An AI interactive AR augmented reality travel culture wall system

Abstract

This invention discloses an AI-interactive AR (Augmented Reality) cultural wall system for cultural tourism, belonging to the field of AR system technology. The surface of the cultural wall is equipped with a pre-embedded feature set, including visual markers, differentiated texture features, 3D depth variation areas, and specifically designed interactive hotspots. These features provide the system with stable visual benchmarks, semantic distinction criteria, depth cues, and precise positioning anchors. The system includes: a feature point acquisition module, used to collect the feature set on the cultural wall through a multimodal feature extraction algorithm to construct a 3D spatial map with semantic annotations; a user positioning and pose estimation module; an AI content generation engine; and a spatial adaptation and rendering module. Through the cultural wall feature set and the multi-module collaborative system, precise spatial integration of virtual content and the physical wall surface is achieved, along with intelligent dynamic content generation based on user behavior, effectively enhancing the immersion, interactivity, and personalization of the AR cultural experience.

Description

Technical Field

[0001] This invention relates to the field of AR system technology, and in particular to an AI interactive AR augmented reality cultural tourism wall system. Background Technology

[0002] Cultural walls, as important carriers of cultural dissemination in cultural tourism settings, are commonly found in museums, historical and cultural blocks, tourist attractions, and other places. Traditional cultural walls mainly adopt static display forms such as graphic panels and relief murals. Although they can display certain cultural content, they have obvious shortcomings such as a single display method, limited information capacity, and a lack of interactive experience. With the development of augmented reality (AR) technology, existing technologies have emerged that allow cultural wall display solutions to trigger virtual content by scanning specific markers with a mobile phone.

[0003] In terms of implementation technology, existing solutions typically face the following technical challenges: First, the spatial registration accuracy is insufficient, and virtual objects are prone to positional drift or inaccurate alignment with physical walls; second, content generation lacks intelligence and cannot dynamically generate virtual content that fits the cultural theme based on real-time scenes; third, the coordination between system modules is poor, with feature acquisition, user positioning, content generation, and rendering display being independent of each other, making it difficult to form an efficient closed-loop system; and fourth, there is a lack of deep perception and understanding of user behavior, making it impossible to provide an adaptive interactive experience. Summary of the Invention

[0004] To address the aforementioned problems, this invention aims to solve the issues described above. One objective of this invention is to provide an AI-interactive AR (Augmented Reality) cultural wall system that solves the problems described above. Through a set of cultural wall features and a multi-module collaborative system, it achieves precise spatial integration of virtual content and physical wall surfaces, as well as intelligent dynamic content generation based on user behavior, effectively enhancing the immersiveness, interactivity, and personalization of the AR cultural experience.

[0005] The solution adopted in this invention is: an AI interactive AR augmented reality cultural tourism wall system, in which the surface of the cultural wall is provided with a pre-embedded feature set, including visual marker points, differentiated texture features, three-dimensional depth variation areas and specially designed interactive hot zones, which provide the system with stable visual reference, semantic distinction basis, depth clues and precise positioning anchor points respectively; The system includes: a feature point acquisition module, used to acquire the feature set of the cultural wall through a multimodal feature extraction algorithm, and construct a 3D spatial map containing semantic annotations; The user positioning and attitude estimation module receives a 3D spatial map from the feature point acquisition module. By integrating computer vision positioning, inertial sensor data and wireless positioning technology, it tracks the user's spatial position, body orientation, joint posture and line of sight relative to the cultural wall in real time. The AI ​​content generation engine dynamically generates virtual content that matches the theme of the cultural wall, based on the semantic features of the cultural wall from the feature point acquisition module and the user spatial relationship data from the user positioning and posture estimation module. The spatial adaptation and rendering module simultaneously receives a 3D spatial map from the feature point acquisition module and virtual content from the AI ​​content generation engine. It precisely registers the virtual content into the 3D coordinate system of the cultural wall, performs physically realistic rendering and compositing, and finally presents the composite effect through an AR application on the user's mobile device. It also integrates media recording functionality to capture and save the composite image containing the user's image and the virtual content. This constructs a complete technological closed loop from environmental perception to content generation to immersive presentation, achieving precise integration of virtual content and the cultural wall.

[0006] The preferred technical solution involves the feature point acquisition module analyzing visual markers and differentiated texture features on the surface of the cultural wall using multi-scale feature extraction technology. Combined with semantic segmentation algorithms, it identifies the outlines and content of each cultural element. 3D scanning technology is used to acquire geometric information of areas with varying depth, and interactive anchor points are set in specially designed interactive hotspots. Ultimately, a database of cultural elements containing semantic tags, spatial coordinates, and interactive attributes is established, providing a data foundation for the accurate mapping of virtual content. This achieves a precise conversion of the physical features of the cultural wall into digital semantics, providing a structured foundation for environmental understanding within the system.

[0007] The preferred technical solution involves a user positioning and posture estimation module employing multi-sensor fusion technology. This involves establishing a positioning benchmark by identifying visual markers on the cultural wall, confirming semantic regions using differentiated texture features, enhancing positioning accuracy by combining 3D depth variation regions, calculating interaction possibilities around interactive hotspots, capturing user movement trajectories with a camera, and obtaining body orientation using an inertial measurement unit. This enables precise tracking of the user's position, posture, and viewing intention. A precise spatial relationship model between the user and the cultural wall is established, providing accurate input for content generation and interactive responses.

[0008] The preferred technical solution lies in that the AI ​​content generation engine is based on a deep learning model and dynamically generates multiple types of virtual content according to user behavior: it generates navigation content when the user approaches a visual marker area, performs semantic interpretation when encountering differentiated texture features, generates 3D displays in areas of varying depth, and creates interactive experiences in interactive hotspots. Intelligent content creation is achieved through natural language generation technology, neural radiation field technology, and generative adversarial networks. This enables context-aware intelligent content creation, ensuring a high degree of relevance between the virtual content and the cultural context.

[0009] The preferred technical solution involves a spatial adaptation and rendering module that utilizes real-time spatial computing technology to calibrate the display baseline based on visual marker points, optimizes visual fusion by combining differentiated texture features, achieves depth fitting using areas of varying three-dimensional depth, enhances interactive feedback for interactive hotspots, establishes a spatial relationship model using visual SLAM technology, and ensures visual harmony through occlusion processing algorithms, thereby achieving a natural integration of virtual content and the cultural wall. This ensures accurate registration and visual harmony of virtual content in physical space, enhancing the immersive experience of augmented reality.

[0010] The preferred technical solution lies in the fact that the system constructs an intelligent response system based on the feature set of the cultural wall. By analyzing the user's dwell time in visual marker areas, movement speed in front of differentiated texture features, observation angle in areas with varying depth, and interaction frequency in interactive hotspots, the system dynamically adjusts the content display strategy, forming a progressive knowledge transfer mechanism. This achieves intelligent navigation based on user behavior analysis, improving the accuracy and effectiveness of cultural dissemination.

[0011] A preferred technical solution involves integrating eye-tracking technology into the system. By detecting the user's gaze duration at visual markers, the area of ​​attention for differentiated texture features, the focus of observation on areas of depth variation, and the degree of attention paid to interactive hotspots, the system intelligently adjusts the timing and level of detail of virtual content display, achieving adaptive content presentation based on visual attention. This eye-tracking-based intelligent adjustment of content presentation enhances the naturalness and smoothness of the user experience.

[0012] The preferred technical solution is that the system employs multimodal fusion recognition technology. By coordinating the interactive behaviors of multiple users in different areas of the cultural wall, when collaborative localization in visual marker areas, collective observation in front of differentiated texture features, synchronous exploration in areas with varying depth, or collaborative interaction in interactive hotspots is detected, distributed rendering technology is used to generate associated virtual content, ensuring the consistency of the multi-user interactive experience. This supports multi-user collaborative interactive experiences, enhancing the social and engaging aspects of cultural dissemination.

[0013] The preferred technical solution involves establishing a personalized visitor profile system. This system records the user's trajectory in visually marked areas, their choices in front of differentiated texture features, their exploration depth in areas with varying 3D depth, and their interaction preferences in interactive hotspots. Machine learning models are then used to construct user interest profiles, optimizing subsequent content recommendation strategies. This enables personalized services based on user profiles, enhancing the relevance and satisfaction of the cultural experience.

[0014] The preferred technical solution is that the system constructs a multi-mode tour guide system. By recognizing the user's basic needs in visual marker areas, willingness to explore in depth in front of differentiated texture features, spatial perception ability in areas with three-dimensional depth changes, and level of participation in interactive hot zones, the system automatically switches tour guide modes. It uses reinforcement learning, path planning, and task generation algorithms to achieve a personalized cultural exploration experience, providing adaptive tour guide mode selection to meet the diverse cultural experience needs of different users.

[0015] Compared with existing technologies, the AI ​​interactive AR augmented reality cultural tourism wall system of this invention has the following technical effects: 1. The AI ​​interactive AR augmented reality cultural wall system proposed in this application achieves high-precision spatial integration of virtual content and cultural wall. By constructing a cultural element database containing semantic tags and spatial coordinates, and using multi-scale feature extraction technology and 3D scanning technology to establish a precise spatial benchmark, it effectively solves the technical problems of inaccurate alignment between virtual content and physical wall surface and easy position drift in traditional AR systems, and realizes accurate registration and stable display of virtual content in three-dimensional space.

[0016] 2. It provides intelligent dynamic content generation capabilities. Based on the user's behavioral dynamics in different characteristic areas of the cultural wall, it intelligently generates matching virtual content through a deep learning model, achieving a technological leap from fixed content display to intelligent content creation. This system can generate diverse virtual content in real time, including navigation content, semantic interpretation, 3D displays, and interactive experiences, based on the user's location and behavioral characteristics.

[0017] 3. An adaptive human-computer interaction experience has been established. By analyzing the user's dwell time, movement speed, observation angle and interaction frequency in the feature area, and combining eye-tracking technology and machine learning models, the system can intelligently adjust the timing, level of detail and presentation of virtual content, and realize personalized content recommendation based on user behavior and interest preferences.

[0018] 4. Supports rich multi-user collaborative interaction. By adopting multimodal fusion recognition technology and distributed rendering technology, the system can coordinate the interactive behavior of multiple users in different areas of the cultural wall, realize multi-user collaborative experience such as collaborative positioning, collective observation, synchronous exploration and collaborative interaction, and significantly enhance the participation and fun of cultural dissemination.

[0019] 5. A complete closed-loop technology system has been formed. Through a tightly coordinated working mechanism from feature point acquisition, user positioning, content generation to rendering, a complete technology chain from environmental perception to content presentation has been constructed. Data sharing and status synchronization between modules have enabled continuous optimization of system performance and continuous improvement of user experience.

[0020] 6. It enhances the efficiency and quality of cultural dissemination. Through the organic combination of precise user spatial positioning and intelligent content generation, the system can display appropriate cultural content to users at the right time and place, effectively improving the efficiency of cultural information transmission and reception, and providing new technical means and experience methods for cultural dissemination in cultural tourism scenarios. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be arbitrarily combined with each other.

[0022] This embodiment describes in detail an AI interactive AR augmented reality cultural tourism wall system. The surface of the cultural wall is provided with a set of pre-embedded features, including visual markers, differentiated texture features, three-dimensional depth variation areas, and specially designed interactive hot zones, which provide the system with stable visual references, semantic distinction criteria, depth clues, and precise positioning anchors, respectively. In practical implementation, visual markers are printed using ArUco codes with specific color encoding. Four to six markers are placed per square meter of wall surface. Users scan these markers with their mobile phone cameras, and the system achieves sub-centimeter-level positioning accuracy based on computer vision detection algorithms. Differentiated texture features are achieved by using different material patterns in different cultural areas. For example, traditional pattern areas use repetitive geometric patterns, while historical figure areas use portrait textures. After users capture images with their mobile phones, the system extracts features using a cloud-based local binary pattern algorithm and histogram of oriented gradients. Three-dimensional depth variation areas are achieved through relief carving, with relief heights ranging from 2 to 15 centimeters. Users capture image sequences from multiple angles with their mobile phones, and the system calculates depth information from the image sequences using a structure-of-motion (SOR) algorithm. Interactive hotspots are marked with high-contrast color blocks, designed to range in size from 30×30 centimeters to 100×100 centimeters. After users identify the boundaries of these hotspots with their mobile phones, the system establishes virtual interactive anchor points in the cloud. This implementation method, which relies entirely on mobile phone data acquisition and cloud computing, provides a stable and reliable foundation for environmental perception through the synergistic effect of multi-dimensional feature sets.

[0023] The system includes a feature point acquisition module, used to collect feature sets from the cultural wall using a multimodal feature extraction algorithm, and construct a 3D spatial map with semantic annotations. In practice, users scan the cultural wall with their mobile phone cameras. The system performs preliminary processing using a lightweight feature extraction algorithm on the mobile phone, and then uploads the image data to the cloud for in-depth analysis. The cloud server employs a multimodal feature extraction algorithm, integrating traditional computer vision features and deep learning features. The SIFT algorithm is used to extract stable feature points, the ORB algorithm for fast feature matching, and a ResNet-50-based convolutional neural network to extract semantic features. Through the collaborative work of these technologies, the 3D spatial map constructed in the cloud achieves centimeter-level accuracy, accurately recording the spatial coordinates and semantic attributes of each cultural element. This implementation method, based on mobile phone acquisition and cloud computing, achieves a precise conversion from a physical cultural wall to a digital model, providing an accurate environmental awareness foundation for subsequent modules.

[0024] The user localization and pose estimation module receives a 3D spatial map from the feature point acquisition module. By fusing computer vision positioning and mobile phone inertial sensor data, it tracks the user's spatial position, body orientation, joint posture, and gaze direction relative to the cultural wall in real time. Specifically, it employs a purely mobile-based vision-inertial fusion positioning scheme: visual positioning is based on markers on the cultural wall, using a mobile-optimized PnP algorithm to calculate 6-DOF pose; the inertial measurement unit uses the phone's built-in IMU sensor, fusing accelerometer, gyroscope, and magnetometer data through complementary filtering; positioning data is fused using a mobile-based Kalman filter, with an output frequency of 60Hz and a positioning latency of less than 20 milliseconds; simultaneously, a MediaPipe-based skeletal keypoint detection algorithm is used to detect 33 human joints in real time via the phone's front-facing camera, achieving complete user state tracking. This purely mobile-based implementation ensures the accuracy and real-time nature of user positioning, providing reliable input data for personalized content generation.

[0025] The AI ​​content generation engine dynamically generates virtual content that matches the theme of the cultural wall, based on the semantic features of the cultural wall from the feature point acquisition module and the user spatial relationship data from the user positioning and pose estimation module. The specific implementation adopts a cloud-based layered generation architecture: the base layer uses a GPT-4-based text generation model to generate corresponding explanatory text based on the user's location; for example, a brief introduction is generated when the user is 2 meters away from the cultural wall, and a detailed explanation is generated when the user is 0.5 meters away. The visual content generation layer uses a Stable Diffusion model to generate corresponding historical scene images based on the texture features of the cultural wall, achieving a resolution of 1024×1024 pixels. 3D content generation uses Instant-NGP technology to quickly generate 3D models based on the depth information of the cultural wall, with a single model generation time of less than 2 seconds. Interactive content generation uses a cloud-based prefabricated system to dynamically load corresponding interactive modules based on the attributes of interactive hotspots. This cloud-based generation and mobile display implementation achieves intelligent and contextualized content generation, significantly improving the relevance and richness of the user experience.

[0026] The spatial adaptation and rendering module simultaneously receives a 3D spatial map from the feature point acquisition module and virtual content from the AI ​​content generation engine. It accurately registers the virtual content into the 3D coordinate system of the cultural wall, performs physically realistic rendering and compositing, and finally presents the composite effect through an AR application on the user's mobile device. It also integrates media recording functionality to capture and save the composite image containing the user's image and the virtual content. Specifically, it uses the mobile ARKit / ARCore underlying framework combined with cloud-based spatial computing algorithms: visual markers provide an initial registration benchmark, and sub-pixel-level alignment accuracy is achieved through mobile feature point matching; differentiated texture features are used to optimize texture mapping, employing physically based rendering technology and using the GGX material model to simulate the reflection characteristics of different surfaces; depth-varying areas achieve depth fitting of virtual content through depth testing, and layered depth buffering technology handles complex occlusion relationships; interactive hotspots use collision detection algorithms, employing a bounding box hierarchy for fast collision lookup; and the media recording function is implemented through mobile frame buffering capture technology, supporting 1080P resolution video recording. This mobile rendering implementation ensures seamless integration of virtual content with the real environment, providing an immersive AR experience.

[0027] A complete technological loop has been constructed, encompassing environmental perception, content generation, and immersive presentation, achieving precise integration of virtual content and the cultural wall. Through the collaborative work of these modules, the system forms a complete data processing pipeline: user mobile phones collect image data, which is uploaded to the cloud-based feature point acquisition module for processing; the user positioning module calculates the user's status in real time based on mobile phone sensor data; the AI ​​content generation engine generates virtual content in the cloud based on environmental features and the user's status; the spatial adaptation and rendering module completes the final registration and rendering on the mobile device; and the end-to-end latency of the entire system is controlled within 100 milliseconds. This cloud-based collaborative implementation ensures efficient collaboration among the system's modules, providing users with a smooth and natural interactive experience.

[0028] A preferred implementation of the feature point acquisition module is that the feature point acquisition module analyzes the visual marker points and differentiated texture features set on the surface of the cultural wall through multi-scale feature extraction technology, identifies the outline and content of each cultural element by combining semantic segmentation algorithm, obtains the geometric information of the three-dimensional depth change area by using motion reconstruction structure technology, and sets interactive anchor points in the interactive hot zone of a specific design. Finally, a database of cultural elements containing semantic tags, spatial coordinates and interactive attributes is established to provide a data foundation for the accurate mapping of virtual content. In specific implementation, multi-scale feature extraction adopts a cloud-based pyramid-style feature extraction architecture, extracting feature points across multiple scales based on images captured by mobile phones to ensure stable feature matching at different viewing distances. Visual marker point analysis employs a cloud-based ArUco code recognition algorithm, combined with adaptive thresholding to ensure recognition stability under different lighting conditions. Differential texture feature analysis uses a combination of cloud-based local binary mode and histogram of oriented gradients to effectively distinguish different texture patterns. The semantic segmentation algorithm uses a cloud-based DeepLabv3+ architecture, pre-trained on the COCO dataset, to accurately identify various cultural elements on the cultural wall. 3D reconstruction uses structure-of-motion reconstruction technology, calculating depth information through multi-view image sequences from mobile phones. Interactive anchor point setting uses a cloud-based feature point clustering algorithm, selecting the most representative feature points within the interactive hotspot as interactive anchor points. The cultural element database uses a cloud-based spatial database management system, supporting rapid spatial queries and semantic retrieval. This implementation method, based on mobile phone acquisition and cloud computing, achieves accurate conversion of the physical features of the cultural wall into digital semantics through multi-level feature analysis, providing a structured environmental understanding foundation for the system.

[0029] A preferred embodiment of the user positioning module is that the user positioning and attitude estimation module adopts mobile phone multi-sensor fusion technology, establishes a positioning benchmark by identifying visual markers on the cultural wall, confirms semantic regions by using differentiated texture features, enhances positioning accuracy by combining three-dimensional depth change regions, calculates interaction possibilities around interactive hot zones, captures user movement trajectories through the mobile phone camera, and obtains body orientation by using the mobile phone inertial measurement unit, thereby achieving accurate tracking of user position, attitude and viewing intention. In specific implementation, multi-sensor fusion adopts a tightly coupled fusion strategy on the mobile device, performing data fusion at the feature level; visual marker recognition uses an improved ARUCO recognition algorithm on the mobile device, maintaining a high recognition rate even under low light conditions; differential texture feature matching uses a bag-of-words model on the mobile device, quantifying texture features into visual words and quickly matching them through an inverted index; the 3D depth variation region utilizes the multi-view geometry principle on the mobile device, calculating depth information through the correspondence of feature points in multiple viewpoints; interaction probability calculation uses a deep learning model based on an attention mechanism on the mobile device, comprehensively considering the user's position, orientation, and movement trajectory; motion trajectory tracking uses a Kalman filter algorithm on the mobile device, effectively predicting the user's movement trend; body orientation estimation uses a quaternion representation on the mobile device to avoid gimbal lock problems; viewing intention recognition uses a long short-term memory network on the mobile device to analyze the user's historical behavior sequence. This purely mobile-based implementation establishes a precise spatial relationship model between the user and the cultural wall, providing accurate input for content generation and interactive responses.

[0030] A preferred implementation of the AI ​​content generation engine is that the AI ​​content generation engine is based on a cloud-based deep learning model and dynamically generates multiple types of virtual content according to user behavior: generating guide content when the user approaches a visual marker area, generating semantic interpretation when facing differentiated texture features, generating three-dimensional displays in areas with varying depth, creating interactive experiences in interactive hot zones, and achieving intelligent content creation through natural language generation technology, neural radiation field technology, and generative adversarial networks. In practical implementation, the deep learning model adopts a cloud-based multi-task learning architecture, sharing the underlying feature extraction network while using dedicated output layers for different tasks. The guide content generation uses a cloud-based Transformer-based text generation model, dynamically adjusting the level of detail based on user distance. Semantic interpretation generation combines cloud-based knowledge graph technology, extracting relevant information from knowledge bases such as Wikipedia. 3D display generation employs cloud-based neural radiation field technology, learning continuous volume representations of scenes from multiple images captured by a mobile phone to achieve rendering from any perspective. Interactive experience creation uses cloud-based physics-based simulation technology to ensure the physical realism of the interaction. Natural language generation technology uses cloud-based GPT series models, fine-tuned for cultural texts. Neural radiation field technology is implemented using cloud-based Instant-NGP, with a training speed 1000 times faster than traditional NeRF. The generative adversarial network uses a cloud-based StyleGAN2 architecture, generating content with a high degree of realism. This cloud-based generation and mobile display implementation achieves context-aware intelligent content creation, ensuring a high degree of relevance between virtual content and cultural scenes.

[0031] The preferred implementation of the spatial adaptation and rendering module is that the spatial adaptation and rendering module uses real-time spatial computing technology on mobile devices to calibrate the display benchmark based on visual marker points, optimize visual fusion by combining differentiated texture features, achieve depth fitting by utilizing three-dimensional depth variation areas, enhance interactive feedback for interactive hot areas, establish a spatial relationship model using mobile device visual SLAM technology, and ensure visual coordination through occlusion processing algorithms, thereby achieving a natural integration of virtual content and cultural wall. In practice, real-time spatial computing employs a mobile parallel computing architecture, leveraging the mobile GPU to accelerate spatial transformation calculations; display benchmark calibration uses a mobile iterative nearest-point algorithm to continuously optimize the alignment accuracy between virtual content and the real environment; visual fusion optimization utilizes a mobile deep learning-based image fusion network to achieve seamless virtual-real fusion; depth fitting employs mobile hierarchical depth testing to ensure correct interaction between virtual content and real depth; enhanced interactive feedback utilizes a mobile haptic rendering pipeline to provide corresponding haptic feedback based on the interaction type; visual SLAM technology uses a mobile ORB-SLAM3 system, supporting mobile monocular cameras; occlusion handling employs mobile depth buffer technology to correctly handle situations where virtual content is occluded by real objects; and the spatial relationship model uses a mobile scene graph representation to effectively manage complex environmental relationships. This mobile rendering implementation ensures accurate registration and visual coordination of virtual content in physical space, enhancing the immersive experience of augmented reality.

[0032] The preferred implementation of the intelligent response system lies in the fact that the system constructs an intelligent response system based on the feature set of the cultural wall. By analyzing the user's dwell time in visual marker areas, movement speed in front of differentiated texture features, observation angle in areas of three-dimensional depth change, and interaction frequency in interactive hotspots, the system dynamically adjusts the content display strategy, forming a progressive knowledge transfer mechanism. Specifically, dwell time analysis uses mobile time series analysis methods, employing an exponentially weighted moving average algorithm to smooth noise; movement speed analysis uses mobile optical flow calculations, combined with Kalman filtering to remove outliers; observation angle analysis uses mobile head posture estimation technology, calculating head orientation through facial key points; interaction frequency analysis uses mobile event counting statistics, incorporating a time decay factor to emphasize recent interactions; content display strategy adjustment uses cloud-based reinforcement learning algorithms, continuously optimizing the strategy based on user feedback; the knowledge transfer mechanism uses cognitive load theory to ensure that information presentation does not exceed the user's cognitive processing capacity; and progressive display uses knowledge graph technology to organize the content presentation order according to concept dependencies. This mobile-based perception and cloud-based decision-making implementation achieves intelligent navigation based on user behavior analysis, improving the accuracy and effectiveness of cultural dissemination.

[0033] A preferred implementation of eye-tracking technology involves integrating mobile eye-tracking technology into the system. This involves intelligently adjusting the timing and level of detail of virtual content display by detecting the user's gaze duration at visual markers, the area of ​​interest for differentiated texture features, the focus of observation on areas of stereoscopic depth variation, and the degree of attention to interactive hotspots. Specifically, eye-tracking employs a deep learning-based method on the mobile device, using a convolutional neural network to directly estimate the gaze direction from eye images. Gaze duration detection uses a threshold segmentation algorithm on the mobile device, combined with temporal clustering to identify valid gaze events. Area of ​​interest analysis uses heatmap generation technology on the mobile device, identifying regions of interest through a Gaussian mixture model. Focus positioning uses 3D gaze mapping technology on the mobile device, back-projecting 2D gaze points into 3D space. Attention level quantification uses a multi-indicator comprehensive evaluation on the mobile device, including gaze duration, number of gazes, and pupil diameter changes. Content display timing control uses a finite state machine on the mobile device, switching display stages according to the user's attention state. Content detail adjustment uses an information density evaluation model on the mobile device, dynamically adjusting the amount of information based on the user's cognitive load. This mobile eye-tracking implementation achieves intelligent adjustment of content presentation through eye tracking, enhancing the naturalness and smoothness of the user experience.

[0034] A preferred implementation of multimodal fusion recognition is that the system employs mobile multimodal fusion recognition technology. By coordinating the interactive behaviors of multiple users in different areas of the cultural wall, when collaborative localization in visual marker areas, collective observation in front of differentiated texture features, synchronous exploration in areas of 3D depth variation, or collaborative interaction in interactive hotspots are detected, associated virtual content is generated using cloud-based distributed rendering technology to ensure the consistency of the multi-user interactive experience. Specifically, multimodal fusion uses a mobile attention mechanism to dynamically adjust the weights of different modal features; collaborative localization detection uses a mobile multi-target tracking algorithm, employing a joint probability data association filter to handle occlusion; collective observation recognition uses a mobile group behavior analysis model, using a social force model to detect aggregation behavior; synchronous exploration analysis uses mobile spatiotemporal consistency detection, employing a dynamic time warping algorithm to analyze the movement patterns of multiple users; collaborative interaction recognition uses mobile multi-agent reinforcement learning to model the collaborative strategies between users; distributed rendering uses a cloud client-server architecture, using a state synchronization protocol to ensure the consistency of multi-user views; and associated virtual content generation uses cloud-based programmatic content generation technology to dynamically generate corresponding content based on group behavior characteristics. This mobile-based sensing and cloud-based generation approach supports multi-user collaborative interaction, enhancing the social and engaging aspects of cultural dissemination.

[0035] A preferred implementation of personalized visitor profiles involves establishing a personalized visitor profile system. This system records user trajectories in visually marked areas, choices made in front of differentiated texture features, exploration depth in areas of varying depth, and interaction preferences in interactive hotspots. It then uses a cloud-based machine learning model to construct user interest profiles, optimizing subsequent content recommendation strategies. Specifically, the visitor profile system employs a cloud-based time-series database for efficient storage and retrieval of user behavior data. Trajectory recording uses a mobile spline interpolation algorithm to smooth the original discrete trajectory points. Selection recording uses a mobile multi-armed machine learning algorithm to model user selection preferences. Exploration depth analysis uses a mobile path integral algorithm, comprehensively considering movement distance and dwell time. Interaction preference analysis uses a cloud-based collaborative filtering algorithm to discover potential user interest patterns. The machine learning model uses a cloud-based gradient boosting decision tree to predict user interests based on multiple features. Interest profile construction uses a cloud-based latent factor model to infer implicit interests from explicit behavior. The content recommendation strategy uses cloud-based multi-objective optimization to balance indicators such as novelty, relevance, and diversity. This mobile-based data collection and cloud-based analysis implementation achieves personalized services based on user profiles, enhancing the relevance and satisfaction of the cultural experience.

[0036] A preferred implementation of the multi-mode guided tour system is that the system constructs a multi-mode guided tour system, which automatically switches guided tour modes by recognizing the user's basic needs in visual marker areas, willingness to explore in depth in front of differentiated texture features, spatial perception ability in areas with three-dimensional depth changes, and participation level in interactive hot zones. It uses cloud-based reinforcement learning, path planning, and task generation algorithms to achieve a personalized cultural exploration experience. In its implementation, the multi-mode guided tour system adopts a cloud-based layered architecture, supporting dynamic combination and switching of modes. Basic needs identification uses a mobile-based Naive Bayes classifier to determine the level of needs based on users' basic behavioral characteristics. In-depth exploration willingness assessment uses a mobile-based logistic regression model, combining historical behavior and real-time status for prediction. Spatial perception ability evaluation uses a mobile-based cognitive ability test task, assessing ability levels through user performance in specific tasks. Participation measurement uses a multi-dimensional indicator system on the mobile device, including interaction frequency, duration, and level of engagement. Guided tour mode switching uses a mobile-based finite state machine to ensure smooth mode transitions. The reinforcement learning algorithm uses cloud-based near-end strategy optimization to achieve a balance between exploration and utilization. Path planning uses a cloud-based A* algorithm, adjusting path weights based on the importance of cultural content. The task generation algorithm uses a cloud-based Monte Carlo tree search to generate exploration tasks that are both interesting and educational. This mobile-based perception and cloud-based decision-making implementation provides adaptive guided tour mode selection, meeting the diverse cultural experience needs of different users.

[0037] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "include," "contain," or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or apparatus that includes a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or apparatus. Without further limitations, an element defined by the phrase "includes…" does not exclude the presence of additional identical elements in the article or apparatus that includes that element.

[0038] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. The present invention has been described in detail with reference to preferred embodiments. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An AI-interactive AR (augmented reality) cultural tourism wall system, characterized in that: The surface of the cultural wall is equipped with a set of pre-embedded features, including visual markers, differentiated texture features, three-dimensional depth variation areas, and interactive hot zones with specific designs, which provide the system with stable visual references, semantic distinction criteria, depth cues, and precise positioning anchors, respectively. The system includes: a feature point acquisition module, used to acquire the feature set of the cultural wall through a multimodal feature extraction algorithm, and construct a 3D spatial map containing semantic annotations; The user positioning and attitude estimation module receives a 3D spatial map from the feature point acquisition module. By integrating computer vision positioning, inertial sensor data and wireless positioning technology, it tracks the user's spatial position, body orientation, joint posture and line of sight relative to the cultural wall in real time. The AI ​​content generation engine dynamically generates virtual content that matches the theme of the cultural wall, based on the semantic features of the cultural wall from the feature point acquisition module and the user spatial relationship data from the user positioning and posture estimation module. The spatial adaptation and rendering module simultaneously receives a 3D spatial map from the feature point acquisition module and virtual content from the AI ​​content generation engine. It accurately registers the virtual content into the 3D coordinate system of the cultural wall, performs physically realistic rendering and compositing, and finally presents the composite effect through an AR application on the user's mobile device. It also integrates media recording functions to capture and save the composite image containing the user's image and virtual content.

2. The AI ​​interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The feature point acquisition module analyzes the visual markers and differentiated texture features set on the surface of the cultural wall through multi-scale feature extraction technology, identifies the outlines and content of each cultural element by combining semantic segmentation algorithm, obtains the geometric information of the three-dimensional depth variation area by using 3D scanning technology, and sets interactive anchor points in the interactive hotspots of a specific design. Finally, it establishes a cultural element database containing semantic tags, spatial coordinates and interactive attributes, providing a data foundation for the accurate mapping of virtual content.

3. The AI ​​interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The user positioning and attitude estimation module adopts multi-sensor fusion technology. It establishes a positioning benchmark by identifying visual markers on the cultural wall, confirms semantic regions by using differentiated texture features, enhances positioning accuracy by combining three-dimensional depth variation regions, calculates interaction possibilities around interactive hotspots, captures user movement trajectories through cameras, and obtains body orientation using inertial measurement units, thereby achieving accurate tracking of user position, attitude, and viewing intention.

4. The AI ​​interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The AI ​​content generation engine is based on a deep learning model and dynamically generates multiple types of virtual content according to user behavior: it generates guide content when the user approaches a visual marker area, generates semantic interpretation when faced with differentiated texture features, generates 3D displays in areas with varying depth, and creates interactive experiences in interactive hot zones. It achieves intelligent content creation through natural language generation technology, neural radiation field technology, and generative adversarial networks.

5. The AI-interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The spatial adaptation and rendering module uses real-time spatial computing technology to calibrate the display benchmark based on visual marker points, optimize visual fusion by combining differentiated texture features, achieve depth fitting by utilizing three-dimensional depth variation areas, enhance interactive feedback for interactive hotspots, establish a spatial relationship model using visual SLAM technology, and ensure visual coordination through occlusion processing algorithms, thereby achieving a natural integration of virtual content and the cultural wall.

6. The AI ​​interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The system constructs an intelligent response system based on the feature set of the cultural wall. By analyzing the user's dwell time in the visual marker area, the movement speed in front of the differentiated texture features, the observation angle in the three-dimensional depth change area, and the interaction frequency in the interactive hot zone, the system dynamically adjusts the content display strategy to form a progressive knowledge transfer mechanism.

7. The AI ​​interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The system integrates eye-tracking technology, which intelligently adjusts the timing and level of detail of virtual content display by detecting the user's gaze duration at visual markers, the area of ​​attention for differentiated texture features, the focus of observation on areas of three-dimensional depth change, and the degree of attention to interactive hot spots, thereby achieving adaptive content presentation based on visual attention.

8. The AI ​​interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The system employs multimodal fusion recognition technology. By coordinating the interactive behaviors of multiple users in different areas of the cultural wall, when it detects collaborative positioning in visual marker areas, collective observation in front of differentiated texture features, synchronous exploration in areas with varying depth, or collaborative interaction in interactive hot zones, it uses distributed rendering technology to generate associated virtual content, ensuring the consistency of the multi-user interactive experience.

9. The AI-interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The system establishes a personalized visitor profile system. By recording the user's trajectory in visual marker areas, choices in front of differentiated texture features, exploration depth in areas with varying 3D depth, and interaction preferences in interactive hotspots, the system uses machine learning models to construct user interest profiles and optimize subsequent content recommendation strategies.

10. The AI-interactive AR augmented reality cultural tourism wall system according to claim 1, characterized in that, The system constructs a multi-mode tour guide system. By recognizing users' basic needs in visual marker areas, their willingness to explore in depth in front of differentiated texture features, their spatial perception ability in areas with changes in three-dimensional depth, and their level of participation in interactive hot zones, the system automatically switches tour guide modes and uses reinforcement learning, path planning, and task generation algorithms to achieve a personalized cultural exploration experience.