AI cloud intelligence fun commemorative screen and commemorative method for religious culture inheritance

Abstract

The invention relates to an A I cloud intelligence fun memorial screen for religious cultural inheritance, which comprises a wall-mounted display terminal and a mobile terminal connected with the display terminal by utilizing the Internet, an APP connected with the display terminal is installed on the mobile terminal, the display terminal comprises a base and a display terminal fixed on the base, and the display terminal is connected with the mobile terminal. A power management module is arranged in the base, a display screen, a high-definition camera and a plurality of control keys are embedded in the front surface of the display terminal, and a network connection module, a timing startup and shutdown module, a cloud storage module, a user authority management module, a remote management module, a management module and an AI compatible module are arranged in the display terminal. Ai learning is utilized to enable the sound and the image of the digital human to reach an expression mode close to that before birth, an A I technology, cloud storage, element universe application and intelligent management are combined, multi-terminal access, two-dimensional code scanning and remote operation are achieved, and use is convenient and fast.

Description

Technical Field

[0001] The present invention relates to the field of virtual interaction technology, and more specifically, to an AI cloud intelligence blessing memorial screen and memorial method for clan culture inheritance. Background Art

[0002] Clan culture is an important part of the traditional culture of the Chinese nation, carrying the core contents of family history, ethics, family traditions and family precepts. In modern society, with the changes in lifestyle and the acceleration of population mobility, the inheritance of clan culture faces many challenges, and the younger generation's understanding and identification of clan culture is gradually weakened. The existing offline ancestral hall memorial inherits the family spirit through family ancestral hall sacrificial activities, but it is difficult to cover distant family members due to the fixed venue. There are significant deficiencies in digitization, interactivity and convenience. In recent years, with the development of technologies such as cloud computing, big data, artificial intelligence (AI) and virtual reality (VR / AR), cultural inheritance and commemoration methods have gradually shifted towards digitization and intelligence. The existing virtual ancestral hall uses simple virtual reality technology to build a clan ancestral hall model, but it has a single function and fails to fully integrate AI and interactive technology. There are significant deficiencies in building immersive scenes and personalized interactions. Summary of the invention

[0003] The present invention provides an AI cloud intelligence and blessing memorial screen for the inheritance of clan culture. It utilizes three-dimensional modeling and dynamic scene technology to replicate the living places and daily scenes of the deceased, construct an immersive clan culture memorial space, and generates a digital person of the deceased through AI technology, simulating their appearance, voice and behavior, and conducting personalized emotional interaction with users.

[0004] The present invention provides an AI Yunzhi Nafu memorial screen for the inheritance of clan culture, comprising a wall-mounted display terminal and a mobile terminal connected to the display terminal by using the Internet, wherein an APP connected to the display terminal is installed on the mobile terminal, and a display screen, a high-definition camera and control buttons are inlaid on the front surface of the display terminal. The mobile terminal scans the QR code of the display screen through the APP or WeChat public account to send a request instruction for uploading photos; the display screen is divided into a plurality of display areas, wherein the display area comprises an automatic scrolling play area and a fixed display area, wherein the automatic scrolling play area is used for automatically playing photos and videos uploaded by users, and the fixed display area is used for long-term fixed display of user-defined content, wherein the size of the display screen is determined based on the best visual experience at a distance of two meters, and the display modes of the display screen include black and white display and color display;

[0005] The display terminal is provided with:

[0006] Network connection module: used for data transmission and communication with external devices, the connection methods include optical fiber line access, WiFi connection and NFC connection;

[0007] Timer power on / off module: realizes the timer power on and off operation of the display screen, and supports automatic power on and off of the device in the specified time period;

[0008] Cloud storage module: used to store and manage photos, videos and related data uploaded by users, and supports fast data migration and import between new and old devices;

[0009] User rights management module: used to control the user's rights to upload data and determine whether the user has paid the fee. The user obtains the upload rights after paying the management fee;

[0010] Remote management module: used to receive remote operation instructions from the administrator, including editing and maintaining the photo and video content stored in the device;

[0011] Management module: AI robots are used to automatically perform intelligent management in the background, including data analysis management, content optimization management, and user interaction management;

[0012] AI compatible module: used to realize the display terminal and AI application and metaverse scene functions, including data interaction unit, scene rendering unit, digital human generation unit, project interaction unit and metaverse space mapping unit, among which:

[0013] The data interaction unit is used to receive and upload multimodal data;

[0014] The scene dynamic rendering unit is used to load and adjust the memorial scene in real time;

[0015] The digital human generation unit is used to generate a digital human and optimize the emotion and behavior performance of the three-dimensional digital human model;

[0016] The item interaction unit is used for semantic recognition and intent analysis to recognize the user's voice and actions, and match the worship interaction items according to the recognition results;

[0017] The Metaverse space mapping unit is used to synchronize device data to the Metaverse platform;

[0018] Interface module: includes power interface and audio output interface. The power interface is used to connect an external power source to the display terminal. Users can connect an external speaker device through the audio output interface according to their needs.

[0019] Preferably, the AI ​​compatible module further includes a project interaction unit, wherein:

[0020] Scene dynamic rendering unit: used to build memorial scenes through multidimensional function space, and the scene template selected by the user through the official account, based on the scene template selected by the user, determine the element set contained in the target scene template; according to the scene template, generate a dynamic scene containing the target element, and render the memorial scene in the metaverse environment; the memorial scene includes a family hall, ancestral hall and landscape courtyard, and selects a preset scene according to the season and regional information, and displays the selected scene on the display screen;

[0021] A digital human generation unit: used to obtain two-dimensional photo data uploaded by a user through a mobile terminal; compare the rendered image of the initial three-dimensional model with the user's photo through an image projection method; optimize the three-dimensional model parameters based on the mean square error so that the rendered image is close to the user's photo; output the optimized three-dimensional appearance model; define a preset basic action set, which includes facial expressions and body movements; assign a time dynamic weight function to each preset action according to the scene requirements selected by the user, generate a continuous action sequence using a weighted combination method, and adjust the order and intensity of the actions to match the scene logic; extract the audio features of the voice data uploaded through the mobile terminal (2), generate a voice model using a recurrent neural network, and output a voice signal with the user's voice features according to the user's scene interaction requirements; process the text description provided by the user through a natural language generation model to generate a personalized expression;

[0022] The project interaction unit is used to obtain the user's voice information through a microphone, obtain the user's action information through a high-definition camera (13), extract the voice information and represent it as a time series, extract the action information and represent it as an action feature sequence, extract the semantic content from the voice features through a preset semantic analysis model, and recognize the action content from the action features through a preset action recognition model; match the most relevant interaction project in the worship interaction project library according to the results of semantic analysis and action recognition, and map the matched worship interaction project to the display space of the display module.

[0023] Preferably, the display terminal supports plane display, immersive display and enhanced display, wherein the plane display is to display the interactive scene on the display screen; the immersive display is to display the interactive scene in a three-dimensional virtual space through VR / AR equipment; the enhanced display is to display the digital human and scene elements superimposed on the real environment through the mobile AR function; the display terminal is used to receive the scene data and adapt it to a format that can be displayed by the terminal device, and at the same time, according to the output of the voice and behavior interaction module, update the digital human's actions, expressions and interactive effects in the scene in real time.

[0024] The present invention also provides an AI cloud wisdom blessing commemoration method for clan culture inheritance, based on the aforementioned AI cloud wisdom blessing commemoration screen for clan culture inheritance, comprising the following steps:

[0025] Scene building step: used to build a memorial scene through a multidimensional function space, the scene template selected by the user through the control button, based on the scene template selected by the user, determine the element set contained in the target scene template; according to the scene template, generate a dynamic scene containing the target element, and render the memorial scene in the metaverse environment; the memorial scene includes a family hall, an ancestral hall and a landscape courtyard, and select a preset scene according to the season and regional information, and display the selected scene through the display screen;

[0026] Item identification step: receiving item data uploaded by the user, the item data includes image, audio and text formats, generating a feature vector through a feature extraction algorithm, the feature vector includes embedded features of the item image extracted based on a deep learning model; extracting feature parameters of the audio based on spectrum analysis, extracting semantic features of the text using natural language processing technology, matching the extracted user item feature vector with the template feature vector in the database, querying the matching template item in the database according to the matching judgment result, and returning the corresponding item information. If there is no matching item, returning an unmatched result and prompting the user to re-enter or adjust the data. If the user agrees, entering the co-creation process, dynamically updating the template feature vector set in the database based on user feedback or newly added item data during the matching process;

[0027] Digital human image generation steps: used to obtain two-dimensional photo data uploaded by the user through the mobile terminal; compare the rendered image of the initial three-dimensional model with the user's photo through the image projection method; optimize the three-dimensional model parameters based on the mean square error so that the rendered image is close to the user's photo; output the optimized three-dimensional appearance model; define a preset basic action set, which includes facial expressions and body movements; according to the scene requirements selected by the user, assign a time dynamic weight function to each preset action, use the weighted combination method to generate a continuous action sequence, and adjust the order and intensity of the actions to match the scene logic;

[0028] Digital human voice expression generation step: used to extract the audio features of the voice data uploaded through the mobile terminal, generate a voice model using a recurrent neural network, and output a voice signal with user voice features according to the user's scenario interaction requirements; process the text description provided by the user through a natural language generation model to generate personalized expressions;

[0029] Interaction steps: used to obtain the user's voice information through the microphone, obtain the user's action information through the high-definition camera, extract the sound information and represent it as a time series, extract the action information and represent it as an action feature sequence, extract the semantic content from the sound features through a preset semantic analysis model, and recognize the action content from the action features through a preset action recognition model; match the most relevant interaction items in the worship interaction item library according to the results of semantic analysis and action recognition, and map the matched worship interaction items to the display space of the display module.

[0030] Preferably, in the item identification step, a feature vector is extracted from the item data uploaded by the user. Matching with the template feature F in the database is expressed as:

[0031] Matching determination

[0032] Among them, d(F,F i ) is the distance metric between feature vectors, ∈ is the matching threshold;

[0033] If the features uploaded by the user are not matched successfully, the co-creation process will be entered, which is represented as follows:

[0034] Feature supplement F′=F+ΔF

[0035] Where ΔF is the feature added by the second user;

[0036] The newly added template F′ is added to the database, represented as:

[0037] Preferably, in the scene generation step, the scene S (x, y, z, t) is represented as a function of the evolution of the three-dimensional space (x, y, z) over time t, expressed as:

[0038]

[0039] Among them, f i (x, y, z) represents the spatial distribution of the i-th element in the scene; g i (t) represents the temporal dynamics of the element;

[0040] Through user-selected scene templates The template is represented by the following formula:

[0041] T(x,y,z,t)={S i (x,y,z,t)|i∈I}

[0042] Where I is the element index set contained in the template.

[0043] Preferably, the f is generated by a preset geometric model or data uploaded by the user. i (x, y, z), the generation steps include: obtaining the three-dimensional shape, position and size parameters of each scene element, mapping the parameters to the distribution function f in three-dimensional space i (x, y, z), used to represent the spatial characteristics of an element;

[0044] Define the time dynamic function g i Step (t) includes: specifying dynamic behavior for each scene element, calculating g using interpolation or dynamic weights i The value of (t) is used to describe the state changes of elements at different time points;

[0045] Preferably, in the digital human image generation step, the appearance of the digital human is represented by a model M(x, y, z), where M is an approximation function based on the user uploaded photo:

[0046]

[0047] Where P(M) is the image obtained by projecting the 3D model onto a 2D plane; D is the photo data uploaded by the user;

[0048] The behavior of the digital human is represented by the action function A(t):

[0049]

[0050] Among them: a j (t) is the time weight function of the action; φ j Is a collection of predefined action bases.

[0051] Preferably, in the digital voice expression generation step, the voice signal V(t) and the text expression T are generated by an AI model trained by user uploaded data:

[0052]

[0053] Where: X is the audio dataset uploaded by the user;

[0054] Y is the text or biographical description data provided by the user;

[0055] and is the generating function trained by deep learning.

[0056] Preferably, in the interaction step,

[0057] Target user voice information V u (t) is expressed as a time series: V u (t)={v1,v2,…,v k}; where vi is the time t i The sound characteristics of

[0058] The target user action information Au(t) is represented as an action feature sequence: A u (t)={a1,a2,…,a m}; where a j is the time t j The action characteristics on

[0059] Define the semantic analysis model as L v , from the sound feature V u (t) Extracting semantic content Expressed as:

[0060]

[0061] Define the action recognition model as L a , from the action feature A u (t) Identify the action content Expressed as:

[0062]

[0063] Define the worship interaction project library as According to semantic analysis and action recognition The result matches the most relevant interaction item P * Expressed as:

[0064]

[0065] in It is the relevance scoring function between interaction items and semantic and action matching;

[0066] The matching worship interaction item P * Mapped to the display space ε of the electronic sacrificial device, it is expressed as:

[0067]

[0068] in: The interactive item P * A function embedded in the interaction field G′(x,y,z,t) and displayed on the display module; (x',y',z') are the display coordinates on the electronic device.

[0069] The beneficial effect of the present invention is that the present invention provides an AI cloud wisdom blessing memorial screen for the inheritance of clan culture. By combining AI technology, cloud storage, metaverse application and intelligent management, the network connection module supports fiber access, WiFi connection and NFC connection, meets the needs of fast data transmission in different network environments, and ensures that the device can operate stably in various scenarios. The timer switch module supports automatic switch in a set time period (such as night to seven in the morning), reduces energy consumption, reduces hardware loss caused by long-term operation of the device, and significantly extends the service life of the device. The cloud storage module realizes the safe storage and remote access of user data, supports the rapid migration and import of data between new and old devices, and avoids data loss caused by equipment replacement. The user authority management module ensures that the upload authority is standardized through the management fee payment mechanism, and provides free upload authority for target users to avoid resource abuse and improve management efficiency. The automatic scrolling playback area can continuously play photos and videos uploaded by users to meet the needs of multiple people sharing and watching. The fixed display area can display user-defined content for a long time, which is suitable for static commemoration needs in specific occasions. The manual search playback time is limited to 8 minutes, and the playback resources are reasonably allocated to reduce the waiting time in queues. The display screen supports color and black and white display modes. Users can choose according to scene requirements or personal preferences to adapt to the display requirements of different environments and cultural atmospheres. The screen size is determined by the optimal viewing distance of 2 meters to ensure that users get the best visual experience when watching. The management module realizes intelligent management of data analysis, content optimization and user interaction through AI robot background management, effectively improving the system's adaptability and operation and maintenance efficiency. Automatically analyze user behavior, optimize display content recommendations, and reduce manual management costs. The AI ​​compatible module integrates five major functions: data interaction, scene rendering, digital human generation, semantic recognition, and metaverse mapping, ensuring that the device can be deeply connected with AI technology and metaverse platform. The remote management module supports the management company to add, delete, modify and maintain photos and videos in the device through remote operation, realizing efficient data update and flexible management, and reducing manual maintenance costs. The interface module supports external speaker function. Users can choose audio playback equipment according to actual needs to increase the expressiveness of the memorial scene. This device can be widely used in scenes such as temples, clan ancestral halls, and above the altar at home to meet the cultural inheritance and commemoration needs of different places. This device is compatible with AI technology and metaverse technology, supports subsequent function expansion and technology iteration, and ensures the long-term sustainable development of the system. Therefore, the present invention is an integrated innovation of AI intelligent management, cloud storage, metaverse mapping and other technologies. Reduce maintenance costs through permission control, remote management and AI background operation and maintenance. Meet the needs of different scenarios through automatic scrolling, fixed display, black and white / color mode. It realizes multi-terminal access, QR code scanning, and remote operation, which is easy to use. The deep integration of digital human technology and scene rendering strengthens cultural and emotional connections.Through the efficient collaboration of the above-mentioned technical solutions and functional modules, the present invention effectively solves the problems of single information, insufficient interaction, and difficult management in the traditional clan culture inheritance method, and provides an innovative and sustainable technical solution for the digital and intelligent inheritance of Chinese clan culture. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be further described below in conjunction with the accompanying drawings and embodiments. 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 creative work:

[0071] Figure 1 This is a display terminal structure diagram of an AI cloud intelligence blessing memorial screen for clan culture inheritance according to an embodiment of the present invention;

[0072] Figure 2 It is a schematic diagram of an AI cloud intelligence blessing memorial method screen for clan culture inheritance according to an embodiment of the present invention;

[0073] Figure 3 It is a flow chart of an AI cloud intelligence blessing memorial method for clan culture inheritance according to an embodiment of the present invention. DETAILED DESCRIPTION

[0074] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the following will be described clearly and completely in combination with the technical solution in the embodiments of the present invention. Obviously, the described embodiments are partial embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0075] like Figure 1 As shown, the present invention provides an AI Yunzhi Nafu memorial screen for clan culture inheritance, including a wall-mounted display terminal 1 and a mobile terminal 2 connected to the display terminal 1 using the Internet, an APP connected to the display terminal 1 is installed on the mobile terminal 2, a display screen 12, a high-definition camera 13 and a control button 11 are inlaid on the front surface of the display terminal 1, and the mobile terminal scans the QR code of the display screen through the APP or WeChat public account to send a request instruction to upload photos; the display screen 12 is divided into multiple display areas, the display area includes an automatic scrolling playback area and a fixed display area, the automatic scrolling playback area is used to automatically play photos and videos uploaded by users, and the fixed display area is used to display user-defined content for a long time. The size of the display screen is determined based on the best visual distance of two meters, and the display mode of the display screen 12 includes black and white display and color display;

[0076] The display terminal has the following features:

[0077] Network connection module: used for data transmission and communication with external devices. Connection methods include fiber optic line access, WiFi connection and NFC connection;

[0078] Timer power on / off module: realizes the timer power on and off operation of the display screen, and supports automatic power on and off of the device in the specified time period;

[0079] Cloud storage module: used to store and manage photos, videos and related data uploaded by users, and supports fast data migration and import between new and old devices;

[0080] User rights management module: used to control the user's rights to upload data and determine whether the user has paid the fee. The user obtains the upload rights after paying the management fee;

[0081] Remote management module: used to receive remote operation instructions from the administrator, including editing and maintaining the photo and video content stored in the device;

[0082] Management module: AI robots are used to automatically perform intelligent management in the background. Intelligent management includes data analysis management, content optimization management, and user interaction management.

[0083] AI compatible module: used to realize the display terminal and AI application and metaverse scene functions, including data interaction unit, scene rendering unit, digital human generation unit, project interaction unit and metaverse space mapping unit, among which:

[0084] The data interaction unit is used to receive and upload multimodal data;

[0085] The scene dynamic rendering unit is used to load and adjust the memorial scene in real time;

[0086] The digital human generation unit is used to generate a digital human and optimize the emotion and behavior performance of the three-dimensional digital human model;

[0087] The item interaction unit is used for semantic recognition and intent analysis to recognize the user's voice and actions, and match the worship interaction items according to the recognition results;

[0088] The Metaverse space mapping unit is used to synchronize device data to the Metaverse platform;

[0089] Interface module: includes power interface and audio output interface. The power interface is used to connect an external power source to the display terminal. Users can connect an external speaker device through the audio output interface according to their needs.

[0090] By combining AI technology, cloud storage, metaverse applications and intelligent management, the network connection module supports fiber access, WiFi connection and NFC connection, meeting the needs of fast data transmission in different network environments and ensuring that the device can operate stably in various scenarios. The timer power on / off module supports automatic power on / off in a set time period (such as night to 7 a.m.), reducing energy consumption, reducing hardware loss caused by long-term operation of the device, and significantly extending the service life of the device. The cloud storage module realizes the secure storage and remote access of user data, supports the rapid migration and import of data between new and old devices, and avoids data loss caused by device replacement. The user rights management module ensures that upload permissions are standardized through the management fee payment mechanism, and provides free upload permissions for target users to avoid resource abuse and improve management efficiency. The automatic scrolling playback area can continuously play photos and videos uploaded by users to meet the needs of multiple people sharing and watching. The fixed display area can display user-defined content for a long time, which is suitable for static commemoration needs in specific occasions. The manual search playback time is limited to 8 minutes, which reasonably allocates playback resources and reduces waiting time in queues. The display screen supports color and black and white display modes. Users can choose according to scene requirements or personal preferences to adapt to the display needs of different environments and cultural atmospheres. The screen size is determined by the optimal viewing distance of 2 meters to ensure that users get the best visual experience when watching. The management module realizes intelligent management of data analysis, content optimization and user interaction through AI robot background management, effectively improving the system's adaptability and operation and maintenance efficiency. Automatically analyze user behavior, optimize display content recommendations, and reduce manual management costs. The AI ​​compatible module integrates five major functions: data interaction, scene rendering, digital human generation, semantic recognition, and metaverse mapping, ensuring that the device can be deeply connected with AI technology and metaverse platform. The remote management module supports the management company to add, delete, modify and maintain photos and videos in the device through remote operation, realize efficient data update and flexible management, and reduce manual maintenance costs. The interface module supports the external speaker function, and users can choose audio playback devices according to actual needs to increase the expressiveness of the memorial scene. This device can be widely used in scenes such as temples, clan ancestral halls, and above the altar at home to meet the cultural inheritance and commemoration needs of different places. This device has the compatibility of AI technology and metaverse technology, supports subsequent function expansion and technology iteration, and ensures the long-term sustainable development of the system. Therefore, the present invention is an integrated innovation of technologies such as AI intelligent management, cloud storage, and metaverse mapping. Reduce maintenance costs through permission control, remote management and AI background operation and maintenance. Meet the needs of different scenarios through automatic scrolling, fixed display, black and white / color mode. Realize multi-terminal access, QR code scanning, remote operation, easy to use. Deeply integrate digital human technology with scene rendering to strengthen cultural and emotional connections.Through the efficient collaboration of the above-mentioned technical solutions and functional modules, the present invention effectively solves the problems of single information, insufficient interaction, and difficult management in the traditional clan culture inheritance method, and provides an innovative and sustainable technical solution for the digital and intelligent inheritance of Chinese clan culture.

[0091] The AI-compatible module also includes a project interaction unit, where:

[0092] Scene dynamic rendering unit: used to build a memorial scene through a multidimensional function space, and the scene template selected by the user through the official account, based on the scene template selected by the user, determine the element set contained in the target scene template; according to the scene template, generate a dynamic scene containing the target element, and render the memorial scene in the metaverse environment; the memorial scene includes a family hall, an ancestral hall, and a landscape courtyard, and select a preset scene according to the season and regional information, and display the selected scene through the display screen 12;

[0093] Digital human generation unit: used to obtain two-dimensional photo data uploaded by the user through the mobile terminal 2; compare the rendered image of the initial three-dimensional model with the user's photo through the image projection method; optimize the three-dimensional model parameters based on the mean square error so that the rendered image is close to the user's photo; output the optimized three-dimensional appearance model; define a preset basic action set, which includes facial expressions and body movements; assign a time dynamic weight function to each preset action according to the scene requirements selected by the user, generate a continuous action sequence using the weighted combination method, and adjust the order and intensity of the actions to match the scene logic; extract the audio features of the voice data uploaded through the mobile terminal 2, generate a voice model using a recurrent neural network, and output a voice signal with user voice features according to the user's scene interaction requirements; process the text description provided by the user through the natural language generation model to generate personalized expressions;

[0094] Project interaction unit: used to obtain the user's voice information through the microphone, obtain the user's action information through the high-definition camera 13, extract the sound information and represent it as a time series, extract the action information and represent it as an action feature sequence, extract the semantic content from the sound features through a preset semantic analysis model, and recognize the action content from the action features through a preset action recognition model; match the most relevant interaction project in the worship interaction project library according to the results of semantic analysis and action recognition, and map the matched worship interaction project to the display space of the display module.

[0095] Scene construction: Use 3D modeling technology to build memorial scenes, such as family halls, ancestral halls, landscape courtyards, etc. Users can choose preset templates or customize scenes to combine with cultural traditions. Support dynamic scenes, such as replicas of places where deceased people have been, re-enactments of daily life scenes, etc., to enhance immersion. Build an exclusive virtual space management platform, and users can access the memorial space through mobile terminals, PC terminals or VR devices. Provide permission management functions to allow users to set access rights to virtual spaces, such as family members, clan members or public access.

[0096] Digital human generation: Using AI deep learning technology, a 3D digital human with a high resemblance to the deceased is generated through photos, audio and video materials. Dynamic adjustment functions are supported, such as repairing low-resolution photos or optimizing appearance details based on user descriptions.

[0097] Voice and expression simulation: Combining natural language processing (NLP) technology with speech synthesis technology, the voice intonation, expression and language habits of the deceased can be simulated. The AI ​​model can be trained through recording materials to give the digital person a personalized language style.

[0098] Emotional and behavioral interaction: Behavior simulation is achieved by giving digital humans common expressions, movements, and gestures, such as smiling, nodding, and shaking hands, through AI algorithms. Voice dialogue and text interaction functions are provided, and users can "talk" with digital humans to obtain stories or wisdom from the memories of old people.

[0099] The display terminal supports plane display, immersive display and enhanced display. The plane display is to display the interactive scene on the display screen; the immersive display is to display the interactive scene in a three-dimensional virtual space through VR / AR equipment; the enhanced display is to superimpose the digital human and scene elements on the real environment through the mobile AR function; the display terminal is used to receive scene data and adapt it to a format that can be displayed by the terminal device, and at the same time, according to the output of the voice and behavior interaction module, update the digital human's actions, expressions and interactive effects in the scene in real time.

[0100] The viewing distance of the display screen is optimized to 1.5-3.0 meters, suitable for home or ancestral hall use. The memorial screen is equipped with a fiber optic module, a WiFi module, and an NFC module, and supports mobile phone hotspot connection function. The memorial screen can be set to a 20 cm wide horizontal screen and supports handwriting input (name + date of birth / password) for quick search. Each screen supports the storage of more than 100,000 photos and can record text biography and resume.

[0101] Data upload supports computers, tablets, mobile phones through network cables, WiFi, NFC and other close-range methods, as well as remote upload on the official account platform. Users can pay to upload more content, including photos, text, etc., using a pay-first-then-service mechanism.

[0102] The memorial screen has the function of metaverse expansion, supports connection with memorial scenes, builds a digital human of the deceased, and simulates the image, voice and expression of the deceased. The digital human supports connection with VR / AR devices to provide relatives with metaverse experience services.

[0103] Mobile terminal users register with the backend server to ensure the uniqueness of user identity. Users use the terminal to capture the features of objects and upload photos, voice or text descriptions for background analysis. After a successful match, relevant service information is obtained; if a match fails, the user co-creation process is entered to enrich the template library through user supplementary data. Users can upload photos / videos, voice or text to supplement the feature information of unmatched objects, and the background will generate a new feature code template.

[0104] The present invention also provides an AI cloud wisdom blessing commemoration method for clan culture inheritance, such as Figure 3 As shown, based on the aforementioned AI cloud intelligence blessing memorial screen for clan culture inheritance, the following steps are included:

[0105] Scene building step: used to build a memorial scene through a multidimensional function space, the scene template selected by the user through the control button, based on the scene template selected by the user, determine the element set contained in the target scene template; according to the scene template, generate a dynamic scene containing the target element, and render the memorial scene in the metaverse environment; the memorial scene includes a family hall, an ancestral hall and a landscape courtyard, and select a preset scene according to the season and regional information, and display the selected scene through the display screen;

[0106] Item identification step: receiving item data uploaded by the user, the item data includes image, audio and text formats, generating a feature vector through a feature extraction algorithm, the feature vector includes embedded features of the item image extracted based on a deep learning model; extracting feature parameters of the audio based on spectrum analysis, extracting semantic features of the text using natural language processing technology, matching the extracted user item feature vector with the template feature vector in the database, querying the matching template item in the database according to the matching judgment result, and returning the corresponding item information. If there is no matching item, returning an unmatched result and prompting the user to re-enter or adjust the data. If the user agrees, entering the co-creation process, dynamically updating the template feature vector set in the database based on user feedback or newly added item data during the matching process;

[0107] Digital human image generation steps: used to obtain two-dimensional photo data uploaded by the user through the mobile terminal; compare the rendered image of the initial three-dimensional model with the user's photo through the image projection method; optimize the three-dimensional model parameters based on the mean square error so that the rendered image is close to the user's photo; output the optimized three-dimensional appearance model; define a preset basic action set, which includes facial expressions and body movements; according to the scene requirements selected by the user, assign a time dynamic weight function to each preset action, use the weighted combination method to generate a continuous action sequence, and adjust the order and intensity of the actions to match the scene logic;

[0108] Digital human voice expression generation step: used to extract the audio features of the voice data uploaded through the mobile terminal, generate a voice model using a recurrent neural network, and output a voice signal with user voice features according to the user's scenario interaction requirements; process the text description provided by the user through a natural language generation model to generate personalized expressions;

[0109] Interaction steps: used to obtain the user's voice information through the microphone, obtain the user's action information through the high-definition camera, extract the sound information and represent it as a time series, extract the action information and represent it as an action feature sequence, extract the semantic content from the sound features through a preset semantic analysis model, and recognize the action content from the action features through a preset action recognition model; match the most relevant interaction items in the worship interaction item library according to the results of semantic analysis and action recognition, and map the matched worship interaction items to the display space of the display module.

[0110] In the item recognition step of this embodiment, feature vectors are extracted from the item data uploaded by the user. Matching with the template feature F in the database is expressed as:

[0111] Matching determination

[0112] Among them, d(F,F i ) is the distance metric between feature vectors (representing the distance metric between feature vectors, commonly used measurement methods include Euclidean distance, cosine similarity or Mahalanobis distance, etc.), ∈ is the matching threshold, which is used to determine whether the feature vectors meet the matching conditions;

[0113] When the matching judgment function When , the system considers that the feature vector matches successfully and returns the matching result. , it means the matching fails and the system will enter the co-creation process.

[0114] If the features uploaded by the user are not matched successfully, the co-creation process will be entered, which is represented as follows:

[0115] Feature supplement F′=F+ΔF

[0116] Wherein ΔF is the feature supplemented by the second user, specifically, the additional feature information supplemented or modified by the user, which may be data such as images, voices, and text descriptions uploaded additionally;

[0117] The newly added template F′ is added to the database, represented as:

[0118] in, represents the original feature template database, F ′ is the new eigenvector after supplementation; It is an updated feature template database, which contains newly recorded feature templates.

[0119] In the item recognition step, a feature vector F is extracted from the item data uploaded by the user (such as photos, videos, audio or other descriptive information). The feature vector belongs to a high-dimensional feature space and is denoted as: Among them, F represents the feature vector of the item uploaded by the user, and n is the dimension of the feature vector, which depends on the algorithm used to extract the feature and the data type (such as image features, sound features, text descriptions, etc.). The system matches the extracted feature vector with the feature template set stored in the database.

[0120] Through accurate matching of feature vectors, the possibility of misidentification and missed identification is reduced. Distance measurement functions and matching thresholds are used to ensure the scientific nature and controllability of matching. When a match fails, a co-creation process is introduced, and users can add additional feature information to continuously improve the database. By updating the feature template database in real time, the system's recognition ability and coverage of new items are improved. Users participate in the co-creation process, supplement item features, and enhance user participation and system interactivity. The continuous expansion of the database forms a sustainable item feature library, which improves the long-term applicability and recognition ability of the system. Each feature supplement and database update is equivalent to a self-learning of the system, and the system's recognition ability will continue to increase. Feature vector supplementation reduces the need for manual intervention, and the system can gradually achieve automatic optimization. Each supplemented feature vector carries the source information uploaded by the user to ensure the traceability and manageability of the data. Improve the transparency of the database and the efficiency of data management.

[0121] In the scene generation step of this embodiment, the scene S (x, y, z, t) is represented as a function of the evolution of the three-dimensional space (x, y, z) over time t, expressed as:

[0122]

[0123] Among them, f i (x, y, z) represents the spatial distribution of the i-th element in the scene; g i(t) represents the temporal dynamics of the element;

[0124] The memorial scene S(x,y,z,t) is represented as a multidimensional function space, where the scene is the dynamic evolution of elements in the three-dimensional space x,y,z at time t, where S(x,y,z,t) represents the complete memorial scene and is a joint function of space and time. i (x, y, z) represents the spatial distribution function of the i-th element in the scene, defining the static characteristics of the element in three-dimensional space, such as position, shape, and color. i (t) represents the time dynamic function of the i-th element in the scene, which defines the state of the element changing over time, such as moving, changing color, disappearing, flashing, etc. n represents the total number of elements contained in the scene.

[0125] A scene is composed of the spatial distribution and temporal dynamics of multiple independent elements. The spatial state of each element is i (x,y,z) and time dynamics g i (t) are multiplied together to form a complete representation of the entire scene.

[0126] Through user-selected scene templates The template is represented by the following formula:

[0127] T(x,y,z,t)={S i (x,y,z,t)|i∈I}

[0128] Where I is the element index set contained in the template. T(x, y, z, t) is the specific scene template selected by the user. S i (x, y, z, t) is the i-th element extracted from the complete scene S. I is the element index set contained in the scene template, which is used to identify the specific elements contained in the template. The scene template T selected by the user is a subset of the complete scene S, which is controlled by a set of specific indexes I. Different template selections correspond to different index sets I, thus forming different scene effects.

[0129] After the user selects a template, the system will extract scene elements according to the index set I and use the rendering engine for dynamic rendering.

[0130] The rendering process includes:

[0131] Spatial rendering: visualize the spatial distribution of elements.

[0132] Time-based dynamic rendering: Animate or process elements with special effects based on time-based dynamic functions.

[0133] The scene will change dynamically according to the time, location and theme specified by the user to achieve the desired display effect.

[0134] Through the combination of spatial distribution function and time dynamic function, the scene not only has a static three-dimensional spatial layout, but also has a dynamic effect that changes over time. A variety of memorial scenes such as dynamic family halls, ancestral halls, landscape courtyards, etc. can be generated to meet the needs of different scenes. Users can select specific scene templates from the scene library for personalized customization. The scene elements are disassembled and combined through mathematical modeling, and the rendering engine can efficiently process spatial and temporal information to ensure real-time display and dynamic update of the scene. The rendering process supports efficient output of different devices (such as VR, AR, and mobile devices) to provide a consistent user experience. The scene template can be continuously expanded according to different user needs. New scene elements can be added to the scene database by supplementing feature information to form a sustainable scene library. Support dynamic scenes such as life scenes of old people, re-enactment of historical events, etc., to enhance the emotional resonance of commemoration. The templated scene selection mechanism improves the reusability of data and reduces the workload of repeated modeling. In the scene generation process, this embodiment accurately describes the spatial distribution and temporal dynamic characteristics of scene elements by representing the scene as a combination of functions of three-dimensional space and time. Users can select specific scenes based on scene templates to achieve personalized configuration.

[0135] Generate f through preset geometric models or user-uploaded data i (x, y, z), the generation steps include: obtaining the three-dimensional shape, position and size parameters of each scene element, mapping the parameters to the distribution function f in three-dimensional space i (x, y, z), used to represent the spatial characteristics of an element;

[0136] Define the time dynamic function g i Step (t) includes: specifying dynamic behavior for each scene element, calculating g using interpolation or dynamic weights i The value of (t) is used to describe the state changes of elements at different time points;

[0137] In the digital human image generation step of this embodiment, the appearance of the digital human is represented by a model M(x, y, z), where M is an approximation function based on the user's uploaded photo:

[0138]

[0139] Where P(M) is the image obtained by projecting the 3D model onto a 2D plane; D is the photo data uploaded by the user;

[0140] The appearance of the digital human is represented by a three-dimensional model M(x, y, z), and its generation goal is to make the three-dimensional model as close as possible to the two-dimensional photo data D uploaded by the user. This goal can be achieved by minimizing the error after the three-dimensional model is projected onto a two-dimensional plane. M(x, y, z) represents the appearance model of the digital human in three-dimensional space. P(M) represents the image generated after projecting the three-dimensional model M onto a two-dimensional plane. D is the two-dimensional photo data uploaded by the user.

[0141] The processing flow is as follows: extract facial and body feature points from the photo data D uploaded by the user. Project the initial 3D model M to obtain a 2D image P(M). Iterate and optimize the 3D model parameters by minimizing the error of ∥P(M)-D∥. Finally, output the optimized 3D appearance model M(x, y, z).

[0142] The behavior of the digital human is represented by the action function A(t):

[0143]

[0144] Among them: a j (t) is the time weight function of the action; φ j Is a collection of predefined action bases.

[0145] The dynamic behavior of the digital human is represented by the action function A(t), which converts different action bases φ j Through the time weight function a j (t) to generate a continuous behavior sequence. Specifically, A(t) is the action behavior of the digital human at time t. j (t) is the time weight function of the jth action base, which represents the proportion or intensity of the action in time. j is a set of predefined action bases, including basic actions such as smiling, nodding, waving, and walking. m is the total number of predefined action bases.

[0146] Processing flow: Select basic action base from the predefined action library j . Determine the time weight function a for each action based on user needs or scene logic j (t). Use weighted combination to generate a complete action sequence A(t). Map the generated action sequence to the digital human 3D model M(x,y,z) to achieve dynamic behavior display.

[0147] In the digital speech expression generation step, the speech signal V(t) and the text expression T are generated by the AI ​​model trained with the user uploaded data:

[0148]

[0149] Where: X is the audio dataset uploaded by the user;

[0150] Y is the text or biographical description data provided by the user;

[0151] and is the generating function trained by deep learning.

[0152] The digital human’s voice signal is generated by the audio data set X uploaded by the user through the speech generation function trained by AI Generate, specifically, V(t) is the speech signal at time t. X is the audio data set uploaded by the user, including information such as intonation, rhythm, and pronunciation features. It is a speech generation function trained by a deep learning model (such as RNN or Transformer).

[0153] Processing flow: Extract characteristic parameters (such as intonation, pronunciation habits, rhythm) from the audio data set uploaded by the user. Train the speech generation model Generate a time series speech signal V(t) based on user input or scenario requirements and output speech data that can be played.

[0154] The textual expression of the digital person is generated by the text or biographical description data Y provided by the user through the AI ​​text generation function Generate, specifically, T is the textual expression content generated by the digital human. T is the text or biographical description data uploaded by the user. It is a text generation function, and deep learning models (such as GPT, Transformer, etc.) are often used for training.

[0155] Processing flow: Extract key information from the text data Y uploaded by the user. Use text generation model Analyze language logic and expression style. Generate text expression T that meets user description and scenario requirements. The generated text content is passed to the voice signal generation module and converted into playable voice.

[0156] The present invention minimizes the projection error through a mathematical model, achieves a high degree of matching between the three-dimensional model and the photos uploaded by the user, and restores the facial features and body contours of the deceased. The optimized three-dimensional model is more natural and realistic visually, which enhances the user's immersion and emotional resonance. Using predefined action bases and time weight functions, the digital human can perform natural and smooth dynamic behaviors, such as smiling, nodding, waving, etc. The action sequence is closely integrated with the scene logic to provide a more realistic interactive experience. The unique intonation, rhythm and pronunciation characteristics of the deceased are restored using the AI ​​voice model. The text description data provided by the user is converted into vivid and natural language expression to enhance the authenticity of emotional expression. The AI ​​model is used to perform deep learning on audio, text, and image data, so that the system can adapt to the data characteristics of different users. As more data is uploaded by users, the restoration effect of the digital human image, behavior and voice will continue to be optimized. Multimodal interaction of vision (digital human image), hearing (voice signal), and action (dynamic behavior) is achieved to enhance the interactivity and user experience of the memorial screen. This embodiment highly integrates the appearance, dynamic behavior and voice expression of the digital human through mathematical modeling and AI deep learning technology.

[0157] In the interaction step of this embodiment,

[0158] Target user voice information V u (t) is expressed as a time series: V u (t)={v1,v2,…,v k}; where v i is the time t i The sound characteristics of

[0159] The target user action information Au(t) is represented as an action feature sequence: A u (t)={a1,a2,…,a m}; where a j is the time t j The action characteristics on

[0160] Define the semantic analysis model as L v , from the sound feature V u (t) Extracting semantic content Expressed as:

[0161]

[0162] Define the action recognition model as L a , from the action feature A u (t) Identify the action content Expressed as:

[0163]

[0164] Define the worship interaction project library as According to semantic analysis and action recognition The result matches the most relevant interaction item P * Expressed as:

[0165]

[0166] in It is the relevance scoring function between interaction items and semantic and action matching;

[0167] Map the matching worship interaction item P to the display space ε of the electronic worship device, expressed as:

[0168]

[0169] in: The interactive item P * A function embedded in the interaction field G′(x,y,z,t) and displayed on the display module; (x′,y′,z′) are the display coordinates on the electronic device.

[0170] In this embodiment, the user's voice information and action information are intelligently analyzed and matched with corresponding worship interaction items, and finally the scene is mapped and rendered on the display screen to achieve natural interaction between the user and the memorial screen.

[0171] Target user's voice information V u (t) is represented as a time series, which contains the sound features V at different time points u (t) is the sound feature sequence of the target user at time t. Specifically, v i is time point t i The sound feature vector on represents the frequency, pitch, tone, speaking speed and other features of the speech. k is the total number of sound feature points.

[0172] Processing flow: Collect the user's voice data. Decompose the sound signal into multiple feature points in the time series. Each feature point contains core features such as the frequency, pitch, and volume of the voice.

[0173] The target user’s action information Au(t) is represented as an action feature sequence, describing the user’s body movements or gestures at different time points. Specifically, Au(t) is the action feature sequence of the target user at time t. j is time point t j The action feature vector on contains features such as gesture, posture, body angle, etc. m is the total number of action feature points.

[0174] Processing flow: Capture the user's motion data through a high-definition camera. Extract multiple motion feature points in the time series. Each feature point contains the position information, angle change, and strength characteristics of the body movement.

[0175] Define the semantic analysis model L v , used to extract semantic content from sound feature sequences, specifically, is the semantic content extracted from the sound features. u (t) Analyze sound sequences based on AI algorithms (such as natural language processing models) to extract key semantic information from speech.

[0176] Define the action recognition model L a , used to extract the action feature sequence A u (t) identifies the action content, specifically, is the action content extracted by the action feature. To analyze the action sequence based on computer vision algorithms (such as convolutional neural network CNN or skeleton tracking algorithm) and extract the user's key action intentions.

[0177] Processing flow: The voice features are used to extract the user's semantic intention through the semantic analysis model. The action features are used to extract the user's body language meaning through the action recognition model.

[0178] Define the worship interaction project library Each interaction item P i Represents a specific worship interaction scenario. According to the semantic analysis results L v And the action recognition result L a , matching the most relevant worship interaction item P * , specifically, P * The interactive item that best matches the user's voice and actions. is the relevance scoring function, measuring the worship project P i And the semantic analysis result L v And the action recognition result L a The matching degree P is the worship interaction project library.

[0179] Processing flow: Take the semantic analysis results and action recognition results as input, calculate the relevance score of each item, and select the item with the highest relevance as the final interactive content.

[0180] Map the matching worship interaction items to the display space of the electronic worship device. Specifically,

[0181] ε(x′, y′, z′, t) is the rendering space on the display screen. To make the interactive project P *A function that is embedded into the scene G′(x,y,z,t) and rendered. (x′,y′,z′) are the coordinates on the display screen. t is the time dynamics.

[0182] Processing flow: Embed the matching interactive items into the scene. Render the worship interactive items on the display screen. Users can intuitively see the interactive effects on the display screen.

[0183] Through voice analysis and motion recognition, the system can accurately understand the user's intentions and achieve natural human-computer interaction. The correlation scoring function is used to match interactive items to ensure that user needs are highly consistent with the scene display. The interactive content can be dynamically mapped to the display screen to present an immersive interactive effect. The three major interactive methods of sound, action, and vision are integrated to provide a highly realistic interactive experience. The worship interaction project library can be expanded according to needs to meet the needs of diverse scenarios. Through the combination of mathematical models and AI algorithms, the real-time response speed and interaction fluency of the system are improved. Through time series modeling, semantic analysis, motion recognition and scene mapping, a natural, accurate and highly immersive interactive experience is achieved between users and worship equipment. This solution not only improves the intelligence level of the memorial screen, but also provides sustainable technical support for future AI and metaverse scene interactions.

[0184] The present invention generates dynamic scenes through three-dimensional modeling technology and multi-dimensional function space, so that the memorial scene not only includes traditional family halls, ancestral halls and landscape courtyards, but also can be dynamically adjusted according to seasons and regional information, enhancing the user's sense of immersion and ritual. Supporting rendering and displaying scenes in the metaverse environment, users can experience realistic virtual memorial spaces from a variety of devices (such as VR / AR head displays, mobile terminals). Highly similar three-dimensional digital human models are optimized and generated through two-dimensional photos uploaded by users, so that the image of the deceased is truly reproduced. AI technology is used to generate personalized voice expressions and natural language descriptions to achieve the reproduction of the deceased's tone, language habits and emotional expressions during his lifetime, providing emotional resonance for users. Through semantic analysis models and action recognition models, the interactive intention can be accurately extracted from the user's voice and actions, and the most relevant worship items can be matched. The dynamically generated digital human behavior and voice response are closely integrated with the scene logic, providing a natural and smooth interactive experience and enhancing the user's sense of participation. Provide digital and dynamic means of clan culture inheritance to solve the limitations of traditional sacrificial forms that are subject to factors such as region and time. It supports the reproduction of deceased people's life scenes and the reproduction of family historical events in dynamic scenes, giving traditional clan culture more scientific and technological connotations. By combining user interaction data with the dynamic behavior of digital humans, memorial activities are made more interactive and educational, stimulating the younger generation's interest in family culture and the responsibility of inheritance. Through this invention, traditional memorial forms have been fully digitalized, intelligentized, and immersive, providing innovative and practical technical means for the protection and inheritance of clan culture.

[0185] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. An AI cloud intelligence blessing memorial screen for clan culture inheritance, comprising a wall-mounted display terminal (1) and a mobile terminal (2) connected to the display terminal (1) via the Internet, characterized in that: The mobile terminal (2) is installed with an APP connected to the display terminal (1); the front surface of the display terminal (1) is inlaid with control buttons (11), a display screen (12) and a high-definition camera (13); the mobile terminal scans a QR code on the display screen through the APP or WeChat public account to send a photo upload request instruction; the display screen (12) is divided into a plurality of display areas, the display areas include an automatic scrolling playback area and a fixed display area, the automatic scrolling playback area is used to automatically play photos and videos uploaded by users, and the fixed display area is used to display user-defined content for a long time; the size of the display screen is determined based on the best visual distance of two meters, and the display mode of the display screen (12) includes black and white display and color display; The display terminal is provided with: Network connection module: used for data transmission and communication with external devices, the connection methods include optical fiber line access, WiFi connection and NFC connection; Timer power on / off module: realizes the timer power on and off operation of the display screen, and supports automatic power on and off of the device in the specified time period; Cloud storage module: used to store and manage photos, videos and related data uploaded by users, and supports fast data migration and import between new and old devices; User rights management module: used to control the user's rights to upload data and determine whether the user has paid the fee. The user obtains the upload rights after paying the management fee; Remote management module: used to receive remote operation instructions from the administrator, including editing and maintaining the photo and video content stored in the device; Management module: AI robots are used to automatically perform intelligent management in the background, including data analysis management, content optimization management, and user interaction management; AI compatible module: used to realize the display terminal and AI application and metaverse scene functions, including data interaction unit, scene rendering unit, digital human generation unit, project interaction unit and metaverse space mapping unit, among which: The data interaction unit is used to receive and upload multimodal data; The scene dynamic rendering unit is used to load and adjust the memorial scene in real time; The digital human generation unit is used to generate a digital human and optimize the emotion and behavior performance of the three-dimensional digital human model; The item interaction unit is used for semantic recognition and intent analysis to recognize the user's voice and actions, and match the worship interaction items according to the recognition results; The Metaverse space mapping unit is used to synchronize device data to the Metaverse platform; Interface module: includes power interface and audio output interface. The power interface is used to connect an external power source to the display terminal. Users can connect an external speaker device through the audio output interface according to their needs.

2. The AI ​​cloud wisdom blessing memorial screen for clan culture inheritance according to claim 1 is characterized in that: The AI-compatible module also includes a project interaction unit, where: A scene dynamic rendering unit is used to build a memorial scene through a multidimensional function space, and a scene template selected by a user through a public account, and determine a set of elements contained in a target scene template based on the scene template selected by the user; according to the scene template, a dynamic scene containing target elements is generated, and the memorial scene is rendered in a metaverse environment; the memorial scene includes a family hall, a clan temple, and a landscape courtyard, and a preset scene is selected according to season and regional information, and the selected scene is displayed on a display screen (12); A digital human generation unit is used to obtain two-dimensional photo data uploaded by a user through a mobile terminal (2); compare the rendered image of the initial three-dimensional model with the user's photo through an image projection method; optimize the three-dimensional model parameters based on the mean square error so that the rendered image is close to the user's photo; output the optimized three-dimensional appearance model; define a preset basic action set, which includes facial expressions and body movements; assign a time dynamic weight function to each preset action according to the scene requirements selected by the user, generate a continuous action sequence using a weighted combination method, and adjust the order and intensity of the actions to match the scene logic; extract the audio features of the voice data uploaded through the mobile terminal (2), generate a voice model using a recurrent neural network, and output a voice signal with user voice features according to the user's scene interaction requirements; process the text description provided by the user through a natural language generation model to generate a personalized expression; The project interaction unit is used to obtain the user's voice information through a microphone, obtain the user's action information through a high-definition camera (13), extract the voice information and represent it as a time series, extract the action information and represent it as an action feature sequence, extract the semantic content from the voice features through a preset semantic analysis model, and recognize the action content from the action features through a preset action recognition model; match the most relevant interaction project in the worship interaction project library according to the results of semantic analysis and action recognition, and map the matched worship interaction project to the display space of the display module.

3. The AI ​​cloud wisdom blessing memorial screen for clan culture inheritance according to claim 1 is characterized in that: The display terminal (1) supports plane display, immersive display and enhanced display. The plane display is to display the interactive scene on the display screen (12); the immersive display is to display the interactive scene in a three-dimensional virtual space through VR / AR equipment; the enhanced display is to display the digital human and scene elements superimposed on the real environment through the mobile terminal AR function; the display terminal (1) is used to receive scene data and adapt it to a format that can be displayed by the terminal device, and at the same time, according to the output of the voice and behavior interaction module, update the digital human's actions, expressions and interactive effects in the scene in real time.

4. An AI cloud wisdom blessing commemorative method for clan culture inheritance, based on the AI ​​cloud wisdom blessing commemorative screen for clan culture inheritance described in any one of claims 1-3, characterized in that: The following steps are involved: Scene building step: used to build a memorial scene through a multidimensional function space, the scene template selected by the user through the official account, based on the scene template selected by the user, determine the element set contained in the target scene template; according to the scene template, generate a dynamic scene containing the target element, and render the memorial scene in the metaverse environment; the memorial scene includes a family hall, a clan temple and a landscape courtyard, and selects a preset scene according to season and regional information, and displays the selected scene through a display screen (12); Item identification step: receiving item data uploaded by a user via a mobile terminal (2), wherein the item data includes image, audio and text formats, and generating a feature vector using a feature extraction algorithm, wherein the feature vector includes embedded features extracted from the item image based on a deep learning model; Extract feature parameters of audio based on spectrum analysis, extract semantic features of text using natural language processing technology, match the extracted user item feature vector with the template feature vector in the database, query the matching template item in the database based on the matching judgment result, and return the corresponding item information. If there is no matching item, return an unmatched result and prompt the user to re-enter or adjust the data. If the user agrees, enter the co-creation process, and dynamically update the template feature vector set in the database based on user feedback or newly added item data during the matching process; The digital human image generation step is used to obtain two-dimensional photo data uploaded by the user through the mobile terminal (2); compare the rendered image of the initial three-dimensional model with the user's photo through an image projection method; optimize the three-dimensional model parameters based on the mean square error so that the rendered image is close to the user's photo; Output optimized 3D appearance model; define preset basic action set, including facial expressions and body movements; assign time dynamic weight function to each preset action according to the scene requirements selected by the user, generate continuous action sequence by weighted combination method, and adjust the order and intensity of the action to match the scene logic; The step of generating digital human speech expression is used to extract the audio features of the speech data uploaded by the mobile terminal (2), generate a speech model using a recurrent neural network, and output a speech signal with user speech features according to the user scene interaction requirements; The text description provided by the user is processed through a natural language generation model to generate personalized expressions; The worship project interaction step is used to obtain the user's voice information through a microphone (15), obtain the user's action information through a high-definition camera (13), extract the voice information and represent it as a time series, extract the action information and represent it as an action feature sequence, extract semantic content from the sound features through a preset semantic analysis model, and identify the action content from the action features through a preset action recognition model; match the most relevant interaction project in the worship interaction project library according to the results of semantic analysis and action recognition, and map the matched worship interaction project to the display space of the display module.

5. The Alyunzhinafu memorial method for clan culture inheritance according to claim 4 is characterized in that: In the item recognition step, feature vectors are extracted from the item data uploaded by the user. Matching with the template feature F in the database is expressed as: Among them, d(F, Fi) is the distance measure between feature vectors, ∈ is the matching threshold; If the features uploaded by the user are not matched successfully, the co-creation process will be entered, which is represented as follows: Feature supplement F′=F+ΔF Where ΔF is the feature added by the second user; The newly added template F′ is added to the database, represented as:

6. The Alyunzhinafu memorial method for clan culture inheritance according to claim 4 is characterized in that: In the scene generation step, the scene S(x, y, z, t) is represented as a function of the evolution of the three-dimensional space (x, y, z) over time t, expressed as: Among them, f i (x, y, z) represents the spatial distribution of the i-th element in the scene; g i (t) represents the temporal dynamics of the element; Through user-selected scene templates The template is represented by the following formula: T(x,y,z,t)={S i (x,y,z,t)| i∈I} Where I is the element index set contained in the template.

7. The Alyunzhinafu memorial method for clan culture inheritance according to claim 6 is characterized in that: Generate f through preset geometric models or user-uploaded data i (x, y, z), the generation steps include: obtaining the three-dimensional shape, position and size parameters of each scene element, mapping the parameters to the distribution function f in three-dimensional space i (x, y, z), used to represent the spatial characteristics of an element; Define the time dynamic function g i Step (t) includes: specifying dynamic behavior for each scene element, calculating g using interpolation or dynamic weights i The value of (t) is used to describe the state changes of elements at different time points.

8. The AI ​​cloud wisdom blessing memorial method for clan culture inheritance according to claim 4 is characterized in that: In the digital human image generation step, the appearance of the digital human is represented by a model M(x, y, z), where M is an approximation function based on the user's uploaded photos: Where P(M) is the image obtained by projecting the 3D model onto a 2D plane; D is the photo data uploaded by the user; The behavior of the digital human is represented by the action function A(t): Among them: a j (t) is the time weight function of the action; φ j Is a collection of predefined action bases.

9. The Alyunzhinafu memorial method for clan culture inheritance according to claim 4 is characterized in that: In the digital speech expression generation step, the speech signal V(t) and the text expression T are generated by the AI ​​model trained with the user uploaded data: Where: X is the audio dataset uploaded by the user; Y is the text or biographical description data provided by the user; and is the generating function trained by deep learning.

10. The Alyunzhinafu memorial method for clan culture inheritance according to claim 1 is characterized in that: In the interactive step, Target user voice information V u (t) is expressed as a time series: V u (t) = {v1, v2, ..., v k }; where v i is the time t i The sound characteristics of Target User action information Au(t) is represented as an action feature sequence: A u (t) = {a1, a2, ..., a m }; where a j is the time t j The action characteristics on definition The semantic analysis model is L v , from the sound feature V u (t) Extracting semantic content It is expressed as: Define the action recognition model as L a , from the action feature A u (t) Identify the action content It is expressed as: Define the worship interaction project library as According to semantic analysis and action recognition The result matches the most relevant interaction item P * It is expressed as: in It is the relevance scoring function between interaction items and semantic and action matching; The matching worship interaction item P * Mapped to the display space ε of the electronic sacrificial device, it is expressed as: in: The interactive item P * A function embedded in the interaction field G′(x, y, z, t) and displayed on the display module; (x′, y′, z′) are display coordinates on the electronic device.

Images