Intelligent recognition method, system, device and storage medium for game scenes

Abstract

The present invention relates to a computer technology, and discloses a method, system, device and storage medium for intelligently identifying game scenes. The present invention analyzes the non-game interface and game interface of a target car racing game to obtain target non-game interface features and target game interface features, which constitute target interface features; obtains a preset text area and acquires text display logic; performs multi-feature recognition on the first text area of ​​a first game image to obtain first text feature information and first text content; determines whether the first game image belongs to a preset scene image, and if not, calls a preset scene recognition scheme; collects first hue information and first texture information; constructs a scene recognition model to obtain a first scene recognition result, which includes a first track map. Compared with the existing technology, the present invention can achieve the effect of improving the timeliness and accuracy of game scene recognition, and then provide users with targeted game services based on timely and reliable scene recognition, thereby enhancing user gaming experience.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, system, device and storage medium for intelligently identifying game scenes. Background Art

[0002] With the continuous development of smart terminal devices, games based on smart devices such as mobile phones and computers are becoming increasingly popular among users. Existing games are diverse in variety and scale, and the scenes in each game are ever-changing. Among them, racing games feature a rich variety of tracks to enhance user interest. Furthermore, due to the inherently high-speed nature of racing games, their game scenes are characterized by extremely rapid changes. Existing intelligent scene recognition and analysis for racing games suffers from an inability to timely and accurately identify the user's real-time game scene, especially scene recognition within the game interface, which places even higher demands on real-time recognition. Therefore, research on the intelligent recognition of racing game scenes using computer science and intelligent algorithms is urgently needed. Generally speaking, the drawbacks of existing methods are low recognition efficiency and inaccurate results when intelligently identifying and analyzing real-time scenes during user gameplay.

[0003] Therefore, how to improve the real-time and accuracy of scene recognition during the game, and thus ensure the user's gaming experience, has become an urgent problem to be solved. Summary of the Invention

[0004] To achieve the above objectives, the main purpose of the present invention is to provide a method, system, device and storage medium for intelligent recognition of game scenes, aiming to improve the real-time and accuracy of intelligent recognition of game scenes.

[0005] The invention proposes an intelligent recognition method for game scenes, comprising the following steps:

[0006] Analyze the non-game interface and the game interface of the target racing game in sequence, obtain target non-game interface features and target game interface features respectively, and form the target interface features;

[0007] Obtaining a preset text area based on the target interface features, and obtaining text display logic for the preset text area;

[0008] Performing multi-feature recognition on the first text area of ​​the first game image to obtain first text feature information, and combining the text display logic to obtain first text content;

[0009] determining whether the first game image belongs to a preset scene image based on the first text content, and if not, calling a preset scene recognition solution;

[0010] collecting first hue information and first texture information of the first game image based on the preset scene recognition scheme;

[0011] A scene recognition model is constructed, and the first hue information and the first texture information are input into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map.

[0012] In addition, to achieve the above-mentioned object, the present invention further proposes an intelligent recognition system for game scenes, the intelligent recognition system for game scenes comprising a memory and a processor, wherein the memory stores an intelligent recognition program for game scenes, and when the intelligent recognition program for game scenes is executed by the processor, the following steps are implemented:

[0013] Analyze the non-game interface and the game interface of the target racing game in sequence, obtain target non-game interface features and target game interface features respectively, and form the target interface features;

[0014] Obtaining a preset text area based on the target interface features, and obtaining text display logic for the preset text area;

[0015] Performing multi-feature recognition on the first text area of ​​the first game image to obtain first text feature information, and combining the text display logic to obtain first text content;

[0016] determining whether the first game image belongs to a preset scene image based on the first text content, and if not, calling a preset scene recognition solution;

[0017] collecting first hue information and first texture information of the first game image based on the preset scene recognition scheme;

[0018] A scene recognition model is constructed, and the first hue information and the first texture information are input into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map.

[0019] In addition, to achieve the above-mentioned object, the present invention also provides a computer device, which includes a processor and a memory;

[0020] The processor is used to process and execute the intelligent recognition method of the game scene;

[0021] The memory is coupled to the processor and is used to store the intelligent recognition program of the game scene. When the program is executed by the processor, the system executes the steps of the intelligent recognition method of the game scene.

[0022] In addition, to achieve the above-mentioned purpose, the present invention also proposes a computer-readable storage medium, wherein the computer-readable storage medium stores an intelligent recognition program for game scenes, and the intelligent recognition program for game scenes can be executed by at least one processor so that the at least one processor executes the steps of the intelligent recognition method for game scenes as described in any one of the above items.

[0023] The present invention sequentially analyzes the non-game interface and game interface of a target racing game, obtaining target non-game interface features and target game interface features, respectively, and forming target interface features. Based on the target interface features, a preset text area is obtained, and text display logic for the preset text area is acquired. Multi-feature recognition is performed on the first text area of ​​a first game image to obtain first text feature information, and first text content is obtained in combination with the text display logic. Based on the first text content, a determination is made as to whether the first game image belongs to a preset scene image, and if not, a preset scene recognition scheme is invoked. Based on the preset scene recognition scheme, first hue information and first texture information of the first game image are collected. A scene recognition model is constructed, and the first hue information and the first texture information are input into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map. Compared to existing technologies, the present invention can intelligently analyze and recognize game scenes based on computer science and technology, thereby improving the timeliness and accuracy of game scene recognition. Based on the timely and reliable scene recognition results, targeted game services are provided to users, thereby enhancing their gaming experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 the structures shown in these drawings without paying any creative work.

[0025] Figure 1 A flowchart of an intelligent identification method for game scenes according to the present invention is shown;

[0026] Figure 2 A schematic diagram of a process for obtaining the target interface feature based on the first interface feature in an intelligent recognition method for a game scene of the present invention;

[0027] Figure 3 A schematic diagram of a process of adding the first scene image set to the preset scene image in an intelligent recognition method for a game scene of the present invention;

[0028] Figure 4A schematic diagram of a process for storing the track map database into the scene recognition model in an intelligent recognition method for a game scene according to the present invention;

[0029] Figure 5 A schematic diagram of a process for determining the key frame image set based on the target feature value difference curve in an intelligent recognition method for a game scene of the present invention;

[0030] Figure 6 A schematic diagram of the operating environment of the intelligent recognition program for game scenes of the present invention;

[0031] Figure 7 This is a program module diagram of the intelligent recognition program for game scenes of the present invention.

[0032] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0033] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0034] The present invention provides an intelligent recognition method for game scenes.

[0035] like Figure 1 As shown, Figure 1 The figure is a flow chart of an intelligent identification method for game scenes according to the present invention.

[0036] In this embodiment, the method includes:

[0037] Step S100: Analyzing the non-game interface and the game interface of the target car racing game in sequence, obtaining target non-game interface features and target game interface features respectively, and forming target interface features;

[0038] Step S200: obtaining a preset text area based on the target interface features, and acquiring text display logic of the preset text area;

[0039] Step S300: performing multi-feature recognition on a first text region of a first game image to obtain first text feature information, and combining the text display logic to obtain first text content;

[0040] Step S400: determining whether the first game image belongs to a preset scene image based on the first text content; if not, calling a preset scene recognition solution;

[0041] Step S500: collecting first hue information and first texture information of the first game image based on the preset scene recognition scheme;

[0042] Step S600: constructing a scene recognition model, and inputting the first hue information and the first texture information into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map.

[0043] The target racing game refers to any racing game in which the scenes in the game are to be intelligently identified by an intelligent recognition system. First, by classifying and analyzing the non-game interface and game interface of the target racing game, the target non-game interface features and target game interface features are obtained respectively, and together they constitute the target interface features. Among them, the non-game interface includes all interfaces in the target racing game that are not in the official game state. Exemplary examples include the game settings interface, game user interface, in-game discussion area interface, game activity area interface, etc. The game interface refers to all interfaces of the game user in the official game state, including game interfaces of different competition types and different competition maps. Exemplary examples include the scene interface of users racing in individual qualifying and team qualifying, and the scene interface of users racing in the Madagascar track and the Hokkaido track, etc., all of which are game interfaces. By analyzing the game interface and non-game interface of the target racing game respectively, the basic setting rule characteristics of the game interface, that is, the target interface characteristics, are obtained, which achieves the goal of providing a basis for subsequent intelligent analysis of the text content in the game interface image, and achieves the technical effect of improving the accuracy of text recognition in the game interface image, thereby providing a basis for scene recognition.

[0044] Then, based on the target interface features obtained through analysis, the preset text areas in each game scene can be determined, and the text display logic for the preset text areas can be obtained. The text areas in different game scene interfaces vary, and the same game scene may have multiple text areas. For example, when user A is playing car racing game B, there is navigation bar text on the left side of the game settings page and user information text on the upper right side. The text display logic refers to the text specifications for displaying text in different preset areas in different game interfaces. For example, when user A is playing car racing game B, there is navigation bar text on the left side of the game settings page. The navigation bar text is in size 2, light gray regular font, with a corresponding line spacing of 1.5 times. There is user information text in size 4, white Song font, with a corresponding line spacing of 1 time, and so on. By performing targeted rule statistical analysis on the text areas in each interface image, the text display logic for the preset text areas is obtained, which provides a basis for subsequent intelligent recognition of text areas and achieves the technical effect of improving the accuracy of text recognition in game interface images.

[0045] Next, a random image of the user's game interface is obtained and recorded as the first game image. The corresponding text area, i.e., the first text area, is analyzed and determined, and multi-feature recognition is performed on it to obtain first text feature information. The first text content is obtained by combining the text display logic. The first text content includes all text-related content information in the first text area, specifically including text color, text font, text line spacing, text font size, and text content meaning, etc. The comprehensive, complete, and accurate first text content is obtained through collection and analysis, providing a reliable information basis for the subsequent judgment of whether the first game image belongs to a game scene or a non-game scene, thereby achieving the technical effect of improving the accuracy of scene type judgment.

[0046] Next, based on the first text content, it is determined whether the first game image belongs to a preset scene image. The preset scene image refers to all non-game scene images in the target racing game, including the game setting interface, the game activity interface, the game lobby interface, and the like. When the first game image does not belong to the preset scene image, it proves that the first game image is an interface image of an ongoing game, and at this time, the intelligent recognition system automatically calls the preset scene recognition scheme. The preset scene recognition scheme refers to a method for intelligently analyzing and identifying different game track maps in the target racing game. By dividing the scene interface in the target racing game into non-game interface scenes and game interface scenes, targeted scene recognition is performed for different types of scenes, ultimately achieving the technical effect of improving the accuracy of scene recognition.

[0047] Furthermore, based on the preset scene recognition scheme, the first color tone information and the first texture information of the first game image are collected, and then the first color tone information and the first texture information are input into the scene recognition model, and the scene recognition model automatically analyzes and obtains the first scene recognition result of the first game image. Among them, the scene recognition model refers to an intelligent model that automatically traverses and compares relevant database information to determine the game track situation corresponding to the first game image. Among them, the first scene recognition result includes the first track map. By constructing a scene recognition model, the technical goal of providing a model basis for intelligently determining the scene situation is achieved. The scene recognition model is used to intelligently traverse and match the color tone information and texture information of the game image, and finally determine the track map with the highest matching degree with the first game image, and output it intelligently, thereby achieving the technical effect of intelligently analyzing the user's game scene and identifying the track map corresponding to the scene.

[0048] like Figure 2As shown, in this embodiment, the non-game interface and the game interface of the target racing game are analyzed in sequence to obtain target non-game interface features and target game interface features, respectively, and form target interface features, including:

[0049] Sequentially collecting non-game interfaces of the target car racing game to obtain a target non-game interface image set;

[0050] The game interface of the target racing game is collected to obtain a target game interface image set, and the target interface image set is combined with the target non-game interface image set to form a target interface image set;

[0051] Acquire the first interface image in the target interface image set;

[0052] Dividing the first interface image to obtain a first division result, wherein the first division result includes a first text range and a first non-text range;

[0053] Obtaining coordinate information of the first text range and the first non-text range in sequence, and recording them as first text position features and first non-text position features;

[0054] using the first text position feature and the first non-text position feature as the first interface feature of the first interface image;

[0055] The target interface feature is obtained based on the first interface feature.

[0056] In this embodiment, the first interface image is divided to obtain a first division result, wherein the first division result includes a first text range and a first non-text range, including:

[0057] Get the preset unit length;

[0058] Performing raster segmentation on the first interface image based on the preset unit length to obtain a first segmentation result, wherein the first segmentation result includes a plurality of segmentation blocks;

[0059] Obtaining a first segmented block based on the multiple segmented blocks, and determining whether the first segmented block meets a preset block requirement;

[0060] If the result is not in compliance, performing a first mark on the first segmented block; if the result is not in compliance, performing a second mark on the first segmented block;

[0061] The segmentation block of the first mark is added to the first text range, and the segmentation block of the second mark is added to the first non-text range.

[0062] Screenshots of non-game interfaces in the target racing game are sequentially taken to obtain a target non-game interface image set. For example, screenshots of the game user information interface, game personalization interface, and game notification interface are taken. Next, the game interface of the target racing game is captured to obtain a target game interface image set. For example, screenshots of different track scenes and different race modes are taken. The target non-game interface image set and the target game interface image set are then combined to form a target interface image set. Next, any image in the target interface image set is randomly extracted to obtain a first interface image. The first interface image is then segmented to obtain a first segmentation result, where the first segmentation result includes a first text range and a first non-text range. Next, coordinate information of the first text range and the first non-text range is sequentially obtained and recorded as a first text position feature and a first non-text position feature, respectively. Finally, the first text position feature and the first non-text position feature are used as the first interface feature of the first interface image. The interface features of all images in the target interface image set are collectively used as the target interface feature.

[0063] When segmenting the first interface image to obtain a first segmentation result, the first step is to analyze all text font size information in the target racing game to obtain the font size of any text in the game, and then calculate the length information of the font size and use it as the preset unit length. Next, the first interface image is rasterized and segmented based on the preset unit length to obtain a first segmentation result, wherein the first segmentation result includes multiple segments. Each of the multiple segments is then analyzed and determined to determine whether the content in each segment meets the preset block requirements. If the corresponding segment meets the preset block requirements, it is marked with a first mark; if not, it is marked with a second mark. The preset block requirement refers to the presence of text information in the corresponding block. In other words, each segment is determined to determine whether it contains text content, and segments with text content are marked with one mark, while segments without text content are marked with another mark. Finally, the segments with the first mark are added to the first text range, and the segments with the second mark are added to the first non-text range. Furthermore, by performing multiple segmentations and multiple judgment analyses of the first interface image based on different text sizes and lengths, the first interface image is precisely delineated. This makes the resulting first text range and first non-text range more accurate, reliable, and more consistent with actual conditions. This provides a technical foundation for subsequent game scene recognition.

[0064] like Figure 3As shown, in this embodiment, before determining whether the first game image belongs to a preset scene image based on the first text content, and if not, calling the preset scene recognition solution, the method further includes:

[0065] Screening the target interface features based on the target non-game interface image set to obtain target non-game interface features;

[0066] Wherein, the target non-game interface features include multiple groups of non-game interface features;

[0067] Obtaining a first non-game interface feature from the multiple groups of non-game interface features, and reversely obtaining a first scene image set corresponding to the first non-game interface feature;

[0068] The first scene image set is added to the preset scene image.

[0069] After acquiring the target non-game interface image set for the target racing game, the target interface features are screened based on the target non-game interface features to obtain all interface features of the non-game interface in the game, namely, the target non-game interface features. Specifically, the target non-game interface features include multiple groups of non-game interface features. For example, the features of the user's personal information interface in the game constitute one group of non-game interface features, while the features of the settings section in the game constitute another group of non-game interface features. Next, each of the multiple groups of non-game interface features is analyzed sequentially, for example, analyzing the first non-game interface feature and reversing the analysis to obtain a first set of scene images corresponding to the first non-game interface feature. Finally, the first set of scene images is added to the preset scene images. In other words, all non-game interface scene images in the target racing game serve as the preset scene images, providing a benchmark for determining whether a first interface image is a non-game interface image. This achieves the technical effect of improving the accuracy of intelligent judgment of non-game and game scenes in the game.

[0070] like Figure 4 As shown, in this embodiment, the above-mentioned construction of the scene recognition model includes:

[0071] Obtaining an arbitrary track map and obtaining a patrol video of the arbitrary track map;

[0072] Extracting key frames from the patrol video to obtain a key frame image set;

[0073] Acquire a first key frame in the key frame image set, and analyze to obtain first key frame information of the first key frame, wherein the first key frame information includes first key frame hue information and first key frame texture information;

[0074] Calculating the mean of the first key frame hue information and the first key frame texture information in sequence to obtain a key frame hue mean and a key frame texture mean, respectively;

[0075] The key frame hue mean, the key frame texture mean, and the arbitrary track map have a first mapping relationship;

[0076] Building a track map database based on the key frame hue mean, the key frame texture mean and the first mapping relationship between the key frame hue mean and the arbitrary track map;

[0077] The track map database is stored in the scene recognition model.

[0078] First, a random track map in the target racing game is randomly acquired and filmed from the user's actual gameplay perspective to obtain a filmed video. Keyframes from the filmed video are then intelligently extracted, and all keyframes are combined into a keyframe image set. Each keyframe image in the keyframe image set is then analyzed sequentially to obtain its corresponding image information, namely, first keyframe information. The first keyframe information includes first keyframe hue information and first keyframe texture information. Next, the hue information of all keyframe images in the keyframe image set is averaged to obtain a keyframe hue mean. Simultaneously, the texture information of all keyframe images in the keyframe image set is averaged to obtain a keyframe texture mean. The keyframe hue mean and the keyframe texture mean have a first mapping relationship with the random track map. Finally, based on the keyframe hue mean, the keyframe texture mean, and the first mapping relationship between the keyframe hue mean and the random track map, a database of the random track map is constructed. The databases of all tracks in the game are combined to obtain the track map database, which is then stored in the scene recognition model. By capturing patrol videos from the user's perspective, we provide more realistic game-perspective images for subsequent intelligent recognition of user scenarios, thereby improving the accuracy of scene and track recognition. By compressing the patrol videos, we improve the overall system performance and matching efficiency. By building a track map database and storing it in the scene recognition model, we provide a traversal database foundation for subsequent scene recognition of any game scene image.

[0079] like Figure 5 As shown, in this embodiment, the key frame extraction is performed on the patrol video to obtain a key frame image set, including:

[0080] Compressing the patrol video to obtain a compressed patrol video;

[0081] Obtaining a target image frame set based on the compressed patrol video, wherein the target image frame set includes multiple image frames;

[0082] Obtaining a first image frame from the plurality of image frames, and decoding the first image frame to obtain a feature value;

[0083] A target eigenvalue difference curve is drawn based on the first image frame eigenvalue, and the key frame image set is determined based on the target eigenvalue difference curve.

[0084] First, the patrol video is compressed to obtain a compressed patrol video, for example, the patrol video is automatically compressed using a dynamic image expert group. Then, computer technology is used to extract key video image frames from the compressed patrol video, and all extracted key video image frames constitute the target image frame set. That is, the target image frame set includes multiple image frames. Further, each image frame in the target image frame set is analyzed and calculated in turn to obtain the true feature values ​​of all images, and then the difference between the feature values ​​of two adjacent images is calculated, and a target feature value difference curve is drawn. Finally, the key frame image set is determined based on the peak and valley information in the target feature value difference curve. That is, the peak value in the target feature value difference curve corresponds to a large difference in the feature values ​​of the two image frames, which in turn indicates that the two image frames have a large difference. Therefore, they cannot be used as information replacement frames for each other. All corresponding image frames are important frames and need to be analyzed and discussed in a targeted manner.

[0085] In this embodiment, after obtaining the first key frame in the key frame image set and analyzing to obtain first key frame information of the first key frame, wherein the first key frame information includes first key frame hue information and first key frame texture information, the following further includes:

[0086] determining a first frame code of the first key frame;

[0087] The first frame code has a second mapping relationship with the first key frame tone information and the first key frame texture information;

[0088] Constructing a smart positioning list based on the first key frame tone information, the first key frame texture information and the second mapping relationship between the first key frame tone information and the first frame code;

[0089] The smart positioning list is stored in the track map database.

[0090] After all key frames are extracted, each key frame is encoded, that is, the first frame code of the first key frame is obtained. Among them, the first frame code is the unique code of the first key frame, so the first frame code has a mapping relationship with the first key frame color information and the first key frame texture information, that is, it has a second mapping relationship. Finally, the first key frame color information, the first key frame texture information and the second mapping relationship between it and the first frame code are sorted to obtain the smart positioning list, and the smart positioning list is stored in the track map database. After the intelligent recognition system automatically analyzes and determines the first track map corresponding to the first game image, the first game image is matched again with the information of each key frame in the first track map, wherein the key frame with the best matching color and texture is used as the position recognition result of the first game image, and the frame code of the best matching key frame is obtained in reverse order to achieve rapid positioning of the first game image in the first track map.

[0091] The present invention sequentially analyzes the non-game interface and game interface of a target racing game, obtaining target non-game interface features and target game interface features, respectively, and forming target interface features. Based on the target interface features, a preset text area is obtained, and text display logic for the preset text area is acquired. Multi-feature recognition is performed on the first text area of ​​a first game image to obtain first text feature information, and first text content is obtained in combination with the text display logic. Based on the first text content, a determination is made as to whether the first game image belongs to a preset scene image, and if not, a preset scene recognition scheme is invoked. Based on the preset scene recognition scheme, first hue information and first texture information of the first game image are collected. A scene recognition model is constructed, and the first hue information and the first texture information are input into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map. Compared to existing technologies, the present invention can intelligently analyze and recognize game scenes based on computer science and technology, thereby improving the timeliness and accuracy of game scene recognition. Based on the timely and reliable scene recognition results, targeted game services are provided to users, thereby enhancing their gaming experience.

[0092] The present invention provides an intelligent recognition program for game scenes.

[0093] See also Figure 6 , is a schematic diagram of the operating environment of the intelligent recognition program 60 of the game scene of the present invention.

[0094] In this embodiment, the game scene intelligent recognition program 60 is installed and executed in an electronic device 6. The electronic device 6 may be a computing device such as a desktop computer, a notebook computer, a PDA, or a server. The electronic device 6 may include, but is not limited to, a memory 61, a processor 62, and a display 63. Figure 6 The electronic device 6 is shown only with components 11 - 13 , but it should be understood that implementing all of the illustrated components is not a requirement, and greater or fewer components may alternatively be implemented.

[0095] In some embodiments, the memory 61 may be an internal storage unit of the electronic device 6, such as a hard disk or memory of the electronic device 6. In other embodiments, the memory 61 may also be an external storage device of the electronic device 6, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 6. Furthermore, the memory 61 may include both an internal storage unit of the electronic device 6 and an external storage device. The memory 61 is used to store application software installed in the electronic device 6 and various types of data, such as the program code of the intelligent recognition program 60 for game scenes. The memory 61 may also be used to temporarily store data that has been output or is about to be output.

[0096] In some embodiments, the processor 62 may be a central processing unit (CPU), a microprocessor, or other data processing chip, configured to execute program codes or process data stored in the memory 61, such as executing the intelligent recognition program 60 for game scenes.

[0097] In some embodiments, the display 63 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 63 is used to display information processed by the electronic device 6 and to display a visual user interface. Components 11-13 of the electronic device 6 communicate with each other via a program bus.

[0098] See also Figure 7 , is a program module diagram of the intelligent recognition program 60 of the game scene of the present invention.

[0099] In this embodiment, the intelligent recognition program 60 for game scenes can be divided into one or more modules, one or more modules are stored in the memory 61, and are executed by one or more processors (processor 62 in this embodiment) to complete the present invention. Figure 7In the example, the game scene intelligent recognition program 60 can be divided into a feature building module 701, a logic acquisition module 702, a content acquisition module 703, a solution calling module 704, an information collection module 705, and a scene recognition module 706. The modules described in the present invention refer to a series of computer program instruction segments that can perform specific functions. They are more suitable for describing the execution process of the game scene intelligent recognition program 60 in the electronic device 6 than programs.

[0100] Feature building module 701: sequentially analyzing the non-game interface and the game interface of the target racing game, obtaining target non-game interface features and target game interface features, and forming target interface features;

[0101] Logic acquisition module 702: obtains a preset text area based on the target interface features, and acquires text display logic of the preset text area;

[0102] Content acquisition module 703: performs multi-feature recognition on the first text area of ​​the first game image to obtain first text feature information, and obtains first text content in combination with the text display logic;

[0103] Solution calling module 704: determining whether the first game image belongs to a preset scene image based on the first text content, and if not, calling a preset scene recognition solution;

[0104] Information collection module 705: collects first color tone information and first texture information of the first game image based on the preset scene recognition scheme;

[0105] Scene recognition module 706: constructs a scene recognition model, and inputs the first hue information and the first texture information into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map.

[0106] The present application also provides an electronic device, which includes a processor and a memory;

[0107] The processor is configured to process the steps of executing the method for intelligently identifying a game scene as described in any one of the first embodiments above;

[0108] The memory is coupled to the processor and is used to store a program. When the intelligent recognition program for game scenes is executed by the processor, the system executes the steps of any of the above-mentioned intelligent recognition methods for game scenes.

[0109] Furthermore, the present invention also proposes a computer-readable storage medium, which stores an intelligent recognition program for game scenes. The intelligent recognition program for game scenes can be executed by at least one processor so that the at least one processor executes the intelligent recognition method for game scenes in any of the above embodiments.

[0110] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made by using the contents of the present invention description and drawings under the inventive concept of the present invention, or direct / indirect application in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A method for intelligently identifying game scenes, characterized in that: include: Analyze the non-game interface and the game interface of the target racing game in sequence, obtain target non-game interface features and target game interface features respectively, and form the target interface features; Obtaining a preset text area based on the target interface feature, and obtaining text display logic for the preset text area, wherein the text display logic refers to text specifications for displaying text in different preset areas under different game interfaces; Performing multi-feature recognition on the first text area of ​​the first game image to obtain first text feature information, and combining the text display logic to obtain first text content; determining, based on the first text content, whether the first game image belongs to a preset scene image, wherein the preset scene image refers to all non-game scene images in the target racing game; and if not, invoking a preset scene recognition solution, wherein the preset scene recognition solution refers to a method for intelligently analyzing and recognizing different game track maps in the target racing game; collecting first hue information and first texture information of the first game image based on the preset scene recognition scheme; Building a scene recognition model, and inputting the first hue information and the first texture information into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map; The target non-game interface and the game interface of the racing game are analyzed in sequence to obtain target non-game interface features and target game interface features respectively, and form target interface features, including: Sequentially collecting non-game interfaces of the target car racing game to obtain a target non-game interface image set; The game interface of the target racing game is collected to obtain a target game interface image set, and the target interface image set is combined with the target non-game interface image set to form a target interface image set; Acquire the first interface image in the target interface image set; Dividing the first interface image to obtain a first division result, wherein the first division result includes a first text range and a first non-text range; Sequentially obtaining coordinate information of the first text range and the first non-text range, which are recorded as first text position features and first non-text position features; using the first text position feature and the first non-text position feature as the first interface feature of the first interface image; Obtaining the target interface feature based on the first interface feature; The constructing of the scene recognition model includes: Obtaining an arbitrary track map and obtaining a patrol video of the arbitrary track map; Extracting key frames from the patrol video to obtain a key frame image set; Acquire a first key frame in the key frame image set, and analyze to obtain first key frame information of the first key frame, wherein the first key frame information includes first key frame hue information and first key frame texture information; Calculating the mean of the first key frame hue information and the first key frame texture information in sequence to obtain a key frame hue mean and a key frame texture mean, respectively; The key frame hue mean, the key frame texture mean, and the arbitrary track map have a first mapping relationship; Building a track map database based on the key frame hue mean, the key frame texture mean and the first mapping relationship between the key frame hue mean and the arbitrary track map; The track map database is stored in the scene recognition model.

2. The intelligent recognition method for game scenes according to claim 1, characterized in that: The first interface image is divided to obtain a first division result, wherein the first division result includes a first text range and a first non-text range, including: Get the preset unit length; Performing raster segmentation on the first interface image based on the preset unit length to obtain a first segmentation result, wherein the first segmentation result includes a plurality of segmentation blocks; Obtaining a first segmented block based on the multiple segmented blocks, and determining whether the first segmented block meets a preset block requirement; If the result is not in compliance, performing a first mark on the first segmented block; if the result is not in compliance, performing a second mark on the first segmented block; The segmentation block of the first mark is added to the first text range, and the segmentation block of the second mark is added to the first non-text range.

3. The intelligent recognition method for game scenes according to claim 1, characterized in that: Before determining whether the first game image belongs to a preset scene image based on the first text content, and if not, calling a preset scene recognition solution, the method further includes: Screening the target interface features based on the target non-game interface image set to obtain target non-game interface features; Wherein, the target non-game interface features include multiple groups of non-game interface features; Obtaining a first non-game interface feature from the multiple groups of non-game interface features, and reversely obtaining a first scene image set corresponding to the first non-game interface feature; The first scene image set is added to the preset scene image.

4. The intelligent recognition method for game scenes according to claim 1, characterized in that: The key frame extraction is performed on the patrol video to obtain a key frame image set, including: Compressing the patrol video to obtain a compressed patrol video; Obtaining a target image frame set based on the compressed patrol video, wherein the target image frame set includes multiple image frames; Obtaining a first image frame from the plurality of image frames, and decoding the first image frame to obtain a feature value; A target eigenvalue difference curve is drawn based on the first image frame eigenvalue, and the key frame image set is determined based on the target eigenvalue difference curve.

5. The intelligent recognition method for game scenes according to claim 1, characterized in that: After acquiring the first key frame in the key frame image set and analyzing to obtain first key frame information of the first key frame, wherein the first key frame information includes first key frame hue information and first key frame texture information, the method further includes: determining a first frame code of the first key frame; The first frame code has a second mapping relationship with the first key frame tone information and the first key frame texture information; Constructing a smart positioning list based on the first key frame tone information, the first key frame texture information and the second mapping relationship between the first key frame tone information and the first frame code; The smart positioning list is stored in the track map database.

6. An intelligent recognition system for game scenes, comprising a memory and a processor, characterized in that: The memory stores an intelligent recognition program for game scenes, and when the processor executes the intelligent recognition program for game scenes, the following steps are implemented: Analyze the non-game interface and the game interface of the target racing game in sequence, obtain target non-game interface features and target game interface features respectively, and form the target interface features; Obtaining a preset text area based on the target interface feature, and obtaining text display logic for the preset text area, wherein the text display logic refers to text specifications for displaying text in different preset areas under different game interfaces; Performing multi-feature recognition on the first text area of ​​the first game image to obtain first text feature information, and combining the text display logic to obtain first text content; determining, based on the first text content, whether the first game image belongs to a preset scene image, wherein the preset scene image refers to all non-game scene images in the target racing game; and if not, invoking a preset scene recognition solution, wherein the preset scene recognition solution refers to a method for intelligently analyzing and recognizing different game track maps in the target racing game; collecting first hue information and first texture information of the first game image based on the preset scene recognition scheme; Building a scene recognition model, and inputting the first hue information and the first texture information into the scene recognition model to obtain a first scene recognition result, wherein the first scene recognition result includes a first track map; The target non-game interface and the game interface of the racing game are analyzed in sequence to obtain target non-game interface features and target game interface features respectively, and form target interface features, including: Sequentially collecting non-game interfaces of the target car racing game to obtain a target non-game interface image set; The game interface of the target racing game is collected to obtain a target game interface image set, and the target interface image set is combined with the target non-game interface image set to form a target interface image set; Acquire the first interface image in the target interface image set; Dividing the first interface image to obtain a first division result, wherein the first division result includes a first text range and a first non-text range; Sequentially obtaining coordinate information of the first text range and the first non-text range, which are recorded as first text position features and first non-text position features; using the first text position feature and the first non-text position feature as the first interface feature of the first interface image; Obtaining the target interface feature based on the first interface feature; The constructing of the scene recognition model includes: Obtaining an arbitrary track map and obtaining a patrol video of the arbitrary track map; Extracting key frames from the patrol video to obtain a key frame image set; Acquire a first key frame in the key frame image set, and analyze to obtain first key frame information of the first key frame, wherein the first key frame information includes first key frame hue information and first key frame texture information; Calculating the mean of the first key frame hue information and the first key frame texture information in sequence to obtain a key frame hue mean and a key frame texture mean, respectively; The key frame hue mean, the key frame texture mean, and the arbitrary track map have a first mapping relationship; Building a track map database based on the key frame hue mean, the key frame texture mean and the first mapping relationship between the key frame hue mean and the arbitrary track map; The track map database is stored in the scene recognition model.

7. A computer device, characterized in that: including processor and memory; The processor is configured to process and execute the method according to any one of claims 1 to 5; The memory is coupled to the processor and is used to store an intelligent recognition program for game scenes. When the intelligent recognition program for game scenes is executed by the processor, the system executes the steps of the method described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores an intelligent recognition program for game scenes, and the intelligent recognition program for game scenes can be executed by at least one processor to enable the at least one processor to perform the steps of intelligent recognition of game scenes as described in any one of claims 1-5.

Images