Method and system for digital media rendering in virtual reality

Abstract

A method and system for digital media rendering in virtual reality. The method comprises: acquiring, in real time, an HDR image, depth information and user posture data in a virtual reality environment by means of a high-definition camera and a sensor, and performing time synchronization to generate a synchronous sensing data set; analyzing a data change, adjusting the angle of view and resolution of a virtual scene, and optimizing a rendering effect; monitoring a user interaction behavior, updating a user behavior model, and establishing a dynamic user model; predicting a future interaction behavior of a user, analyzing the impact of interaction, and calculating a key interaction path; and using a multi-kernel learning method to adjust kernel parameters, so as to generate a user behavior analysis result.

Description

A digital media rendering method and system in virtual reality TECHNICAL FIELD

[0001] The present application belongs to the technical field of virtual reality, and specifically relates to a digital media rendering method and system in virtual reality. BACKGROUND

[0002] With the rapid development of virtual reality technology, digital media rendering technology has become a key component in building immersive experiences. However, existing digital media rendering technology still has certain deficiencies in handling complex scenes in virtual reality, efficient rendering, and user interaction, which affects user experience and application effect.

[0003] After searching, a media resource pushing method and device with publication number CN104469477B and publication date August 11, 2017. The patent provides a media resource pushing method and device, which encapsulates the URL and ID3 attribute information of the media resource to be transmitted as metadata and sends it to the digital media renderer (DMR), so as to realize the pushing of the media resource on the Internet to the local DMR for playing. This method simplifies the user's operation and improves the compatibility. However, this technical solution mainly focuses on the transmission and pushing of media resources, and lacks support for complex scene rendering and user interaction in virtual reality environment, especially in the processing of high dynamic range images (HDR) and stereo sound effects. In addition, this method has high requirements for network delay and bandwidth, which may affect real-time performance and smoothness.

[0004] After searching, a method for realizing virtual product visualization and interaction in CAVE environment with publication number CN104657096B and publication date February 23, 2018. The patent proposes a method for realizing virtual product visualization and interaction in CAVE environment, which integrates virtual reality technology, digital mockup technology and digital media technology to optimize the construction and real-time rendering of virtual product visualization model. This method is suitable for virtual product visualization and interaction of multiple helicopter models, and has high practicability and flexibility. However, this technical solution is mainly suitable for specific CAVE systems and professional application scenarios, and its expansibility and adaptability are limited for more extensive virtual reality environment and consumer-level applications. In addition, this method has high requirements for computing resources and rendering speed when dealing with large-scale complex scenes, which may affect user experience and system stability. TECHNICAL PROBLEM

[0005] The aforementioned problems indicate that existing digital media rendering technologies still have certain shortcomings in handling complex scenes, efficient rendering, user interaction support, and effective utilization of computing resources in virtual reality environments. Therefore, this invention provides a digital media rendering technology for virtual reality, aiming to optimize the rendering effects of complex scenes, improve rendering efficiency, enhance user interaction experience, and reduce system resource consumption, thereby meeting the demands of virtual reality applications for efficient and high-quality digital media rendering. Technical solutions

[0006] The purpose of this invention is to address the shortcomings of existing technologies by proposing a digital media rendering method and system for virtual reality.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a digital media rendering method in virtual reality, comprising the following steps:

[0008] High-definition cameras and sensors are used to acquire HDR images, depth information and user posture data in the virtual reality environment in real time, and timestamp information is added to the data. Time synchronization technology is used to align the data of multiple types of sensors to obtain a synchronous perception dataset.

[0009] The system analyzes the data changes in the synchronous perception dataset, adjusts the viewpoint and resolution of the virtual scene according to the user's posture, optimizes the rendering effect based on depth information, and uses deep learning methods to evaluate the potential impact of user interaction, dynamically generates scene configurations that match the current data, and generates an optimized virtual scene.

[0010] Using the optimized virtual scene, monitor current virtual reality data, analyze user interaction behavior, update the user behavior model based on the interaction data, automatically integrate data into the user behavior model, update the data through the automatic update mechanism of the user behavior model, and establish a dynamic user model.

[0011] Based on the dynamic user model, predictions are made to forecast future user interaction behaviors, analyze the impact of interaction behaviors on virtual scenarios, and obtain user behavior prediction and interaction analysis results by calculating key interaction paths.

[0012] The user behavior prediction and interaction analysis results are processed, and the data is analyzed in a high-dimensional space using a multi-core learning method. The parameters of multiple kernels are adjusted to match the data features, and user behavior analysis results after kernel parameter adjustment are generated.

[0013] The synchronous perception dataset specifically includes HDR image data, depth data, and user pose data, the optimized virtual scene includes perspective, resolution, and rendering effect, the dynamic user model includes a user behavior model and an automatic updating mechanism of the user behavior model, and the user behavior prediction and interaction analysis result includes future interaction behavior, interaction influence, and key interaction path.

[0014] As a further scheme of the present application, the step of acquiring the synchronous perception dataset specifically comprises:

[0015] S11, collecting data and adding time stamps: collecting HDR image data, depth data, and user pose data through high-definition cameras and sensors installed on virtual reality devices, adding time stamps to each data point, and obtaining a time-labeled original dataset;

[0016] S12, time alignment processing: applying a time alignment algorithm to process the time-labeled original dataset, aligning the HDR image, depth information, and user pose data on the same time axis by adjusting the data time stamps, and obtaining a time-aligned sensor dataset;

[0017] S13, calculating comprehensive perception indicators: analyzing the time-aligned sensor dataset and calculating the perception indicators of all sensor data at each time point, using the formula: t =αH t +βD t +γU t

[0018] to obtain a comprehensive perception dataset;

[0019] wherein p t represents the comprehensive perception indicator at a single time point, H t represents the HDR image data value at a single time point, D t represents the depth data value at a single time point, U t represents the user pose data value at a single time point, α is the weight coefficient of the HDR image data for adjusting the influence of the HDR image data in the comprehensive perception indicator, β is the weight coefficient of the depth data for adjusting the influence of the depth data in the comprehensive perception indicator, and γ is the weight coefficient of the user pose data for adjusting the influence of the user pose data in the comprehensive perception indicator;

[0020] S14, integrating to generate a synchronous perception dataset: integrating and formatting the comprehensive perception dataset to generate a synchronous perception dataset.

[0021] As a further scheme of the present application, the step of acquiring the optimized virtual scene specifically comprises:

[0022] S21, generating time series posture data: collecting user posture data in the synchronized perception dataset, recording the posture situation of each user at different time points, and generating time series posture data of each user;

[0023] S22, analyzing posture data and generating statistical results: analyzing the motion mode of user posture from the time series posture data of each user, calculating the standard deviation and mean value of user posture, and generating user posture statistical results;

[0024] S23, adjusting the virtual scene configuration according to the statistical results: based on the user posture statistical results, dynamically adjusting the perspective and resolution of the virtual scene according to the user posture, using the formula:

[0025] generate the adjusted virtual scene configuration;

[0026] wherein, C new represents the adjusted virtual scene configuration of the user, p current represents the real-time posture of the user, P max represents the maximum posture range in all users during observation, C old represents the original virtual scene configuration of the user, and ∈ is a small adjustment factor introduced to avoid zero denominator and increase the stability of calculation, used to enhance the adjustment coefficient of the virtual scene configuration;

[0027] S24, applying the adjusted configuration and optimizing the virtual scene: applying the adjusted virtual scene configuration to the entire virtual scene, dynamically matching user posture changes according to real-time data, optimizing the perspective and resolution of the virtual scene, and generating an optimized virtual scene.

[0028] As a further scheme of the present application, the acquisition step of the dynamic user model is specifically:

[0029] S31, monitoring virtual reality data and obtaining a preliminary interaction dataset: using the optimized virtual scene to monitor the current virtual reality data, analyzing user interaction behavior, and obtaining a preliminary user interaction dataset;

[0030] S32, updating the user behavior model and generating an updated model matrix: updating the user behavior model according to the user interaction data in the preliminary user interaction dataset, calculating and adjusting the statistical values of user behavior, and generating an updated user behavior model matrix;

[0031] S33, integrating the updated model matrix and generating a dynamic user model: automatically integrating the updated user behavior model matrix into the user behavior model, using the formula:

[0032] updating the user behavior model, dynamically predicting user behavior and system state according to real-time data, and generating a dynamic user model;

[0033] Wherein, P(A|B) represents the conditional probability of user A behavior occurring under the condition that user B behavior occurs, reflecting the predictability of user A behavior occurring under the condition that user B behavior occurs, P(A) represents the marginal probability of user A, P(B) represents the marginal probability of user B, and P(B|A) represents the conditional probability of user B behavior occurring under the condition that user A behavior occurs, which indicates the probability of user B behavior occurring when user A behavior has occurred.

[0034] As a further scheme of the present application, the step of obtaining the user behavior prediction and interaction analysis result is specifically:

[0035] S41, predicting future interaction behavior: based on the dynamic user model, predicting future interaction behavior of the user by using statistical analysis method, and obtaining prediction data of future interaction behavior;

[0036] S42, analyzing the influence of user behavior: analyzing the influence of user behavior in the virtual scene from the prediction data of future interaction behavior, referring to the characteristics of the virtual scene and the user preference, and generating an analysis result of the influence of user behavior;

[0037] S43, calculating the key interaction path: according to the analysis result of the influence of user behavior, calculating the key interaction path, using the formula:

[0038] The calculation result of the key interaction path is generated;

[0039] Wherein, I represents the key influence strength of user interaction, p i represents the probability contribution of user i, is the average value of probability, and σ p is the standard deviation of probability, d i represents the distance from user i to the next user, and n represents the total number of users in the path;

[0040] S44, generating the user behavior prediction and interaction analysis result: using the calculation result of the key interaction path, integrating the virtual scene model and the probability data, and generating the user behavior prediction and interaction analysis result.

[0041] As a further scheme of the present application, the step of obtaining the user behavior analysis result after adjusting the kernel parameters is specifically:

[0042] S51, generating a preliminary user behavior analysis result: analyzing the user behavior prediction and interaction analysis result, using a multi-core learning method to process data in a high-dimensional space, assigning a preliminary kernel parameter to each data point, and generating a user behavior analysis result;

[0043] S52, adjusting kernel parameters and generating an adjustment scheme: based on the user behavior analysis result, adjusting the parameters of each kernel to match the characteristics of the data, adjusting the weights through linear algebra and optimization algorithms, and generating a parameter adjustment scheme;

[0044] S53, applying the adjustment scheme and generating the final analysis result: applying the parameter adjustment scheme, recalculating the user behavior, and using the formula:

[0045] reflecting the adjusted user behavior through a weighted Gaussian function, generating a user behavior analysis result after adjusting the kernel parameters;

[0046] where R(x) represents the response of the user behavior, reflecting the prediction result obtained by the multi-core learning algorithm through differentiated kernel function weighted summation, w j is the adjustment parameter of the jth kernel, representing the weight of the kernel in the total user behavior, x is the input variable, representing the data characteristics, μ j is the mean parameter of the jth kernel function, describing the center position of the data distribution corresponding to the kernel, and σ j is the standard deviation parameter of the jth kernel function, indicating the dispersion degree of the data distribution.

[0047] A digital media rendering system in virtual reality, the digital media rendering system in virtual reality is used to execute the above-mentioned digital media rendering method in virtual reality, the system comprises:

[0048] The data acquisition module acquires the HDR image, depth information and user posture data in the virtual reality environment in real time through the high-definition camera and sensor, adds timestamp information, aligns the multi-type sensor data in time using time synchronization technology, and generates a synchronous perception dataset;

[0049] The scene optimization module adjusts the perspective and resolution of the virtual scene based on the HDR image data, depth data and user posture data in the synchronous perception dataset, updates the rendering effect, evaluates the potential impact of user interaction, reconfigures the virtual scene, and generates an optimized virtual scene;

[0050] The behavior modeling module monitors the current HDR image data, depth data and user posture data based on the optimized virtual scene, analyzes the user interaction behavior, updates the user behavior model, integrates the data into the user behavior model, and establishes a dynamic user model;

[0051] The prediction analysis module predicts based on the dynamic user model, predicts future interaction behavior of the user, analyzes the influence of the user behavior in the virtual scene, calculates a key interaction path, obtains a user behavior prediction and interaction analysis result, processes the user behavior prediction and interaction analysis result, analyzes HDR image data, depth data and user posture data using a multi-core learning method, adjusts kernel parameters, and generates a user behavior analysis result. Advantages:

[0052] In the present application, through the construction and real-time updating of the synchronous perception dataset, detailed monitoring of user behavior and scene changes in the virtual reality environment is provided. By dynamically analyzing the changes in user posture and depth information, the perspective and resolution of the virtual scene are effectively optimized, enhancing the immersion of users in the virtual reality environment. Using the automatic updating mechanism of the user behavior model and the dynamic user model, rapid prediction and response to future user interaction behavior are achieved, improving prediction accuracy and decision efficiency. In addition, through the application of multi-core learning methods, user behavior can be more accurately analyzed and predicted in high-dimensional space, thereby improving user experience and system stability and smoothness of virtual reality applications. BRIEF DESCRIPTION OF DRAWINGS

[0053] Fig. 1 is a workflow diagram of a digital media rendering method in virtual reality according to the present application;

[0054] Fig. 2 is a flowchart of the acquisition steps of the synchronous perception dataset according to the present application;

[0055] Fig. 3 is a flowchart of the acquisition steps of the optimized virtual scene according to the present application;

[0056] Fig. 4 is a flowchart of the acquisition steps of the dynamic user model according to the present application;

[0057] Fig. 5 is a flowchart of the acquisition steps of the user behavior prediction and interaction analysis result according to the present application;

[0058] Fig. 6 is a flowchart of the acquisition steps of the user behavior analysis result after kernel parameter adjustment according to the present application;

[0059] Fig. 7 is a flowchart of a digital media rendering system in virtual reality according to the present application. Embodiments of the present application

[0060] The present application provides a digital media rendering method and system in virtual reality, which is characterized by real-time acquisition and synchronous processing of multi-sensor data, dynamic optimization of virtual scene configuration, automatic updating of user behavior model, and high-dimensional data analysis based on multi-core learning, to achieve accurate prediction and efficient response of user interaction behavior in virtual reality environment. The specific embodiments of the present application are described in detail below with reference to Figs. 1 to 7.

[0061] In practical applications, the implementation process of the present application first starts from the data acquisition module, as shown in FIG. 1. This module acquires HDR image data, depth information and user pose data in real time through high-definition cameras and various sensors installed on virtual reality devices, and adds timestamp information to each data point. For example, in a specific embodiment, the high-definition camera collects HDR image data at a speed of 30 frames per second, while the depth sensor records environmental depth information at the same frequency, and the inertial measurement unit (IMU) is used to capture user head pose and hand action data. After these data are marked with accurate timestamps, a time-labeled raw data set is formed. Subsequently, the data is processed by a time alignment algorithm to ensure that all sensor data is aligned on the same time axis. Specifically, the time alignment algorithm adjusts the time reference of the data according to the timestamp differences of each sensor, thereby generating a time-aligned sensor data set. On this basis, the comprehensive perception index at each time point is further calculated, and the formula is as follows: p t = aH t + bD t + gU t

[0062] where p t represents the comprehensive perception index at a single time point, H t represents the HDR image data value at a single time point, D t represents the depth data value at a single time point, U t represents the user pose data value at a single time point, and a, b and g are weight coefficients of the HDR image data, depth data and user pose data, respectively, used to adjust the influence of each type of data in the comprehensive perception index. By performing the above calculation on the time-aligned sensor data set, a synchronized perception data set is finally generated, as shown in FIG. 2.

[0063] Next, the scene optimization module dynamically adjusts the perspective and resolution of the virtual scene based on the HDR image data, depth data and user pose data in the synchronized perception data set, and optimizes the rendering effect. The specific implementation steps are shown in FIG. 3. First, collect the user pose data in the synchronized perception data set, record the pose of each user at different time points, and generate time series pose data for each user. Then, analyze these time series pose data, calculate the standard deviation and mean of the user pose, and generate user pose statistical results. Based on these statistical results, the perspective and resolution of the virtual scene are dynamically adjusted using the following formula:

[0064] where C new represents the adjusted virtual scene configuration of the user, C old represents the original virtual scene configuration of the user, and p currentP represents the real-time pose of the user, max P represents the maximum pose range among all users during observation, ∈ is a small adjustment factor to avoid the denominator being zero and increase the stability of calculation, k is an adjustment coefficient of the augmented virtual scene configuration. Through the above formula, the system can dynamically match the changes of the virtual scene according to the real-time pose of the user, thereby optimizing the perspective and resolution of the virtual scene. In addition, depth information is also used to optimize rendering effects, such as adjusting the distribution of light and shadow through depth data, making the virtual scene more realistic.

[0065] After generating the optimized virtual scene, the behavior modeling module begins to monitor the current virtual reality data, analyze user interaction behavior, and update the user behavior model. As shown in FIG. 4, the module first uses the optimized virtual scene to monitor the current HDR image data, depth data, and user pose data, analyzes the user's interaction behavior, and obtains a preliminary user interaction data set. Then, according to the user interaction data in the preliminary user interaction data set, the user behavior model is updated, the statistical value of the user behavior is calculated and adjusted, and the updated user behavior model matrix is generated. Specifically, the process of updating the user behavior model uses the following formula:

[0066] Where P(A|B) represents the conditional probability of user A behavior occurring under the condition that user B behavior occurs, reflecting the estimated nature of user A behavior occurring under the condition that user B behavior occurs; P(A) and P(B) represent the marginal probability of user A and user B, respectively; P(B|A) represents the conditional probability of user B behavior occurring under the condition that user A behavior occurs. Through the above formula, the system can dynamically predict user behavior and system state, and generate a dynamic user model.

[0067] On the basis of the dynamic user model, the prediction analysis module further predicts the user's future interaction behavior and analyzes the impact of the interaction behavior in the virtual scene. As shown in FIG. 5, the module first makes predictions based on the dynamic user model, predicts the user's future interaction behavior using statistical analysis methods, and obtains prediction data of future interaction behavior. Then, from the prediction data of future interaction behavior, the impact of user behavior in the virtual scene is analyzed, the analysis results of user behavior impact are generated by referring to the characteristics of the virtual scene and user preferences. Finally, according to the user behavior impact analysis results, the key interaction path is calculated, and the following formula is used:

[0068] Where I represents the key influence strength of user interaction, p i represents the probability contribution of user i, is the average value of probability, σ p is the standard deviation of probability, d iLet N represent the distance from user i to the next user, and N be the total number of users in the path. Using the above formula, the system can generate calculation results for key interaction paths and, by integrating the virtual scene model and probability data, generate user behavior prediction and interaction analysis results.

[0069] Finally, to further improve the accuracy of user behavior analysis, the system employs a multi-core learning method to process data in a high-dimensional space, adjusting the parameters of multiple kernels to match data features. As shown in Figure 6, this process first analyzes the user behavior prediction and interaction analysis results, assigns preliminary kernel parameters to each data point, and generates user behavior analysis results. Then, based on the user behavior analysis results, the parameters of each kernel are adjusted to ensure that each parameter matches the characteristics of the data. Weights are adjusted using linear algebra and optimization algorithms to generate a parameter adjustment scheme. Specifically, the process of recalculating user behavior uses the following formula:

[0070] Where R(x) represents the user's response, reflecting the prediction result obtained by the multi-kernel learning algorithm through weighted summation of differentiated kernel functions; w j μ is the adjustment parameter of the j-th kernel, representing the weight of the kernel in the total user behavior; x is the input variable, representing the data feature; μ j σ is the mean parameter of the j-th kernel function, describing the central location of the data distribution corresponding to the kernel; j is the standard deviation parameter of the j-th kernel function, indicating the dispersion of the data distribution. Using the above formula, the system can generate user behavior analysis results after adjusting the kernel parameters.

[0071] To implement the aforementioned digital media rendering method in virtual reality, a digital media rendering system in virtual reality is provided, as shown in Figure 7. The system includes:

[0072] The data acquisition module acquires HDR images, depth information and user posture data in the virtual reality environment in real time through high-definition cameras and sensors, adds timestamp information, and uses time synchronization technology to time-align data from multiple types of sensors to generate a synchronous perception dataset.

[0073] The scene optimization module adjusts the viewpoint and resolution of the virtual scene, updates the rendering effect, assesses the potential impact of user interaction, reconfigures the virtual scene, and generates an optimized virtual scene based on the HDR image data, depth data, and user pose data in the synchronous perception dataset.

[0074] Based on the optimized virtual scene, the behavior modeling module monitors the current HDR image data, depth data, and user posture data, analyzes user interaction behavior, updates the user behavior model, integrates data into the user behavior model, and establishes a dynamic user model.

[0075] The predictive analysis module makes predictions based on the dynamic user model, predicts the user's future interactive behavior, analyzes the impact of user behavior in the virtual scene, calculates key interaction paths, obtains user behavior prediction and interaction analysis results, processes the user behavior prediction and interaction analysis results, uses a multi-core learning method to analyze HDR image data, depth data and user pose data, adjusts kernel parameters, and generates user behavior analysis results.

[0076] In summary, this invention, through the collaborative work of a data acquisition module, a scene optimization module, a behavior modeling module, and a predictive analysis module, achieves detailed monitoring and dynamic response to user behavior and scene changes in a virtual reality environment. In practical applications, such as virtual reality games or virtual training systems, this invention can significantly enhance user immersion and interactive experience, while improving system stability and smoothness.

Claims

1. A method for rendering digital media in virtual reality, the method comprising: The method comprises the following steps: Real-time acquisition of HDR image data, depth information and user posture data in a virtual reality environment through high-definition cameras and sensors, and addition of timestamp information to the data, time alignment of multiple types of sensor data through time synchronization technology, and obtaining of a synchronized perception dataset; Analysis of data changes in the synchronized perception dataset, adjustment of the perspective and resolution of the virtual scene according to the user posture, optimization of the rendering effect according to the depth information, simultaneous evaluation of the potential impact of user interaction using a deep learning method, dynamic generation of a scene configuration matching the current data, and generation of an optimized virtual scene; Using the optimized virtual scene, monitoring of the current virtual reality data, analysis of user interaction behavior, updating of the user behavior model according to the interaction data, automatic integration of the data into the user behavior model, data updating through the automatic updating mechanism of the user behavior model, and establishment of a dynamic user model; Based on the dynamic user model, prediction of future user interaction behavior, analysis of the impact of the interaction behavior in the virtual scene, and obtaining of user behavior prediction and interaction analysis results through calculation of a key interaction path; Processing the user behavior prediction and interaction analysis results, analyzing the data in a high-dimensional space using a multi-core learning method, adjusting the parameters of multiple cores to match the data characteristics, and generating a user behavior analysis result after adjustment of the core parameters.

2. The method of claim 1, wherein, The synchronized perception dataset specifically comprises HDR image data, depth data and user posture data, the optimized virtual scene comprises perspective, resolution and rendering effect, the dynamic user model comprises a user behavior model and an automatic updating mechanism of the user behavior model, the user behavior prediction and interaction analysis results comprise future interaction behavior, interaction impact and a key interaction path, and the user behavior analysis result comprises learning method application, data analysis dimension and core parameter adjustment.

3. The method of claim 2, wherein, The acquisition step of the synchronized perception dataset specifically comprises: S11, collecting data and adding a timestamp: collecting HDR image data, depth data and user posture data through high-definition cameras and sensors installed on a virtual reality device, adding a timestamp to each data point, and obtaining a time-labeled original dataset; S12, time alignment processing: applying a time alignment algorithm to process the time-labeled original dataset, aligning the HDR image, depth information and user posture data on the same time axis by adjusting the data timestamp, and obtaining a time-aligned sensor dataset; S13, calculating a comprehensive perception index: analyzing the time-aligned sensor dataset and calculating the perception index of all sensor data at each time point, using the formula: p t = aH t + bD t + gU t to obtain a comprehensive perception dataset; wherein p t represents the integrated perception index at a single time point, H t represents the HDR image data value at a single time point, D t represents the depth data value at a single time point, U t represents the user pose data value at a single time point, a is a weight coefficient of the HDR image data, used to adjust the influence of the HDR image data in the integrated perception index, β is a weight coefficient of the depth data, used to adjust the influence of the depth data in the integrated perception index, and γ is a weight coefficient of the user pose data, used to adjust the influence of the user pose data in the integrated perception index. S14, integrating and generating a synchronized perception dataset: integrating and formatting the comprehensive perception dataset to generate a synchronized perception dataset.

4. The method of claim 3, wherein, The acquisition step of the optimized virtual scene specifically comprises: S21, generating time series posture data: collecting user posture data in the synchronized perception dataset, recording the posture of each user at different time points, and generating time series posture data for each user; S22, analyze the posture data and generate statistical results: analyze the motion patterns of the user posture from the time series posture data of each user, calculate the standard deviation and mean of the user posture, and generate user posture statistical results; S23, adjusting the virtual scene configuration according to the statistical result: based on the user posture statistical result, dynamically adjusting the perspective and resolution of the virtual scene according to the user posture, using the formula: generate an adjusted virtual scene configuration; where C new represent the user-adjusted virtual scene configuration, p current represent the user's real-time pose, P max represent the maximum pose range in all users during observation, C old represent the user's original virtual scene configuration, and ε is a small adjustment factor introduced to avoid a zero denominator and increase the stability of the calculation, used to enhance the adjustment coefficient of the virtual scene configuration; S24, apply the adjusted configuration and optimize the virtual scene: apply the adjusted virtual scene configuration to the entire virtual scene, dynamically match user posture changes according to real-time data, optimize the perspective and resolution of the virtual scene, and generate an optimized virtual scene.

5. The method of claim 4, wherein, The obtaining step of the dynamic user model is specifically: S31, monitor virtual reality data and obtain a preliminary interaction data set: monitor the current virtual reality data using the optimized virtual scene, analyze user interaction behavior, and obtain a preliminary user interaction data set; S32, update the user behavior model and generate an updated model matrix: update the user behavior model according to the user interaction data in the preliminary user interaction data set, calculate and adjust the statistical values of the user behavior, and generate an updated user behavior model matrix; S33, integrate the updated model matrix and generate a dynamic user model: the updated user behavior model matrix is automatically integrated into the user behavior model, using the formula: update the user behavior model, dynamically predict user behavior and system state according to real-time data, and generate a dynamic user model; Wherein, P(A|B) represents the conditional probability of user A behavior occurring under the condition that user B behavior occurs, reflecting the predictability of user A behavior occurring under the condition that user B behavior occurs, P(A) represents the marginal probability of user A, P(B) represents the marginal probability of user B, and P(B|A) represents the conditional probability of user B behavior occurring under the condition that user A behavior occurs, which indicates the probability of user B behavior occurring when user A behavior has already occurred.

6. The method of claim 5, wherein, The obtaining step of the user behavior prediction and interaction analysis result is specifically: S41, predict future interaction behavior: based on the dynamic user model, predict future interaction behavior using statistical analysis methods to obtain prediction data of future interaction behavior; S42, analyze the influence of user behavior: analyze the influence of user behavior in the virtual scene from the prediction data of future interaction behavior, refer to the characteristics of the virtual scene and user preferences, and generate analysis results of the influence of user behavior; S43, calculating the key interaction path: according to the user behavior influence analysis result, calculating the key interaction path, using the formula: generate a calculation result of the key interaction path; where I denotes the key influence strength of the user interaction, p i representing the probability contribution of user i, is the average value of the probability, σ p is the standard deviation of the probability, d i represents the distance from user i to the next user, is the total number of users in the path; S44, generate user behavior prediction and interaction analysis results: use the key interaction path calculation result, integrate the virtual scene model and probability data, and generate user behavior prediction and interaction analysis results. The digital media rendering method in virtual reality according to claim 6, wherein the obtaining step of the user behavior analysis result after adjusting the kernel parameters is specifically: S51, generate a preliminary user behavior analysis result: analyze the user behavior prediction and interaction analysis result, use a multi-kernel learning method to process data in a high-dimensional space, assign a preliminary kernel parameter to each data point, and generate a user behavior analysis result; S52, adjust the kernel parameters and generate an adjustment scheme: based on the user behavior analysis result, adjust the parameters of each kernel to ensure that each parameter matches the characteristics of the data, adjust the weights through linear algebra and optimization algorithms, and generate a parameter adjustment scheme; S53, apply adjustment scheme and generate final analysis results: apply the parameter adjustment scheme, recalculate user behavior, using the formula: The user behavior analysis result after the kernel parameter adjustment is generated by reflecting the adjusted user behavior through a weighted Gaussian function. where R(x) represents the response of user behavior, reflecting the prediction result obtained by the multi-kernel learning algorithm through the weighted summation of differentiated kernel functions, w j is the adjustment parameter of the jth kernel, representing the weight of the kernel in the total user behavior, x is the input variable, representing the data characteristics, μ j is the mean parameter of the jth kernel function, describing the center position of the data distribution corresponding to the kernel, σ j is the standard deviation parameter of the jth kernel function, indicating the dispersion degree of the data distribution.

7. A digital media rendering system in virtual reality, characterized by, The system comprises a method for rendering digital media in virtual reality according to any one of claims 1-7. The data acquisition module acquires HDR image data, depth information and user posture data in real time in the virtual reality environment through a high-definition camera and a sensor, adds timestamp information, aligns the multi-type sensor data in time by using a time synchronization technology, and generates a synchronous perception dataset. The scene optimization module adjusts the perspective and resolution of the virtual scene, updates the rendering effect, evaluates the potential impact of user interaction, reconfigures the virtual scene, and generates an optimized virtual scene based on the HDR image data, depth data and user posture data in the synchronous perception dataset. The behavior modeling module monitors the current HDR image data, depth data and user posture data, analyzes user interaction behavior, updates the user behavior model, integrates data into the user behavior model, and establishes a dynamic user model based on the optimized virtual scene. The prediction analysis module predicts future user interaction behavior, analyzes the impact of user behavior in the virtual scene, calculates a key interaction path, obtains user behavior prediction and interaction analysis results, processes the user behavior prediction and interaction analysis results, analyzes the HDR image data, depth data and user posture data by using a multi-core learning method, adjusts the kernel parameters, and generates a user behavior analysis result based on the dynamic user model.

8. The digital media rendering system in virtual reality of claim 8, wherein, The data acquisition module comprises a high-definition camera and a sensor, the high-definition camera is used to acquire HDR image data, the sensor is used to acquire depth data and user posture data, and the data acquisition module further comprises a timestamp generation unit configured to add a timestamp to each data point.

9. The digital media rendering system in virtual reality of claim 8, wherein, The prediction analysis module comprises a user behavior prediction unit, an interaction impact analysis unit and a key interaction path calculation unit, the user behavior prediction unit is configured to predict future user interaction behavior, the interaction impact analysis unit is configured to analyze the impact of user behavior in the virtual scene, and the key interaction path calculation unit is configured to calculate a key interaction path.

Images