial layout chess record is greater than the similarity coefficient threshold, the initial layout chess record is marked as an estimated chess record, thereby screening the estimated chess records of the target chessboard images of each failed game of the chess player, and extracting the similarity coefficients of the estimated chess records of the target chessboard images of each failed game of the chess player.

[0043] The beneficial effects of the present invention are as follows: (1) Step 1 of the present invention, chess game acquisition, obtains historical chess games and their information, thereby facilitating subsequent analysis.

[0044] (2) Step 2 of the present invention is to perform preliminary image analysis. By analyzing the chessboard images of the chess game, it is convenient to calculate how many points the two chess pieces each occupy at each time point in each chess game. Not only are pictures with hands eliminated, but also because long periods of thinking are prone to occur in Go games, many identical images often appear. These repeated images are eliminated, and only the image with the latest time point among the repeated images is retained to facilitate subsequent re-analysis.

[0045] (3) Step 3, analysis of winning games and step 4, analysis of losing games of the present invention, analyze winning games and losing games separately, so as to facilitate the location of the step where the chess player made a mistake. For losing games, the opponent's initial layout chess record is added. Since it is difficult for the chess record to be exactly the same in the actual chess game process, the similarity analysis of the initial layout chess record is performed to facilitate the subsequent review.

[0046] (4) Step 5 of the present invention is to perform a review process, by sending the chessboard images of each time point of interest to the chess player and providing a detailed explanation of each estimated chess record, so as to facilitate the chess player to locate the time point of interest, improve the efficiency of the review, and reduce the difficulty of the chess player's review. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Reference Figure 1 As shown, the present invention provides a Go review method based on video image analysis, including: Step 1. Game acquisition: obtaining video recordings, win-loss status and chess pieces of each game played by a player, and extracting chessboard images at each time point of each game played by the player.

[0051] In a specific embodiment, the method for obtaining the video recordings, win / loss status and chess pieces of each chess game played by the chess player is as follows: obtaining the video recordings, win / loss status and chess pieces of each chess game played by the chess player from a local database.

[0052] It should be noted that the local database is used to store video recordings of each chess game played by the chess player, the win-loss status and the chess pieces held, the ratio of the side length value of the small squares of the standard chessboard to the side length value of the entire chessboard, the predefined placement determination radius, the white piece gray value threshold, the black piece gray value threshold, the occupation point reduction coefficient threshold, the number of winning points in each winning game played by the chess player, the error reflection coefficient threshold, the number of losing points in each losing game played by the chess player, the positions of each ranked chess piece in each initial layout chess record, the target value, the side length value of the small squares of the standard chessboard, the predefined sorting strategy, the similarity coefficient threshold, and the detailed explanation of the game of each estimated chess record.

[0053] It should be noted that the video recording of the chess game was shot from directly above the chessboard.

[0054] It should be noted that the win-loss status includes: victory and defeat.

[0055] It should be noted that the chess pieces held include black pieces and white pieces.

[0056] It should be noted that the chessboard is a standard Go chessboard of 19*19.

[0057] Step 1 of the present invention is to obtain chess games, by obtaining historical chess games and their information, so as to facilitate subsequent analysis.

[0058] Step 2. Preliminary image analysis: Filter the chessboard images of each time point without hand movements of each chess game played by the chess player, count the number of occupied points of the black and white pieces in the chessboard images of each time point without hand movements of each chess game played by the chess player, and filter the chess games played by the chess player in failure and victory.

[0059] In a specific embodiment of the present invention, the specific screening method for screening chessboard images at each time point without hand movement of each chess game played by the chess player is as follows: based on the chessboard images at each time point of each chess game played by the chess player, and through the hand image recognition technology, the chessboard images at each hand time point of each chess game played by the chess player are screened, and the chessboard images at the remaining time points of each chess game played by the chess player are marked as the chessboard images at each chessboard time point, if the chessboard image at a certain chessboard time point of a chess game played by the chess player is consistent with the chessboard image at another chessboard time point, then the chessboard image at the earlier chessboard time point in the video recording of the chess game is deleted, so as to obtain the chessboard images at each non-repeated chessboard time point in each chess game played by the chess player, and mark them as the chessboard images at each hand-free time point of each chess game played by the chess player.

[0060] In a specific embodiment, the hand image recognition technology is used to filter the chessboard images of each hand time point of each chess game played by a chess player. The specific method is as follows: the existing hand image recognition technology is relatively mature. The existing hand image recognition technology is used to identify the presence of hands in the chessboard image at a certain time point of a chess game played by a chess player. The time point is marked as the hand time point, thereby filtering the chessboard images of each hand time point of each chess game played by the chess player.

[0061] In a specific embodiment of the present invention, the specific statistical method for counting the number of occupied points of black and white pieces in the chessboard image at each no-hand time point of each chess game played by the chess player is as follows: the chessboard image at each no-hand time point of each chess game played by the chess player is divided into xy coordinate axes, with the lower left corner of the chessboard image as the coordinate origin, the direction from the lower left corner to the lower right corner of the chessboard as the positive direction of the x-axis, and the direction from the lower left corner to the upper left corner of the chessboard as the positive direction of the y-axis.

[0062] The ratio of the side length of the small squares of the standard chessboard to the side length of the entire chessboard is obtained from the local database, so as to obtain the coordinates of each point of the chessboard image at each hand-free time point of each chess game played by the chess player.

[0063] Analyze the chess piece placement status of each point in the chessboard image at each time point without a move in each chess game played by the chess player.

[0064] It should be noted that the states of placing pieces include: black piece state, white piece state, and no piece state.

[0065] The number of points in the chessboard image at each time point without a hand movement in each chess game played by the chess player is counted, and the number of points in the chessboard image at each time point without a hand movement in each chess game played by the chess player is counted.

[0066] If the placement status of a certain point in the chessboard image at a certain hand-free time point in a chess game to which the chess player belongs is a no-piece state, and the placement status of the points connected to this point in all directions are all white, then this point is used as the white-occupied point in the chessboard image at the hand-free time point in the chess game to which the chess player belongs, thereby screening the white-occupied points in the chessboard images at each hand-free time point in each chess game to which the chess player belongs, and counting the number of white-occupied points in the chessboard images at each hand-free time point in each chess game to which the chess player belongs, plus the number of points in the white state in the chessboard images at each hand-free time point in each chess game to which the chess player belongs, and calculating the number of white-occupied points in the chessboard images at each hand-free time point in each chess game to which the chess player belongs.

[0067] It should be noted that the connection points in each direction of the point are the connection points in eight directions of the point, and the eight directions are: upper left, upper, upper right, right, lower right, lower, lower left, and left.

[0068] Similarly, the number of black pieces' occupied points in the chessboard image at each time point without a move in each chess game played by the player is calculated.

[0069] In a specific embodiment of the present invention, the specific analysis method for analyzing the chess piece placement status of each point in the chessboard image at each no-hand time point of each chess game played by the chess player is: obtaining a predefined chess piece placement determination radius, white piece gray value threshold, and black piece gray value threshold from a local database.

[0070] Taking the coordinates of each point of the chessboard image at each hand-free time point of each chess game played by the chess player as the center of the circle, and taking the predefined chess piece placement judgment radius as the radius, the chess piece placement judgment range of each point of the chessboard image at each hand-free time point of each chess game played by the chess player is obtained, and the grayscale value of each pixel point in the chess piece placement judgment range of each point of the chessboard image at each hand-free time point of each chess game played by the chess player is obtained, and the grayscale value of each pixel point in the chess piece placement judgment range is averaged, so as to obtain the grayscale average value of the chess piece placement judgment range of each point of the chessboard image at each hand-free time point of each chess game played by the chess player.

[0071] The chess player's chess placement status at each point in the chessboard image at each hand-free time point in each chess game is determined by the grayscale average value of the chess placement determination range at each point.

[0072] In a specific embodiment, the chess player determines the chess piece placement status of each point of the chessboard image at each hand-free time point of each chess game by using the grayscale average value of the chess piece placement determination range at each point. The specific method is as follows: compare the grayscale average value of the chess piece placement determination range of each point of the chessboard image at each hand-free time point of each chess game to which the chess player belongs with the white piece grayscale value threshold. If the grayscale average value of the chess piece placement determination range of a certain point of the chessboard image at a hand-free time point of a chess game to which the chess player belongs is less than the white piece grayscale value threshold, If the gray value threshold of the chess piece is less than the white piece gray value, the state of the chess piece at that point is recorded as the white piece state. Otherwise, the gray value average of the chess piece placement range at that point in the chessboard image at that time point without the player's hand in the chess game to which the player belongs is compared with the black piece gray value threshold. If the gray value average of the chess piece placement range at that point in the chessboard image at that time point without the player's hand in the chess game to which the player belongs is greater than the black piece gray value threshold, the state of the chess piece at that point is recorded as the black piece state. Otherwise, the state of the chess piece at that point is recorded as the no-piece state.

[0073] Step 2 of the present invention is a preliminary image analysis. By analyzing the chessboard images of the chess game, it is convenient to calculate how many points are occupied by the two chess pieces at each time point in each chess game. Not only are pictures with hands eliminated, but also because long-term thinking is prone to occur in Go games, many identical images often appear. These repeated images are eliminated, and only the image with the latest time point in the repeated images is retained to facilitate subsequent re-analysis.

[0074] Step 3. Analysis of winning chess games: Filter chessboard images of each winning chess game of a player at each time point of interest.

[0075] In a specific embodiment of the present invention, the chessboard images of each focus time point of each winning game of the chess player are screened, and the specific screening method is: calculating the occupation point reduction coefficient of each hand-free time point of each winning game of the chess player.

[0076] It should be noted that the larger the coefficient of reduction in occupied points is, the more dead pieces are taken by the opponent between two steps, and the chessboard image with more dead pieces needs more attention.

[0077] Analyze the chessboard images at each focus point in time of each winning game of the chess player.

[0078] In a specific embodiment, the specific method for analyzing the chessboard images of each focus time point of each victorious game played by the chess player is as follows: obtaining an occupation point reduction coefficient threshold from a local database, comparing the occupation point reduction coefficient of each hand-free time point of each victorious game played by the chess player with the occupation point reduction coefficient threshold; if the occupation point reduction coefficient of a hand-free time point of a victorious game played by the chess player is greater than the occupation point reduction coefficient threshold, marking the hand-free time point as a focus time point, thereby screening the focus time points of each victorious game played by the chess player; and based on the chessboard images of each hand-free time point of each game played by the chess player, extracting the chessboard images of each focus time point of each victorious game played by the chess player.

[0079] In a specific embodiment of the present invention, the occupation point reduction coefficient of each handless time point of each winning game of the chess player is calculated by: obtaining the victory point number a of each winning game of the chess player from the local database x , where x represents the number of each winning game, x=1,2,...,y, and y is a positive integer greater than 2.

[0080] According to the number of occupied points of black and white pieces in the chessboard images at each time point without hand movements in each chess game played by the chess player, the number of occupied points of black and white pieces in the chessboard images at each time point without hand movements in each chess game played by the chess player is extracted.

[0081] According to the chess pieces held by the chess player in each chess game, the chess pieces held by the chess player in each winning chess game are extracted, so as to obtain the number of occupied points b of the chess pieces held by the chess player at each hand-free time point in each winning chess game. xn , where n represents the number of each time point without moves in the winning game, n=1,2,...,m, and m is a positive integer greater than 2.

[0082] Calculate the reduction coefficient of the occupied points at each handless time point in each winning game of the chess player Where e represents a natural constant.

[0083] Step 4. Failed game analysis: Filter the chessboard images of each focused time point of each failed game of the chess player, and analyze the estimated chess records and similarity coefficients of the initial layout of the opponent of each failed game of the chess player.

[0084] In a specific embodiment of the present invention, the chessboard images of the time points of interest in each failed game of the chess player are screened, and the specific screening method is: based on the chess pieces held in each game of the chess player, the chess pieces held in each failed game of the chess player are extracted, and based on the time points without hands in each failed game of the chess player, the time points without hands used for the player's own chess moves in each failed game of the chess player are extracted, and they are marked as the chess-playing time points of each failed game of the chess player.

[0085] Calculate the reduction coefficient δ of the occupied points at each chess time point of each failed chess game of the chess player ip , where i represents the number of each failed game, i=1,2,...,j, j is a positive integer greater than 2, and p represents the number of each no-move time point in the failed game, p=1,2,...,q, q is a positive integer greater than 2.

[0086] Calculate the thinking hazard coefficient ε of each chess player at each chess-playing time point in each failed chess game ip .

[0087] It should be noted that the greater the thinking hazard coefficient, the shorter the player's thinking time during the game. The shorter the thinking time, the easier it is to make mistakes in judgment, and the chessboard image at this time needs more attention.

[0088] Calculate the error reflection coefficient φ of each chess player at each chess-playing time point in each failed chess game ip =δ ip +ε ip .

[0089] Through the error reflection coefficient, the chessboard images of each focus time point of each failed game of the chess player are analyzed.

[0090] In a specific embodiment, the chessboard images of the respective focus time points of the failed chess games of the chess player are analyzed through the error reflection coefficient. The specific method is: obtaining the error reflection coefficient threshold from the local database, comparing the error reflection coefficient of each chess-playing time point of each failed chess game of the chess player with the error reflection coefficient threshold; if the error reflection coefficient of a chess-playing time point of a failed chess game of the chess player is greater than the error reflection coefficient threshold, marking the chess-playing time point as the focus time point, thereby screening the focus time points of each failed chess game of the chess player; and extracting the chessboard images of the respective focus time points of each failed chess game of the chess player based on the chessboard images of each hand-free time point of each chess game of the chess player.

[0091] In a specific embodiment of the present invention, the occupation point reduction coefficient of each chess-playing time point of each failed chess game of the chess player is calculated by: obtaining the number of failed points c of each failed chess game of the chess player from the local databasei .

[0092] According to the chess pieces held by the chess player in each failed game, the number of occupied points d of the chess pieces held by the chess player at each chess-playing time point in each failed game is obtained. ip , and obtain the number of occupied positions f of the chess pieces held by the chess player at the next no-hand time point at each chess-playing time point of each failed chess game ip .

[0093] It should be noted that the next non-move time point of each chess-playing time point in each failed game of the chess player is represented by the opponent's chess-playing time point after the chess player completes each move.

[0094] Calculate the reduction coefficient of the occupied points at each chess time point of each failed chess game of the chess player

[0095] In a specific embodiment of the present invention, the thinking hazard coefficient of each chess-playing time point of each failed chess game of the chess player is calculated by: according to each chess-playing time point of each failed chess game of the chess player, the time length h from each chess-playing time point of each failed chess game of the chess player to the previous chess-playing time point is calculated. ip .

[0096] Calculate the thinking hazard coefficient of each chess player at each chess-playing time point in each failed chess game Where q represents the number of chess playing time points.

[0097] In a specific embodiment of the present invention, the estimated chess records and similarity coefficients of the initial layouts of the opponents of each failed game of the chess player are analyzed, and the specific analysis method is: obtaining the position of each sorted chess piece of each initial layout chess record, the target value z, the side length value g of the small square of the standard chessboard, and the predefined sorting strategy from the local database.

[0098] According to the chess pieces held by the chess player in each failed game, the chessboard images of each hand-free time point of the opponent in each failed game of the chess player are obtained, and according to the order of time from far to near, the chessboard image of the zth hand-free time point of the opponent in each failed game of the chess player is selected and marked as the target chessboard image of each failed game of the chess player, the position of each chess piece of the opponent in the target chessboard image of each failed game of the chess player is extracted, and the chess pieces are sorted according to a predefined sorting strategy, so as to obtain the position of each sorted chess piece of the opponent in the target chessboard image of each failed game of the chess player.

[0099] According to the positions of the chess pieces in each initial layout chess record, the positions of the chess pieces of the opponent in the target chessboard image of each failed chess game of the chess player and the positions of the nearest chess pieces in each initial layout chess record are screened, and the distance t between the positions of the ranked chess pieces of the opponent in the target chessboard image of each failed chess game of the chess player and the positions of the corresponding ranked chess pieces in each initial layout chess record is calculated. iru , where r represents the number of each sorted chess piece, r=1,2,...,s, s is a positive integer greater than 2, and u represents the number of each initial layout chess record, u=1,2,...,w, w is a positive integer greater than 2.

[0100] Calculate the similarity coefficient between the target chessboard image of each failed game of the chess player and each initial layout chess record

[0101] A similarity coefficient threshold is obtained from a local database, and similarity coefficients of target chessboard images of each failed game of the chess player and each initial layout chess record are compared with the similarity coefficient threshold. If the similarity coefficient between the target chessboard image of a failed game of the chess player and an initial layout chess record is greater than the similarity coefficient threshold, the initial layout chess record is marked as an estimated chess record, thereby screening the estimated chess records of the target chessboard images of each failed game of the chess player, and extracting the similarity coefficients of the estimated chess records of the target chessboard images of each failed game of the chess player.

[0102] Step 3 of the present invention, analysis of winning games and step 4 of the present invention, analyze winning games and losing games separately, so as to facilitate the location of the step where the chess player made a mistake. For losing games, the opponent's initial layout chess record is added. Since it is difficult for the chess record to be exactly the same in the actual chess playing process, the initial layout chess record is analyzed for similarity, so as to facilitate the subsequent review.

[0103] Step 5. Replay processing: Get detailed explanations of the games for each estimated chess record, send the estimated chess records of the initial layouts of the opponents of each failed game of the chess player and their similarity coefficients and detailed explanations of the games, chessboard images at each time point of interest, and chessboard images at each time point of interest for each winning game to the chess player's mobile phone APP.

[0104] In a specific embodiment, the detailed explanation of each estimated chess game is obtained by a specific method: the detailed explanation of each estimated chess game is obtained from a local database.

[0105] Step 5 of the present invention is to perform a review process, by sending chessboard images of each time point of interest to the chess player, and providing a detailed explanation of each estimated chess record, so as to facilitate the chess player to locate the time point of interest, improve the efficiency of the review, and reduce the difficulty of the chess player's review.

[0106] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they shall all fall within the protection scope of the present invention.

Claims

1. A Go review method based on video image analysis, It is characterized in that include: Step 1. Game acquisition: obtaining video recordings, win / loss status, and chess pieces of each chess game played by a chess player, and extracting chessboard images at each time point of each chess game played by the chess player; Step 2. Preliminary image analysis: Screen the chessboard images of each time point without a hand in each chess game played by the chess player, count the number of occupied points of the black and white pieces in the chessboard images of each time point without a hand in each chess game played by the chess player, and screen the chess games played by the chess player in failure and victory; Step 3. Analysis of winning chess games: screening chessboard images of each winning chess game of the chess player at each time point of interest; The specific screening method of screening chessboard images of each winning chess game of each chess player at each focus time point is as follows: Calculate the reduction coefficient of the occupied points at each handless time point in each winning game of the chess player; Analyze the chessboard images at each time point of interest of each winning game of the chess player; The specific calculation method of the occupation point reduction coefficient of each hand-free time point in each winning game of the chess player is: Get the number of wins of each winning game of the chess player from the local database a x , where x represents the number of each winning game, x=1,2,...,y, and y is a positive integer greater than 2; According to the number of occupied points of the black pieces and the white pieces in the chessboard images at each time point without a hand in each chess game played by the chess player, the number of occupied points of the black pieces and the white pieces in the chessboard images at each time point without a hand in each chess game played by the chess player is extracted; According to the chess pieces held by the chess player in each chess game, the chess pieces held by the chess player in each winning chess game are extracted, so as to obtain the number of occupied points b of the chess pieces held by the chess player at each hand-free time point in each winning chess game. xn , where n represents the number of each time point without moves in the winning game, n = 1, 2, ..., m, and m is a positive integer greater than 2; Calculate the reduction coefficient of the occupied points at each handless time point in each winning game of the chess player Where e represents a natural constant; The specific method of analyzing the chessboard images of each focus time point of each winning chess game of the chess player is as follows: Obtaining an occupation point reduction coefficient threshold from a local database, comparing the occupation point reduction coefficient of each hand-free time point of each winning game of the chess player with the occupation point reduction coefficient threshold, if the occupation point reduction coefficient of a hand-free time point of a winning game of the chess player is greater than the occupation point reduction coefficient threshold, marking the hand-free time point as a focus time point, thereby screening the focus time points of each winning game of the chess player, and extracting chessboard images of each focus time point of each winning game of the chess player based on the chessboard images of each hand-free time point of each game of the chess player; Step 4. Failed game analysis: screening chessboard images of each focused time point of each failed game of the chess player, and analyzing each estimated chess record and similarity coefficient of the initial layout of the opponent of each failed game of the chess player; The specific screening method of screening chessboard images of each focused time point of each failed chess game of each chess player is as follows: According to the chess pieces held by the chess player in each chess game, the chess pieces held by the chess player in each failed chess game are extracted, and according to the no-hand time points of each failed chess game, the no-hand time points used for the chess player to play chess in each failed chess game are extracted, and marked as the chess playing time points of each failed chess game; Calculate the reduction coefficient δ of the occupied points at each chess time point of each failed chess game of the chess player ip , where i represents the number of each failed game, i=1,2,...,j, j is a positive integer greater than 2, p represents the number of each no-move time point of the failed game, p=1,2,...,q, q is a positive integer greater than 2; Calculate the thinking hazard coefficient ε of each chess player at each chess-playing time point in each failed chess game ip ; Calculate the error reflection coefficient φ of each chess player at each chess-playing time point in each failed chess game ip =δ ip +ε ip ; By using the error reflection coefficient, the chessboard images of the chess players at each focus time point of each failed game are analyzed; The specific calculation method of the occupation point reduction coefficient of each chess-playing time point of each failed chess game of the chess player is as follows: Get the number of failed points c of each failed game of the chess player from the local database i ; According to the chess pieces held by the chess player in each failed game, the number of occupied points d of the chess pieces held by the chess player at each chess-playing time point in each failed game is obtained. ip , and obtain the number of occupied positions f of the chess pieces held by the chess player at the next no-hand time point at each chess-playing time point of each failed chess game ip ; Calculate the reduction coefficient of the occupied points at each chess time point of each failed chess game of the chess player The specific calculation method of the thinking hazard coefficient of each chess-playing time point of each failed chess game of the chess player is: According to the chess-playing time points of each chess-playing game of the chess player, the time length h between each chess-playing time point of each chess-playing game of the chess player and the previous chess-playing time point is calculated. ip ; Calculate the thinking hazard coefficient of each chess player at each chess-playing time point in each failed chess game Where q represents the number of chess playing time points; Step 5. Replay processing: Get detailed explanations of the games for each estimated chess record, send the estimated chess records of the initial layouts of the opponents of each failed game of the chess player and their similarity coefficients and detailed explanations of the games, chessboard images at each time point of interest, and chessboard images at each time point of interest for each winning game to the chess player's mobile phone APP.

2. A Go game review method based on video image analysis according to claim 1, It is characterized in that The specific screening method of screening chessboard images of each chess game played by a chess player at each time point without a hand is as follows: According to the chessboard images at each time point of each chess game played by the chess player, and through the hand image recognition technology, the chessboard images at each hand time point of each chess game played by the chess player are screened, and the chessboard images at the remaining time points of each chess game played by the chess player are marked as the chessboard images at each chessboard time point; if the chessboard image at a chessboard time point of a chess game played by the chess player is consistent with the chessboard image at another chessboard time point, the chessboard image at the earlier chessboard time point in the video recording of the chess game is deleted, so as to obtain the chessboard images of each non-repeated chessboard time point in each chess game played by the chess player, and mark them as the chessboard images without hand time points of each chess game played by the chess player.

3. A Go game review method based on video image analysis according to claim 1, It is characterized in that The specific statistical method for counting the number of occupied points of black and white pieces in the chessboard image at each time point without a move in each chess game played by the chess player is as follows: The chessboard image of each time point without a move in each chess game of the chess player is divided into xy coordinate axes, with the lower left corner of the chessboard image as the coordinate origin, the direction from the lower left corner to the lower right corner of the chessboard as the positive direction of the x-axis, and the direction from the lower left corner to the upper left corner of the chessboard as the positive direction of the y-axis; Obtaining the ratio of the side length of the small squares of the standard chessboard to the side length of the entire chessboard from the local database, thereby obtaining the coordinates of each point of the chessboard image at each hand-free time point of each chess game played by the chess player; Analyze the chess piece placement status of each point of the chessboard image at each time point without a hand in each chess game played by the chess player; Counting the number of points in the chessboard image at each time point without a hand in each chess game played by the chess player where the chess piece is in a white state, and counting the number of points in the chessboard image at each time point without a hand in each chess game played by the chess player where the chess piece is in a black state; If the chess piece placement state of a certain point of the chessboard image at a certain hand-free time point of a chess game to which the chess player belongs is a no-piece state, and the chess piece placement states of the connected points in all directions of the point are all white, then the point is used as the white-occupied point of the chessboard image at the hand-free time point of the chess game to which the chess player belongs, thereby screening the white-occupied points of the chessboard images at each hand-free time point of each chess game to which the chess player belongs, and counting the number of white-occupied points of the chessboard images at each hand-free time point of each chess game to which the chess player belongs, plus the number of points whose chess piece placement state is white in the chessboard images at each hand-free time point of each chess game to which the chess player belongs, and calculating the number of white-occupied points of the chessboard images at each hand-free time point of each chess game to which the chess player belongs; Similarly, the number of black pieces' occupied points in the chessboard image at each time point without a move in each chess game played by the player is calculated.

4. A Go game review method based on video image analysis according to claim 3, It is characterized in that The specific analysis method of analyzing the chess piece placement status of each point of the chessboard image at each time point without a hand in each chess game of the chess player is as follows: Obtain the predefined drop determination radius, white piece gray value threshold, and black piece gray value threshold from the local database; Taking the coordinates of each point of the chessboard image at each time point without hand movements of each chess game played by the chess player as the center of the circle and taking the predefined chessboard move determination radius as the radius, obtaining the chessboard move determination range of each point of the chessboard image at each time point without hand movements of each chess game played by the chess player, and obtaining the grayscale value of each pixel point of the chessboard move determination range of each point of the chessboard image at each time point without hand movements of each chess game played by the chess player, calculating the average grayscale value of each pixel point of the chessboard move determination range, thereby obtaining the grayscale average value of the chessboard move determination range of each point of the chessboard image at each time point without hand movements of each chess game played by the chess player; The chess player's chess placement status at each point in the chessboard image at each hand-free time point in each chess game is determined by the grayscale average value of the chess placement determination range at each point.

5. A Go game review method based on video image analysis according to claim 4, It is characterized in that The specific analysis method of the estimated chess records and similarity coefficients of the initial layouts of the opponents of the chess players in the failed chess games is as follows: Obtain the position of each sorted chess piece of each initial layout chess record, the target value z, the side length g of the small square of the standard chessboard, and the predefined sorting strategy from the local database; According to the chess pieces held by the chess player in each failed game, the chessboard images of each time point without hands of the opponent in each failed game of the chess player are obtained, and according to the order of time from far to near, the chessboard images of the z-th time point without hands of the opponent in each failed game of the chess player are selected and marked as the target chessboard images of each failed game of the chess player, the positions of each chess piece of the opponent in the target chessboard images of each failed game of the chess player are extracted, and the chess pieces are sorted according to a predefined sorting strategy, so as to obtain the positions of each sorted chess piece of the opponent in the target chessboard images of each failed game of the chess player; According to the positions of the chess pieces in each initial layout chess record, the positions of the chess pieces of the opponent in the target chessboard image of each failed chess game of the chess player and the positions of the nearest chess pieces in each initial layout chess record are screened, and the distance t between the positions of the ranked chess pieces of the opponent in the target chessboard image of each failed chess game of the chess player and the positions of the corresponding ranked chess pieces in each initial layout chess record is calculated. iru , where r represents the number of each sorted chess piece, r=1,2,...,s, s is a positive integer greater than 2, u represents the number of each initial layout chess record, u=1,2,...,w, w is a positive integer greater than 2; Calculate the similarity coefficient between the target chessboard image of each failed game of the chess player and each initial layout chess record A similarity coefficient threshold is obtained from a local database, and similarity coefficients of target chessboard images of each failed game of the chess player and each initial layout chess record are compared with the similarity coefficient threshold. If the similarity coefficient between the target chessboard image of a failed game of the chess player and an initial layout chess record is greater than the similarity coefficient threshold, the initial layout chess record is marked as an estimated chess record, thereby screening the estimated chess records of the target chessboard images of each failed game of the chess player, and extracting the similarity coefficients of the estimated chess records of the target chessboard images of each failed game of the chess player.

Citation Information

Patent Citations

  • A Go game review method based on video image analysis

    CN112507926B

  • Go replaying method, device and equipment and storage medium

    CN117582651A

  • Weiqi playing process recalling method based on video image analysis

    CN112507926A

  • Wargame game strategy generation method and device and storage medium

    CN115222304A

  • Chinese chess killing method identification method and device and storage medium

    CN115463406A