Multimedia teaching method and teaching system

Abstract

The invention relates to the technical field of multimedia teaching, in particular to a multimedia teaching method and teaching system. The method comprises the following steps: obtaining multimedia courseware and marking the content structure of the multimedia courseware in a refined manner, and generating three-dimensional demonstration content structure refined data; secondly, quantizing frame-skipping jitter of the space interaction behavior, generating nonlinear frame-skipping strength regression data, and fitting rendering failure degrees based on the data; and finally, designing a self-adaptive rendering architecture according to the rendering failure fitting data, and sending the architecture to terminal equipment to execute the multimedia teaching method. Through optimization of rendering and interaction technologies, the teaching experience is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of multimedia teaching, and in particular to a multimedia teaching method and a teaching system. Background Art

[0002] When faced with complex teaching content, traditional text-based and image-based teaching methods often lack sufficient interactivity and immersion. Therefore, the introduction of multimedia technology has become a key breakthrough in teaching innovation. By combining multiple media formats, such as images, sound, video, and animation, students can simultaneously access information through different sensory channels, thereby enhancing their understanding and retention of knowledge. In particular, with the assistance of technologies such as 3D teaching and virtual reality (VR), teaching content can be presented in a more vivid and intuitive manner. Students can not only learn on a two-dimensional surface, but also achieve a more immersive learning experience through interactive methods such as 3D models and virtual experiments. Spatial interaction technology allows students to independently choose their own learning paths and explore the different dimensions of the teaching content, thereby increasing their learning initiative and engagement. This multimedia-based teaching method is particularly suitable for subjects that require spatial understanding and dynamic presentation, such as biology, physics, and engineering. However, a traditional multimedia teaching method suffers from inaccurate analysis of 3D courseware rendering failures, resulting in poor stability of adaptive 3D rendering of 3D courseware during multimedia teaching. Summary of the Invention

[0003] Based on this, it is necessary to provide a multimedia teaching method and teaching system to solve at least one of the above technical problems.

[0004] To achieve the above object, a multimedia teaching method is provided, the method comprising the following steps: Step S1: Acquire multimedia interactive courseware; perform content structure refinement annotation on the multimedia interactive courseware to obtain three-dimensional demonstration content structure refinement data; Step S2: quantifying spatial interactive behavior frame skipping jitter on the 3D presentation content structure refinement data to generate nonlinear frame skipping intensity regression data; performing polygon rendering failure degree fitting on the 3D presentation content structure refinement data based on the nonlinear frame skipping intensity regression data to obtain rendering failure degree fitting data; Step S3: designing an adaptive rendering architecture based on the rendering failure degree fitting data, thereby obtaining a 3D demonstration content adaptive rendering architecture; and sending the 3D demonstration content adaptive rendering architecture to a multimedia device terminal to execute the multimedia teaching method.

[0005] Preferably, step S1 includes the following steps: Step S11: Obtain multimedia interactive courseware; Step S12: Verify the content accuracy of the multimedia interactive courseware and generate multimedia interactive verification courseware; Step S13: extracting the three-dimensional demonstration content in the multimedia interactive verification courseware; Step S14: performing content structure refinement annotation on the three-dimensional demonstration content to obtain three-dimensional demonstration content structure refinement data.

[0006] Preferably, step S2 includes the following steps: Step S21: Obtain basic performance attributes of the multimedia device; Step S22: quantifying spatial interaction behavior frame skipping and jitter on the 3D presentation content structure refinement data according to the basic performance attributes of the multimedia device to obtain spatial interaction behavior frame skipping and jitter data; Step S23: performing nonlinear regression analysis on the frame skipping intensity of the spatial interaction behavior frame skipping jitter data to generate nonlinear frame skipping intensity regression data; Step S24: performing polygon rendering failure degree fitting on the three-dimensional presentation content structure refinement data based on the nonlinear frame skipping intensity regression data, thereby obtaining rendering failure degree fitting data.

[0007] Preferably, step S22 includes the following steps: Step S221: simulating and calculating a GPU performance load instability fluctuation curve in the basic performance attributes of the multimedia device; Step S222: Evaluate the dynamic GPU core computing power requirement range for content structure rotation / scaling / slicing in the 3D presentation content structure refinement data; Step S223: performing dynamic demand vertex transformation geometry increment analysis on the dynamic GPU core computing power demand interval of content structure rotation / scaling / slicing to obtain dynamic demand vertex transformation geometry increment data; Step S224: Calculating the computing power constraint ratio of the dynamically required vertex transformation geometry increment data according to the GPU performance load instability fluctuation curve to generate the GPU computing power constraint ratio; Step S225: quantify the frame skipping jitter of the spatial interaction behavior according to the GPU computing power constraint ratio to obtain the frame skipping jitter data of the spatial interaction behavior.

[0008] Preferably, step S23 includes the following steps: Step S231: performing disorder jitter intensity analysis on the frame skipping jitter data of the spatial interaction behavior to obtain the frame skipping disorder jitter intensity; Step S232: performing a logarithmic transformation of the jitter intensity time series variance of the spatial interaction behavior frame skipping jitter data according to the frame skipping disorder jitter intensity to obtain jitter intensity time series variance distribution characteristic data; Step S233: performing skewness distribution coupling on the jitter intensity time series variance distribution characteristic data to obtain jitter intensity variance skewness coupling data; Step S234: performing frame skipping intensity nonlinear regression analysis on the jitter intensity variance skewness coupling data to generate nonlinear frame skipping intensity regression data.

[0009] Preferably, step S24 includes the following steps: Step S241: performing polygon mesh topology analysis on the 3D presentation content structure refinement data to obtain content structure polygon mesh data; Step S242: Calculating the frame displacement mutation ratio on the nonlinear frame skipping intensity regression data, thereby obtaining the frame skipping intensity frame displacement mutation ratio; Step S243: performing structural polygon angular aliasing fuzzy analysis on the content structure polygon mesh data according to the frame skip strength and frame displacement mutation ratio to obtain structural angular aliasing fuzzy data; Step S244: performing interactive motion blur quantization based on the structural angular aliasing blur data and the frame skipping intensity / frame displacement mutation ratio to obtain interactive motion blur quantization data; Step S245: performing polygon rendering failure degree fitting based on the interactive operation motion blur quantization data, thereby obtaining rendering failure degree fitting data.

[0010] Preferably, step S244 includes the following steps: Based on the frame skipping intensity and frame displacement mutation ratio, a fracture interpolation process is performed between adjacent frames to obtain frame skipping fracture interpolation data between adjacent frames; The probability density of the displacement vector is calculated according to the frame skipping intensity and the frame displacement mutation ratio, thereby obtaining the probability density of the frame skipping displacement vector; Performing random process decomposition of the structural polygonal angular sawtooth shape on the content structure polygonal mesh data according to the probability density of the frame skip displacement vector and the frame skip fracture interpolation data between adjacent frames to obtain random process decomposition data of the sawtooth shape; The sharp sawtooth distribution data of the sawtooth morphology random process decomposition is identified by displacement mutation, and the sharp sawtooth distribution data is obtained; According to the sharp sawtooth distribution data, the fuzzy analysis of the structural polygonal edges and corners is performed to obtain the fuzzy data of the structural polygonal edges and corners.

[0011] Preferably, step S3 includes the following steps: Step S31: normalizing the rendering failure degree fitting data to obtain rendering failure degree fitting normalized data; Step S32: designing an adaptive rendering architecture based on the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data, thereby obtaining an adaptive rendering architecture for 3D presentation content; Step S33: Sending the three-dimensional demonstration content adaptive rendering architecture to the multimedia device terminal to execute the multimedia teaching method.

[0012] Preferably, step S32 includes the following steps: Step S321: performing adaptive frame rate synchronization processing according to the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data to obtain adaptive frame rate synchronization data; Step S322: performing view structure rendering resolution matching on the adaptive frame rate synchronization data according to the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data, thereby obtaining view structure rendering resolution matching data; Step S323: performing a reset frame buffer logic design based on the adaptive frame rate synchronization data and the view structure rendering resolution matching data, and generating a reset frame buffer design logic; Step S324: Adaptive rendering architecture design is performed through adaptive frame rate synchronization data, perspective structure rendering resolution matching data and resetting frame buffer design logic, thereby obtaining a three-dimensional presentation content adaptive rendering architecture.

[0013] Preferably, the present invention further provides a multimedia teaching system for executing the multimedia teaching method described above, the multimedia teaching system comprising: The content structure refinement module is used to obtain multimedia interactive courseware; perform content structure refinement annotation on the multimedia interactive courseware to obtain three-dimensional demonstration content structure refinement data; The rendering failure degree fitting module is used to quantify the frame skipping jitter of spatial interactive behavior on the 3D presentation content structure refinement data to generate nonlinear frame skipping intensity regression data; based on the nonlinear frame skipping intensity regression data, the polygon rendering failure degree is fitted on the 3D presentation content structure refinement data to obtain the rendering failure degree fitting data; The content adaptive rendering architecture module is used to design an adaptive rendering architecture based on the rendering failure degree fitting data, thereby obtaining a three-dimensional demonstration content adaptive rendering architecture; and sending the three-dimensional demonstration content adaptive rendering architecture to the multimedia device terminal to execute the multimedia teaching method.

[0014] The beneficial effect of the present invention is that by acquiring multimedia interactive courseware and performing structured and detailed annotation on its content, the three-dimensional presentation content in the courseware can be accurately extracted and organized. Through this detailed annotation, the various elements and interaction methods in the courseware will be more clearly defined, laying the foundation for subsequent analysis and processing. The detailed annotation not only helps to understand the content structure of the courseware, but also reveals the hierarchy of different interactive links and information presentation, allowing the courseware content to be accurately classified in spatial and temporal dimensions. This process ensures that subsequent spatial interaction behavior quantification and rendering failure analysis can be performed based on a clear data model, thereby improving the presentation effect of the courseware on multimedia devices. By quantifying the frame skipping and jitter of the spatial interaction behavior of the three-dimensional presentation content, it is possible to effectively identify and quantify instabilities such as freezes and delays that occur during dynamic presentations. This process further reveals the complex relationship between frame skipping and presentation content by generating nonlinear frame skipping intensity regression data. Based on this data, a fitting analysis of the failure degree of polygon rendering in the multimedia courseware can be performed to identify which links experience rendering failures. This quantitative analysis can accurately identify bottlenecks in the rendering process and provide a scientific basis for the design of an adaptive rendering architecture, thereby optimizing the presentation of courseware on different devices and improving fluency and stability. Based on the fitting data of the rendering failure degree, an adaptive rendering architecture is designed that adapts to different device performance and network environments. This architecture can dynamically adjust the rendering process according to the processing power of the device, ensuring that the rendering effect of the courseware is not reduced due to hardware limitations, thereby achieving more efficient multimedia presentation. This adaptive rendering architecture not only avoids rendering failures caused by device performance differences, but also ensures a smooth user experience in real-time interaction. By sending this optimized architecture to the multimedia device terminal, the multimedia teaching method can be smoothly run on various devices, improving the teaching quality and the student learning experience. Therefore, the present invention is an optimization process for a traditional multimedia teaching method, solving the problem of inaccurate analysis of 3D courseware rendering failure in a traditional multimedia teaching method, which results in poor stability of the adaptive rendering of 3D courseware during the teaching process of the multimedia device. It improves the accuracy of the analysis of 3D courseware rendering failure and improves the stability of the adaptive rendering of 3D courseware during the teaching process of the multimedia device. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flowchart of the steps of a multimedia teaching method is provided; Figure 2 for Figure 1 Detailed implementation steps of step S2 in FIG. Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0016] See also Figures 1 to 3 , a multimedia teaching method, the method comprising the following steps: Step S1: Acquire multimedia interactive courseware; perform content structure refinement annotation on the multimedia interactive courseware to obtain three-dimensional demonstration content structure refinement data; Step S2: quantifying spatial interactive behavior frame skipping jitter on the 3D presentation content structure refinement data to generate nonlinear frame skipping intensity regression data; performing polygon rendering failure degree fitting on the 3D presentation content structure refinement data based on the nonlinear frame skipping intensity regression data to obtain rendering failure degree fitting data; Step S3: designing an adaptive rendering architecture based on the rendering failure degree fitting data, thereby obtaining a 3D demonstration content adaptive rendering architecture; and sending the 3D demonstration content adaptive rendering architecture to a multimedia device terminal to execute the multimedia teaching method.

[0017] In the embodiment of the present invention, reference Figure 1 The above is a flowchart of the steps of a multimedia teaching method of the present invention. In this example, the multimedia teaching method includes the following steps: Step S1: Acquire multimedia interactive courseware; perform content structure refinement annotation on the multimedia interactive courseware to obtain three-dimensional demonstration content structure refinement data; In an embodiment of the present invention, a specified multimedia interactive courseware is downloaded from a teaching content management server through a local area network environment. The courseware contains three-dimensional model data, interactive script control files, material texture resources, and animation controller parameter files. The courseware is expanded step by step through a custom courseware unpacking program. The unpacked courseware file is scanned and traversed through the three-dimensional scene nodes using a content parsing module. The parent-child relationship of the three-dimensional objects is recursively identified based on the hierarchical tree structure. The grid topology structure, material reference path, animation binding information, and coordinate transformation matrix parameters are extracted for each three-dimensional model node. The spatial structure of the model space layout relationship is reconstructed using an octree space partitioning algorithm. At the same time, floating-point precision error statistics are performed on each grid vertex component. The floating-point error threshold is set to 0.0001. The vertex coordinate points whose errors exceed the threshold are subjected to coordinate correction processing. The corrected three-dimensional coordinate data is structured and stored in JSON format to generate a three-dimensional demonstration content structure refinement data file.

[0018] Step S2: quantifying spatial interactive behavior frame skipping jitter on the 3D presentation content structure refinement data to generate nonlinear frame skipping intensity regression data; performing polygon rendering failure degree fitting on the 3D presentation content structure refinement data based on the nonlinear frame skipping intensity regression data to obtain rendering failure degree fitting data; In the embodiment of the present invention, the basic attribute parameters of the hardware performance of the target multimedia device are obtained through the device performance detection module, including the GPU model, video memory capacity, core frequency and the number of parallel processing units, wherein the GPU model is NVIDIAGTX 1660, the video memory capacity is 6GB, the core frequency is 1530MHz, and the number of parallel CUDA cores is 1408. The theoretical maximum vertex processing throughput is calculated based on the GPU core frequency and the number of CUDA cores. The sampling frequency is set to 60Hz, and the spatial interactive behavior simulation is performed on the structural refinement data of the three-dimensional demonstration content. During the simulation process, the interactive input events are recorded frame by frame, including the rotation angle change rate, the scaling change rate and the depth change gradient of the cutting operation. The input behavior data is denoised using the Kalman filter algorithm, and then the time difference method is used to perform the spatial interactive behavior simulation. The inter-frame displacement velocity statistics of the 3D model vertex coordinate changes between consecutive frames were calculated. The maximum inter-frame displacement and displacement fluctuation variance were extracted. The frame skip detection threshold was set to 0.5 pixels / frame. The sample sequence was marked as a frame skip segment based on the detection threshold. The displacement change curve within the frame skip segment was nonlinearly fitted using polynomial regression analysis. A third-order polynomial fitting model was used, and the fitting accuracy was controlled to be greater than 0.95. Finally, nonlinear frame skip intensity regression data was obtained. The frame skip intensity distribution was optimized using the least squares method to ensure that the error of the regression curve for the entire interval data was less than 5%. Based on the nonlinear frame skipping intensity regression data, a polygon rendering failure degree fitting operation is performed on the three-dimensional demonstration content structure refinement data. First, the normal vector consistency of all polygonal faces is detected using the mesh topology analysis algorithm. The adjacent facet areas with a normal vector angle change rate greater than 15° are marked for corner partitioning. Combined with the peak segment in the nonlinear frame skipping intensity regression data, the normal vector mutation points at the corresponding vertex index position are regionally mapped. The Laplacian smoothing algorithm is used to perform geometric noise analysis on the local mesh of the model, and the regional average curvature change rate is extracted. The fuzzy logic discriminant function is used to perform correlation analysis between the average curvature change rate and the frame skipping intensity peak. The rendering failure degree evaluation interval is established, and the failure degree quantization coefficient is set between 0 and 1. The rendering failure degree fitting data is obtained through interval normalization.

[0019] Step S3: designing an adaptive rendering architecture based on the rendering failure degree fitting data, thereby obtaining a 3D demonstration content adaptive rendering architecture; and sending the 3D demonstration content adaptive rendering architecture to a multimedia device terminal to execute the multimedia teaching method.

[0020] In an embodiment of the present invention, linear normalization processing is performed on the rendering failure degree fitting data, all failure degree coefficients are scaled to the range of 0 to 1, and an interval linear mapping method is used for normalization operation. Then, the frame skipping intensity change curve in the nonlinear frame skipping intensity regression data is combined, and the frame rate synchronization analysis is performed based on the frame skipping intensity time series and the rendering failure degree series. The sliding window mean filtering method is used to calculate the frame rate fluctuation interval per second, and the dynamic frame rate adjustment threshold is determined. The threshold is set to 5fps, and the frame buffer is reconfigured based on the threshold. A double buffering mechanism is used for frame data management. The buffer size is automatically expanded according to the highest frame skipping intensity segment, and the buffer expansion upper limit is set to twice the original buffer size. At the same time, the perspective structure rendering resolution is matched according to the rendering failure degree fitting normalization data, and the setting The resolution is dynamically adjusted between 720p and 1080p. The target rendering resolution is adjusted in real time according to the current GPU load. A block rendering strategy is used to prioritize the 3D model area. The rendering quality of blocks with high failure rates is reduced first. The invisible patch areas are cropped in real time through the depth buffer culling method to reduce the pixel fill rate. Finally, an adaptive rendering architecture is designed based on the frame rate synchronization data, resolution matching data and buffer configuration logic. The architecture configuration file is output in XML format. The file records the frame rate synchronization parameters, resolution adjustment strategy and buffer management rules in detail. Finally, the adaptive rendering architecture of the 3D demonstration content is sent to the multimedia device terminal through the network interface protocol. The terminal executes the adaptive rendering process of the multimedia teaching content according to the received architecture file.

[0021] Step S1 includes the following steps: Step S11: Obtain multimedia interactive courseware; Step S12: Verify the content accuracy of the multimedia interactive courseware and generate multimedia interactive verification courseware; Step S13: extracting the three-dimensional demonstration content in the multimedia interactive verification courseware; Step S14: performing content structure refinement annotation on the three-dimensional demonstration content to obtain three-dimensional demonstration content structure refinement data.

[0022] In an embodiment of the present invention, a specified multimedia interactive courseware file is downloaded from a teaching resource management server via a local area network transmission protocol. The main content resources contained therein include three-dimensional model data, animation control data, material mapping resources, script behavior control files, and metadata description files. The download process uses the TCP protocol for data transmission, and adopts a fixed data packet size of 4KB for segmented buffer transmission. The CRC32 algorithm is used for data integrity verification during the transmission process. After the download is completed, the binary data of the courseware file is unpacked by the resource loading engine, and after unpacking, three-dimensional object resource files, audio and video resource files, script behavior files, and metadata files are generated in sequence. The content consistency check module is used to check the content accuracy of the unpacked multimedia interactive courseware. First, the topological structure consistency of the 3D model data is verified. The boundary connectivity of all triangular faces is checked using the topological scanning algorithm. The boundary isolation threshold is set to 2 during the inspection process, that is, the same vertex can only be shared by two adjacent faces. The vertex area exceeding the threshold is marked as a topological abnormal area. The key frame time series consistency check is performed on the animation control data part. The time series linear incremental judgment method is used to perform incremental analysis on all key frame timestamp data. If the timestamp is reversed, it is determined to be a time series abnormality. Usually, the path reference consistency of material map resources is checked, and the MD5 value of all material reference paths is calculated by the path reference hash verification method. If the material path does not exist or the MD5 is inconsistent, it is recorded as a material loss error. The syntax integrity of the script behavior file is checked, and the AST abstract syntax tree parsing method is used to generate a syntax tree for the script source file and traverse the syntax nodes. If there is a syntax tree parsing error, the script is marked as a syntax exception. Finally, a multimedia interactive verification courseware is generated based on various test results. The verification courseware file structure contains the resource files that have passed the verification and the error log record text of the abnormal resources. The 3D resource extraction module is used to accurately extract the 3D demonstration content in the multimedia interactive verification courseware, batch load all resource files belonging to this category, and use vertex buffer reading technology to decode and extract the vertex coordinates, normal vectors, UV coordinates and vertex color data of each 3D model from the binary buffer. Little-Endian byte order is used for memory mapping during the decoding process, and texture path mapping conversion is performed on the map resources. The texture reference paths in the material files are uniformly converted into local absolute paths. During the parsing process, the hierarchical structure of all Mesh resources is restored, and the scene node tree is reconstructed based on the parent-child hierarchical relationship. The output 3D demonstration content data includes model mesh data, texture binding information, hierarchical tree relationship table and animation skin binding weight matrix. After the extraction is completed, a 3D demonstration content resource package is generated.The structured annotation module is used to refine the annotation of the extracted 3D demonstration content. First, the AABB bounding box space partitioning algorithm is used to divide each model mesh into spatial regions. The block dimension threshold is set to 512 cubic voxel units. The normal vector average angle statistics are performed on the vertex set in each spatial block. The statistical interval is set to a segment interval of every 10°. The surface roughness classification is performed on the average angle interval. The roughness classification threshold is set to 30°. The part exceeding the threshold is marked as a high-roughness patch area. The polygon patch density statistics are performed on the high-roughness area. The statistical standard is per square unit. The number of triangles is set, and the density statistical threshold is set to 50 faces per square unit. The low-density area is spatially sectioned, and the sectioning area is hierarchically divided using the BSP tree partitioning method. At the same time, the spatial coordinate index range, normal vector change gradient, material map distribution characteristics, and vertex color distribution deviation are recorded for each sectioning level node. By parsing the model hierarchy LOD level, the grid level labels of different LOD levels are written into the annotation data file. Finally, all multi-dimensional annotation information such as spatial division, face density, normal vector characteristics, material distribution and LOD level are integrated to generate a 3D presentation content structure refinement data file.

[0023] Step S2 includes the following steps: Step S21: Obtain basic performance attributes of the multimedia device; Step S22: quantifying spatial interaction behavior frame skipping and jitter on the 3D presentation content structure refinement data according to the basic performance attributes of the multimedia device to obtain spatial interaction behavior frame skipping and jitter data; Step S23: performing nonlinear regression analysis on the frame skipping intensity of the spatial interaction behavior frame skipping jitter data to generate nonlinear frame skipping intensity regression data; Step S24: performing polygon rendering failure degree fitting on the three-dimensional presentation content structure refinement data based on the nonlinear frame skipping intensity regression data, thereby obtaining rendering failure degree fitting data.

[0024] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: Obtain basic performance attributes of the multimedia device; In the embodiment of the present invention, a device performance identification module is used to extract hardware attributes of the target multimedia device. First, the GPU model, video memory capacity, core frequency and number of rendering pipelines are obtained through the Direct3D device query interface in the DirectX API. The GPU model is NVIDIA GeForce GTX 1660, with a video memory capacity of 6GB, a core frequency of 1530MHz, and a number of rendering pipelines of 22 groups. The CPU main frequency, number of cores and system available memory capacity are obtained using the WMI service interface. The CPU model is Intel i5-10400, with a main frequency of 2.9GHz, a core number of 6 cores, and a system available memory capacity of 8GB. The floating-point computing capability of the device is obtained through the OpenCL environment query, and the calculated floating-point computing peak reaches 5.2 TFLOPS, using the PCI bus interface to query the bus bandwidth of the current device, the detection result is 8GB / s. All the above basic performance attributes are normalized, and the maximum and minimum normalization algorithm is used to normalize the units of each performance indicator. After normalization, attribute data such as GPU computing power, memory bandwidth, CPU computing power and bus transmission capacity are stored in the performance attribute vector array. The vector dimension is 4, and each dimension corresponds to a performance indicator.

[0025] Step S22: quantifying spatial interaction behavior frame skipping and jitter on the 3D presentation content structure refinement data according to the basic performance attributes of the multimedia device to obtain spatial interaction behavior frame skipping and jitter data; In the embodiment of the present invention, a spatial interactive behavior simulation module is used to perform frame skipping and jitter quantification processing on the three-dimensional demonstration content structure refinement data. First, the performance basic attribute vector obtained in step S21 is read, and the device rendering bottleneck factor is calculated according to the ratio of GPU computing power to memory bandwidth. The bottleneck factor threshold is set to 0.75. When the bottleneck factor is less than the threshold, the device is judged to be in a low-performance environment. Subsequently, the three-dimensional demonstration content structure refinement data is loaded, and spatial behavior simulation is performed on all three-dimensional mesh models. The segmented time step integration method is used to simulate three-dimensional interactive behaviors such as user rotation, scaling and cutting. The time step is set to 16.7ms, the simulation period is set to 10s, and a total of 600 frames are sampled. The continuous frames are analyzed by the time difference method. The vertex coordinate changes are statistically analyzed frame by frame, and the maximum vertex displacement, average displacement rate, and minimum distance change rate between frames are extracted. The displacement rate sequence is smoothed using the exponentially weighted sliding average method. The sliding window size is set to 5 frames, and a threshold is applied to the displacement change of each frame. The frame skipping threshold is set to 0.5 pixels / frame. Frame segments exceeding this threshold are marked as frame skipping events. The threshold is dynamically adjusted based on the GPU core occupancy and memory usage, with a maximum adjustment range of ±20%. Finally, the start and end frame numbers, maximum displacement amplitude, and average inter-frame speed of all frame skipping segments are quantized and encoded in a floating-point array format to generate frame skipping jitter data for spatial interaction behavior.

[0026] Step S23: performing nonlinear regression analysis on the frame skipping intensity of the spatial interaction behavior frame skipping jitter data to generate nonlinear frame skipping intensity regression data; In the embodiment of the present invention, the spatial interaction behavior frame skipping jitter data obtained in step S22 is sorted in time series to ensure that all data are arranged in ascending order by frame number. Subsequently, a local weighted regression method is used to perform nonlinear regression analysis on the frame skipping intensity change curve. The regression window size is set to 15 frames, and the kernel function is selected as a cubic spline kernel function. The frame skipping intensity time series data is locally fitted, and the frame skipping intensity change trend curve within each window is calculated. The least squares method is used to perform residual optimization on the fitting error. The fitting error threshold is set to 5%, and the window area where the residual exceeds the threshold is re-fitted and iterated. The maximum number of iterations does not exceed 10. At the same time, a polynomial fitting measurement is performed on the frame skipping intensity data sequence. The third-order polynomial regression model is selected to calculate the determination coefficient R² of the fitting curve. The R² value is controlled above 0.95. The regression curve is smoothed and corrected using the local deviation curve method. The correction goal is to reduce the oscillation amplitude of the regression curve in the frame skipping peak range. The smoothing coefficient is set to 0.8. Finally, the frame skipping intensity change trend, local fitting curve coefficient, fitting error distribution and smoothing correction parameters obtained by regression analysis are uniformly encoded into nonlinear frame skipping intensity regression data. The data is stored in a floating-point sequence, and each frame corresponds to a frame skipping intensity prediction value and a fitting error residual value.

[0027] Step S24: performing polygon rendering failure degree fitting on the three-dimensional presentation content structure refinement data based on the nonlinear frame skipping intensity regression data, thereby obtaining rendering failure degree fitting data.

[0028] In the embodiment of the present invention, the nonlinear frame skipping intensity regression data generated in step S23 and the three-dimensional demonstration content structure refinement data extracted in step S21 are first read, the topological structure of the three-dimensional mesh model is analyzed at the polygon level, and the depth-first traversal method is used to comprehensively extract all vertex indexes, face vertex sequences, and normal vector data in the model. The local curvature radius of each vertex is calculated using a curvature analysis algorithm, and the calculation radius threshold is set to 0.005m. The frame number interval corresponding to the peak segment in the frame skipping intensity regression data is selected, and the corresponding vertex displacement change is extracted frame by frame. The vertex area with a curvature radius change rate greater than 15% is marked as a failure risk, and the face normal vector angle change rate is used to compare the adjacent face angles. The segment is locally segmented, and the angle change rate threshold is set to 10°. The average curvature fluctuation variance statistics are performed on the segmented polygonal area, and the statistical window size is set to 20 patches. The areas with curvature fluctuation variance greater than the set threshold of 0.002 are divided into failure degree levels. The level division is based on the interval mapping of the normalized peak value of the frame skipping intensity regression data. The mapping interval is divided into five levels, from the lowest level 0 to the highest level 4. Each level corresponds to a different degree of rendering failure. The failure level of all patch areas is encoded, and an index array is used to correspond one-to-one with the floating-point failure degree factor, finally forming the rendering failure degree fitting data. The data structure includes polygon face index, failure level mark and corresponding failure intensity factor.

[0029] Step S22 includes the following steps: Step S221: simulating and calculating a GPU performance load instability fluctuation curve in the basic performance attributes of the multimedia device; Step S222: Evaluate the dynamic GPU core computing power requirement range for content structure rotation / scaling / slicing in the 3D presentation content structure refinement data; Step S223: performing dynamic demand vertex transformation geometry increment analysis on the dynamic GPU core computing power demand interval of content structure rotation / scaling / slicing to obtain dynamic demand vertex transformation geometry increment data; Step S224: Calculating the computing power constraint ratio of the dynamically required vertex transformation geometry increment data according to the GPU performance load instability fluctuation curve to generate the GPU computing power constraint ratio; Step S225: quantify the frame skipping jitter of the spatial interaction behavior according to the GPU computing power constraint ratio to obtain the frame skipping jitter data of the spatial interaction behavior.

[0030] In an embodiment of the present invention, an OpenCL parallel computing environment is first used to perform load simulation calculations on a multimedia device GPU. A high-density floating-point load simulation program based on pixel-by-pixel operations is constructed by calling a GPU instruction-level rendering instruction set. The program uses a 512×512 resolution virtual rendering viewport as the calculation basis. The single simulation run time is set to 300 seconds. During the simulation, a fixed primitive fill rate of 1 million triangles per second is set. A continuous multi-frame rendering task queue is used to periodically impact the GPU performance. The task queue length is set to 120 frames, and the number of shader executions called per frame is fixed to 500. At the same time, dynamically changing rendering state parameters are used to perform instantaneous load perturbations on the GPU core execution unit. The specific perturbation method is to increase or decrease the number of shader core threads by 10% every 2 seconds during the simulation cycle. At the same time, rendering pipeline parameters such as texture sampling rate, depth test state, and alpha blending factor are periodically modulated in the form of a sine wave. The modulation period is set to 5s, and the modulation amplitude is controlled within the range of ±20% of the basic parameters. During the simulation, the built-in OpenCL The performance counter interface samples the GPU's current core frequency, instruction execution rate, rendering pipeline fill rate, and memory bandwidth utilization in real time, with a sampling period set to 10ms. The fluctuation variance is calculated for all sampling points, and the degree of GPU load fluctuation is statistically analyzed using a weighted sliding variance algorithm with a sliding window length of 200 sampling points. A short-time Fourier transform is performed on the GPU core frequency sequence to extract the frequency domain fluctuation energy spectral density curve and determine the instability peak frequency interval. The instability determination threshold is set as the fundamental frequency fluctuation energy density peak greater than 25% of the total energy density. The GPU main frequency fluctuation amplitude within the instability interval is normalized to the maximum and minimum values ​​to obtain the relative GPU frequency fluctuation rate within a unit interval. Combining the load disturbance intensity and energy spectral density, the GPU performance response over the entire simulation cycle is integrated within the interval. Finally, a GPU performance load instability fluctuation curve is formed by jointly fitting multiple parameters such as time point, fluctuation amplitude, frequency domain energy density, and disturbance factor. The curve data is accurate to six decimal places and is represented in floating-point time series format, with each sampling point corresponding to a relative GPU main frequency fluctuation amplitude value.

[0031] Read the GPU performance load instability fluctuation curve and three-dimensional demonstration content structure refinement data generated in step S221, and divide the spatial interactive behavior parameters such as rotation, scaling and cutting in the three-dimensional content into intervals. First, use the segmented interval behavior modeling method to divide the rotation angular velocity into a low speed interval of 0 to 30° / s, a medium speed interval of 30 to 90° / s and a high speed interval greater than 90° / s. The scaling behavior is divided into a small scaling interval of 0 to 1.1 times, a medium scaling interval of 1.1 to 1.5 times and a large scaling interval greater than 1.5 times. The cutting behavior is divided into a shallow cutting interval of 0 to 0.2m / s, a medium cutting interval of 1.1 to 1.5 times and a large cutting interval greater than 1.5 times according to the cutting depth change rate. The middle-layer sectioning interval is 0.2 to 0.5 m / s and the deep-layer sectioning interval is greater than 0.5 m / s. For each behavior interval, the GPU core execution unit load simulation module is used to simulate the computing power demand. The simulation process uses the frame-by-frame vertex transformation counting method to count the total number of vertex transformations that the GPU needs to perform per second in each behavior interval. The linear relationship between the number of vertex transformations and the behavior speed parameter is modeled to construct a dynamic GPU core computing power demand interval data set. Each element in the data set contains the behavior type, behavior rate interval, corresponding vertex transformation load, and GPU cycle computing power consumption, in millions of floating-point operations per second. Based on the dynamic GPU core computing power requirement interval data obtained in step S222, the vertex transformation geometric incremental analysis method is used to calculate the vertex position change in different behavior intervals. First, the vertex index of each triangle face in the three-dimensional demonstration content structure refinement data is reordered, and the XYZ axis sequence is sorted according to the spatial position. The frame-by-frame displacement difference method is used to calculate the difference between the XYZ coordinate changes of the same vertex in consecutive frames, and the spatial displacement increment of each vertex between consecutive frames is calculated in units of m. The dynamic behavior rate interval mapping method is used to classify each vertex transformation increment into the corresponding behavior interval. At the same time, the total amount of all vertex increments in the behavior interval is accumulated to obtain the sum of the vertex transformation geometric increments of each behavior interval. The incremental data of all intervals are normalized, and the minimum and maximum normalization method is used to scale all incremental data to the range of 0 to 1. Finally, the dynamic demand vertex transformation geometric incremental data is generated. The data content includes behavior type, rate interval, number of vertices and corresponding spatial displacement increment factor.The dynamic demand vertex transformation geometry increment data generated in step S223 and the GPU performance load instability fluctuation curve generated in step S221 are read and processed using the computing power constraint ratio calculation algorithm. First, the high fluctuation section in the GPU instability fluctuation curve is interval-integrated to calculate the average fluctuation amplitude per second in MHz. The vertex transformation demand corresponding to each behavior interval in the dynamic demand vertex transformation geometry increment data is normalized and converted into millions of floating-point operations per second (MFLOPS). The computing power benchmark value is set to the GPU theoretical maximum floating-point performance peak value of 5.2. TFLOPS, for each behavior interval, calculate the ratio of the current computing power demand to the baseline value to form the computing power demand coefficient. Perform a point-to-point ratio operation on each computing power demand coefficient and the corresponding GPU instability fluctuation amplitude to calculate the GPU computing power constraint ratio, which is a dimensionless proportional coefficient. Perform linear interpolation on the GPU computing power constraint ratios of all behavior intervals with an interpolation step size of 0.01. Perform Savitzky-Golay filtering on the constraint ratio sequence with a filter window length of 11 points and a polynomial order of 3. After smoothing, the final GPU computing power constraint ratio sequence is formed. The output sequence length is consistent with the number of dynamic behavior intervals, and each interval corresponds to a GPU computing power constraint ratio.

[0032] The GPU computing power constraint ratio sequence generated in step S224 is read, and the spatial interactive behavior frame skipping jitter quantification processing is performed on the 3D presentation content structure refinement data. First, the behavior intervals are dynamically graded according to the constraint ratio sequence, and the frame skipping jitter weight factor is set to the square root of the constraint ratio. The inter-frame displacement fluctuation variance analysis is performed on the vertex transformation sequence of each behavior interval. The sliding window method is used to perform variance statistics on the vertex displacement within 10 frames. The sliding step size is set to 1 frame. The variance value in each sliding window is multiplied by the corresponding frame skipping jitter weight factor to obtain The weighted frame skipping jitter intensity is calculated and all behavior intervals are sorted by frame skipping intensity. The top 20% intensity intervals are selected for frame skipping marking. The frame skipping marking method uses a binary threshold judgment, and the threshold is set to the weighted intensity mean plus one standard deviation. The marked area is checked for temporal consistency, and isolated frame skipping points are filled using the temporal connectivity analysis method. The filling threshold is set to within 3 frames. The frame number interval, maximum vertex displacement, average displacement speed, and frame skipping intensity level are output for the final frame skipping marked area to form the frame skipping jitter data of spatial interaction behavior.

[0033] Step S23 includes the following steps: Step S231: performing disorder jitter intensity analysis on the frame skipping jitter data of the spatial interaction behavior to obtain the frame skipping disorder jitter intensity; Step S232: performing a logarithmic transformation of the jitter intensity time series variance of the spatial interaction behavior frame skipping jitter data according to the frame skipping disorder jitter intensity to obtain jitter intensity time series variance distribution characteristic data; Step S233: performing skewness distribution coupling on the jitter intensity time series variance distribution characteristic data to obtain jitter intensity variance skewness coupling data; Step S234: performing frame skipping intensity nonlinear regression analysis on the jitter intensity variance skewness coupling data to generate nonlinear frame skipping intensity regression data.

[0034] In the embodiment of the present invention, the frame sequence of the frame skipping jitter data of the spatial interaction behavior is first reordered to ensure that the frame numbers are correctly sorted in ascending order. Then, the sliding interval discrete statistics method is used to perform uneven amplitude difference statistics on the vertex displacement changes between consecutive frames. The sliding window size is set to 10 frames and the step size is set to 1 frame. The discrete variance method is used to calculate the variance of the displacement amplitude sequence in each window to obtain the inter-frame displacement fluctuation variance sequence. Then, the third-order central moment of the variance sequence is calculated, and the sample bias measurement index is used to analyze the disorder characteristics of the frame skipping intensity sequence. The skewness coefficient threshold is set to 0.3. Intervals exceeding this threshold are marked as disordered frame skipping segments. Subsequently, the range statistics of the maximum inter-frame displacement increment within the disordered frame skipping segments are performed. Regions with ranges greater than 1.5 times the average range are marked as high-intensity disordered frame skipping segments. Finally, the disordered frame skipping intensities of all intervals are normalized and the intensity data are scaled to the range of 0 to 1 using the maximum and minimum normalization method. This results in a disordered frame skipping jitter intensity sequence. Each item in the sequence corresponds to the normalized intensity value of a frame skipping segment, and the unit is the normalization amplitude factor. Read the frame skipping disordered jitter intensity sequence generated in step S231, perform frame sequence matching on the spatial interaction behavior frame skipping jitter data, ensure that each frame skipping segment intensity data corresponds one-to-one with the original frame number, then calculate the inter-frame variance of the disordered frame skipping intensity sequence, use the sliding window method to calculate the variance of the frame skipping intensity data within every 20 consecutive frames, set the sliding step to 1 frame, perform logarithmic transformation on the obtained time series variance sequence, and use the logarithmic transformation function with the natural logarithm as the base to perform logarithmic mapping on each variance value. Set the variance non-zero lower limit threshold to 0 before processing. To avoid mathematical overflow problems in logarithmic calculations, the time series variance logarithm sequence is normalized after transformation. The normalization method uses Z-score standardization. After standardization, the mean is 0 and the standard deviation is 1. The normalized sequence is divided into segmented intervals. The k-means clustering method is used to divide the entire sequence into three intervals. Each interval corresponds to a variance distribution characteristic category, including low volatility area, medium volatility area and high volatility area. Finally, the jitter intensity time series variance distribution characteristic data is generated. The data structure is a sequential floating-point array. Each frame corresponds to a standardized variance logarithm eigenvalue and a cluster classification label number. The jitter intensity time series variance distribution characteristic data generated in step S232 is read. First, the sliding window skewness analysis method is used to calculate the sliding interval skewness coefficient of the standardized variance logarithm feature sequence. The sliding window size is set to 25 frames and the step size is 1 frame. The third-order central moment formula is used to calculate the skewness of the eigenvalue distribution in each sliding window. The skewness threshold is set to 0.4. All window intervals with skewness coefficients exceeding the threshold are marked as high-skew segments. The variance distribution in the high-skew segments is subjected to range analysis. The standard of range greater than 1.3 times the mean range of the entire sequence is used as the high-volatility anomaly discrimination condition. The segment is defined as the skewness coupling interval. The Pearson correlation coefficient is calculated for the logarithmic sequence of inter-frame variance within the skewness coupling interval, and a point-to-point correlation analysis is performed with the frame skipping disorder jitter intensity sequence. Intervals with calculation results less than 0.6 are smoothed by a secondary sliding average with a smoothing window size of 5 frames. All skewness coupling interval data are uniformly encoded. Each interval data contains the skewness coefficient, range statistics, mean square error within the sliding window, and correlation coefficient. Finally, the jitter intensity variance skewness coupling data is obtained. The data structure is a segmented floating-point sequence. Each segment contains the start and end frame numbers, skewness characteristic values, and fluctuation intensity indicators. The jitter intensity variance skewness coupling data obtained in step S233 is read, and all segment data are reorganized into time series according to the start and end frame numbers. Then, the jitter intensity and time series are nonlinearly fitted using the polynomial regression method. The order of the regression model is set to third order, and the regression factors include the segment skewness coefficient, the extreme value and the mean square error. The third-order polynomial regression coefficient is calculated for each segment using the least squares method. The residual variance test is performed on the regression residual sequence, and the residual variance threshold is set to 0.02. The residual exceeding the threshold segment is tested using local weighting. The regression method is used for quadratic fitting correction. The weighting factor is set as the square value of the segment correlation coefficient. Segment splicing is performed on all segment fitting curves. The segment connection points are smoothed using the cubic spline interpolation method. The interpolation nodes are set at the midpoint of the start and end frame numbers of each segment. The mean square error statistics are performed on the fitting curve of the entire sequence. A mean square error of less than 0.015 is used as the fitting termination condition. Finally, nonlinear frame skipping intensity regression data is generated. The data consists of a polynomial coefficient sequence, a segment residual variance index, and a splicing smoothing parameter.

[0035] Step S24 includes the following steps: Step S241: performing polygon mesh topology analysis on the 3D presentation content structure refinement data to obtain content structure polygon mesh data; Step S242: Calculating the frame displacement mutation ratio on the nonlinear frame skipping intensity regression data, thereby obtaining the frame skipping intensity frame displacement mutation ratio; Step S243: performing structural polygon angular aliasing fuzzy analysis on the content structure polygon mesh data according to the frame skip strength and frame displacement mutation ratio to obtain structural angular aliasing fuzzy data; Step S244: performing interactive motion blur quantization based on the structural angular aliasing blur data and the frame skipping intensity / frame displacement mutation ratio to obtain interactive motion blur quantization data; Step S245: performing polygon rendering failure degree fitting based on the interactive operation motion blur quantization data, thereby obtaining rendering failure degree fitting data.

[0036] In an embodiment of the present invention, the three-dimensional presentation content structure refinement data is read, the vertex index of all triangular facets is extracted and reconstructed, the half-edge data structure is used to map the adjacency relationship of the polygon mesh, a vertex connection table is established based on the adjacency edge relationship of each vertex, a depth-first traversal algorithm is used to perform topological traversal on each connected sub-region, the direction of the mesh face normal vector, the vertex normal vector angle and the number of boundary edges are counted one by one, a mesh face connectivity matrix is ​​generated through a two-way edge table, the shared edge information between the facets is recorded, the normal vectors of all triangular facets are normalized, the normal vector accuracy is controlled to six significant digits after the decimal point, the number of facets per cubic meter is counted using a mesh density calculation method, the space division unit size is set to 0.05m, the number of facets, the average normal vector angle and the boundary density in all space units are recorded, and finally the content structure polygon mesh data is generated, which includes a vertex coordinate sequence, a facet index list, a normal vector sequence, an adjacency matrix and spatial density statistics. Read the nonlinear frame skip intensity regression data, perform first-order difference processing on the frame skip intensity prediction value of each frame in the continuous frame time series, use the forward difference method to calculate the change rate of the continuous frame skip intensity values, and obtain the inter-frame frame skip intensity increment sequence. Perform absolute value processing on the increment sequence, and the unit is normalized intensity increment ratio. Then use the threshold discrimination method to identify the mutation segment, and set the mutation identification threshold to 1.5 times the standard deviation of the frame skip intensity sequence. All frame numbers greater than the threshold are marked as mutations, and the ratio of the frame skip intensity increment of all mutation frames to the intensity value of the previous frame is calculated. Finally, the frame skip intensity frame displacement mutation ratio is generated. The data structure is a time series floating-point array, and each item corresponds to a frame number and its mutation ratio. Read the content structure polygon mesh data generated in step S241 and the frame skip strength frame displacement mutation ratio of step S242, use the mesh edge feature extraction method to measure the angle of all polygon boundary edges, use the normal vector angle calculation formula to calculate the angle between adjacent face normal vectors in radians, the unit is radian, and the calculation accuracy is controlled to six decimal places. All boundary areas with angles less than 0.1rad are marked as smooth areas, and boundary areas with angles greater than 0.7rad are marked as sharp areas. Then, all marked areas are marked as sharp areas. Weight mapping is performed on the mutation ratio of frame skipping intensity and frame displacement. The linear weight mapping method is used to map the mutation ratio to the fuzzy weight coefficient interval within the angular area, with a value range of 0 to 1. Angular fuzziness calculation is performed on all grid areas. The fuzziness is defined as the weighted angle variance within the local angular area. The sliding window size is set to 20 adjacent facets. All facet fuzzy sequences are normalized to finally generate structural angular aliasing fuzzy data. The data content includes facet number, fuzzy weight coefficient, angle variance value and fuzzy distinction category label.The structural angular jagged blurred data generated in step S243 and the frame skipping intensity frame displacement mutation ratio in step S242 are read, and the two types of data are mapped in a frame-level linkage manner using a multidimensional interval fusion algorithm. First, the frame skipping intensity frame displacement mutation ratio sequence is subjected to sliding interval weighted averaging processing, and the sliding window size is set to 15 frames to obtain a smooth mutation intensity sequence. Each patch fuzzy weight coefficient in the structural angular jagged blurred data is subjected to point-to-point multiplication mapping with the mutation intensity to generate the patch-level motion blur initial weight value, the unit of which is the normalized fuzzy influence factor, and spatial convolution processing is performed on the initial weight values ​​of all patches. The convolution kernel size is set to 5×5 patch adjacent units, and the Gaussian weighted kernel is used to perform weighted summation on the fuzzy weights of the surrounding patches to generate a spatial diffusion fuzzy weight matrix. The fuzzy weight matrix is ​​weighted accumulated in time series, and the time weight coefficient is exponentially decayed according to the time difference between the current frame number and the reference frame number. The exponential decay factor is set to 0.1. The final weighted accumulation result is Z-score normalized, and the value range after normalization is controlled between -3 and 3. Finally, the interactive motion blur quantization data is generated. The data structure contains the patch number, the time weighted fuzzy factor and the spatial diffusion fuzzy influence coefficient. The interactive operation motion blur quantization data generated in step S244 is read, and the time-weighted blur factor and the spatial diffusion blur influence coefficient of each patch are nonlinearly weighted and synthesized. The two factors are fused using the Sigmoid nonlinear weighting function, with the function parameters set to a slope coefficient of 1.5 and a threshold offset of 0.5. The fusion blur strength of all patches is fitted with the least squares surface method. The patch number and the fusion strength are nonlinearly fitted using the quadratic surface function. The convergence threshold of the fitting residual variance is set to 0.01. The high residual segments in the fitting process are corrected by quadratic fitting using the local spline interpolation method, and the interpolation node spacing is set to 5 patches. The global fitting surface is tested for continuity, and the smoothness of the fitting surface is corrected using a continuous second-order derivative constraint. The fusion strength values ​​of all patches on the final fitting surface are interval-normalized, with the value range limited to 0 to 1. Finally, the rendering failure degree fitting data is generated. The data is indexed by the patch number, and each patch corresponds to a normalized rendering failure fitting strength value.

[0037] Step S244 includes the following steps: Based on the frame skipping intensity and frame displacement mutation ratio, a fracture interpolation process is performed between adjacent frames to obtain frame skipping fracture interpolation data between adjacent frames; The probability density of the displacement vector is calculated according to the frame skipping intensity and the frame displacement mutation ratio, thereby obtaining the probability density of the frame skipping displacement vector; Performing random process decomposition of the structural polygonal angular sawtooth shape on the content structure polygonal mesh data according to the probability density of the frame skip displacement vector and the frame skip fracture interpolation data between adjacent frames to obtain random process decomposition data of the sawtooth shape; The sharp sawtooth distribution data of the sawtooth morphology random process decomposition is identified by displacement mutation, and the sharp sawtooth distribution data is obtained; According to the sharp sawtooth distribution data, the fuzzy analysis of the structural polygonal edges and corners is performed to obtain the fuzzy data of the structural polygonal edges and corners.

[0038] In an embodiment of the present invention, in an embodiment of performing intermittent interpolation processing between adjacent frames based on a frame skip intensity frame displacement mutation ratio, a frame skip intensity frame displacement mutation ratio sequence is first read, differential statistics are performed on the frame skip intensity differences between all consecutive frames, a forward difference method is used to calculate the rate of change of the frame skip intensity ratio of adjacent frames, a fracture discrimination threshold is set to 1.8 times the average rate of change of the entire sequence, and fracture points are marked for frame pairs with a value greater than the threshold. Subsequently, linear interpolation processing is performed on the spatial displacement vector difference between each pair of fractured frames. An inter-frame displacement vector linear interpolation algorithm is used to linearly interpolate the displacement vectors of the frames before and after the fracture, with the interpolation step size set to the length of a single frame. The interpolation result is subjected to Euclidean distance correction in a three-dimensional coordinate system to ensure the monotonicity of the interpolation point sequence on the spatial path. Finally, frame skip fracture interpolation data between adjacent frames is generated. The data structure includes a frame number, three-dimensional coordinates of the interpolated displacement vector, and a fracture segment marker label. In an embodiment of calculating the probability density of a displacement vector based on the mutation ratio of the frame skipping intensity to the frame displacement amount, the frame skipping intensity to the frame displacement amount mutation ratio sequence is first divided into intervals, and the mutation ratio interval length is set to 0.05. The histogram method is used to count the frequency of the mutation ratio in each interval, and the three-dimensional modulus length of the inter-frame displacement vector in each interval is calculated in units of m. Kernel density estimation is performed on all displacement modulus length sequences, and the Gaussian kernel function is used as the kernel function. The bandwidth parameter is set to 0.02. The estimation results are interval normalized to ensure that the integral of the probability density function is 1. Finally, the probability density sequence of the frame skipping displacement vector is obtained, and each interval corresponds to a mutation ratio interval, a displacement modulus length density value, and a normalization weight factor. In an embodiment of performing random process decomposition of structural polygonal angular sawtooth morphology on content structure polygonal mesh data based on the probability density of frame skip displacement vectors and frame skip break interpolation data between adjacent frames, local angular regions are first extracted from the content structure polygonal mesh data. All angular facet regions with angles greater than 0.6 radians are selected using a normal vector angle extraction algorithm. Vertex coordinate perturbation mapping is performed on these regions based on the frame skip break interpolation data between adjacent frames. The perturbation amount is proportional to the interpolation displacement vector of the corresponding frame, and the scale factor is set to 1.2. Time series difference analysis is performed on the perturbed vertex position sequence. Multidimensional white noise coupling is performed on the perturbation sequence of each vertex in combination with the probability density of the frame skip displacement vector. A spatial perturbation sequence is generated using the white noise superposition method, with the standard deviation set to 0.15 times the standard deviation of the original displacement modulus. Markov segment cutting is performed on all perturbed vertex sequences. A sliding window Markov chain state transition matrix method is used with a window length set to 10 frames. Finally, random process decomposition data of the sawtooth morphology is generated. The data content includes facet number, perturbation sequence, state transition probability matrix, and spatial noise distribution coefficient.In an embodiment of identifying sharp sawtooth distributions with displacement mutations in sawtooth-shaped random process decomposition data, first, local maximum gradients are identified for vertex disturbance sequences in all decomposed data, the instantaneous gradient change rate of the disturbance sequence is calculated using the second-order difference method, the gradient threshold is set to 1.5 times the average gradient change rate, and vertex sequence segments exceeding the threshold are marked with sharp jumps. Subsequently, the marked segments are calculated for time series range, with the range standard being greater than 1.3 times the local mean. A local extreme point detection algorithm is applied to all segments with excessive range to detect the locations of peak and valley points, and the ratio of the jump height between peaks and valleys to the number of continuous frames is calculated. The lower limit of the number of continuous frames is set to 3 frames, and the lower limit of the jump height is set to 0.02m. Sharp sawtooth segment encoding is performed on all segments that meet the conditions, and finally sharp sawtooth distribution data is generated. The data content includes vertex number, gradient change rate, range value, peak and valley position, and segment duration length. In an embodiment of performing fuzzy analysis of structural polygonal corner sawtooth based on sharp sawtooth distribution data, first, the local adjacency relationship of the corners is extracted for all facets marked as sharp sawtooth segments, and the half-edge topology table is used to perform spatial diffusion area division on the adjacent facets of the target area, and the division depth is set to the second-order adjacent layer. The variance of all vertex positions in each adjacent facet area is calculated to obtain the spatial divergence of the regional vertex, and the divergence calculation unit is mm². Then, linear weighting is performed based on the divergence value and the extreme difference value in the sharp sawtooth distribution data of the corresponding segment, and the weight coefficient is set to the extreme difference value normalization ratio. The weighted result is spatially smoothed, and a bidirectional weighted moving average algorithm is used. The sliding window size is set to 5 facets. All smoothed corner fuzzy data are interval normalized, and the normalized target interval is 0 to 1. Finally, structural corner sawtooth fuzzy data is generated, and the data includes the facet number, spatial divergence value, weighted fuzzy factor and adjacent layer diffusion coefficient.

[0039] Step S3 includes the following steps: Step S31: normalizing the rendering failure degree fitting data to obtain rendering failure degree fitting normalized data; Step S32: designing an adaptive rendering architecture based on the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data, thereby obtaining an adaptive rendering architecture for 3D presentation content; Step S33: Sending the three-dimensional demonstration content adaptive rendering architecture to the multimedia device terminal to execute the multimedia teaching method.

[0040] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: normalizing the rendering failure degree fitting data to obtain rendering failure degree fitting normalized data; In the embodiment of the present invention, for all the faces in the rendering failure degree fitting data, the rendering failure values ​​are normalized using the range normalization method, the normalization interval is set to 0 to 1, the maximum failure value and the minimum failure value in the rendering failure degree fitting data are extracted, and marked as the maximum reference value and the minimum reference value respectively, the failure degree of each face is linearly scaled, the calculation process uses single-precision floating-point precision, and the range less than 0 is used in the processing process. The sample points are assigned a fixed normalization value of 0.5 to ensure numerical stability, and the normalized results are checked for errors. The maximum and minimum normalization checks are performed on the normalized data sequence to ensure that all data points are strictly distributed in the closed interval of 0 to 1. The normalized results are numbered and mapped, and the patch numbers and the corresponding normalized failure degree values ​​are recorded to finally form the rendering failure degree fitting normalized data.

[0041] Step S32: designing an adaptive rendering architecture based on the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data, thereby obtaining an adaptive rendering architecture for 3D presentation content; In an embodiment of the present invention, in an embodiment of an adaptive rendering architecture design based on nonlinear frame skipping intensity regression data and rendering failure degree fitting normalization data, first, the nonlinear frame skipping intensity regression data is timestamp aligned, and the frame skipping intensity regression sequences at different time scales are uniformly time-base adjusted using a frame-by-frame interpolation method, and the interpolation step is set to the time length of a single frame. Subsequently, the time-aligned frame skipping intensity regression data and the rendering failure degree fitting normalization data are jointly interval classified, and three interval thresholds are set according to the frame skipping intensity value and the failure normalization value, respectively. The frame skipping intensity interval division thresholds are 0.3 and 0.7, and the failure degree interval division thresholds are 0.4 and 0.8, which are combined to form nine types of joint priority intervals. Rendering task priority encoding is performed on facet sets in different joint intervals, and the encoding levels range from P1 to P9, where P1 represents the highest priority and P9 represents the lowest priority. Full-resolution real-time rendering mode is adopted for faces in the high-priority interval, and the high-speed rendering channel of the GPU is used to allocate the maximum texture. Buffer and minimum rendering delay, medium resolution rendering is used for medium priority interval faces, medium level texture filtering and anti-aliasing processing is applied, a degraded rendering strategy is used for low priority interval faces, dynamic LOD control algorithm is enabled, GPU load is reduced by reducing the number of polygons and texture resolution, the frame buffer size of the overall rendering architecture is dynamically configured, the buffer frame range is set between 4 frames and 10 frames, the buffer depth is dynamically adjusted according to the current GPU utilization rate, the rendering instruction stream is reordered, and the P1 interval face rendering instructions are processed first according to the priority encoding, and then the instructions of the P2 to P9 intervals are processed, the GPU bandwidth allocation is adjusted in real time during each frame rendering process, and the bandwidth adjustment frequency is set to 60 times per second, the rendering time slice occupancy ratio of each type of interval is monitored in real time, and the time slice allocation ratio is 50%, 30% and 20% respectively. The final output is a 3D presentation content adaptive rendering architecture including a face priority table, a rendering channel allocation table and a buffer frame depth table.

[0042] Step S33: Sending the three-dimensional demonstration content adaptive rendering architecture to the multimedia device terminal to execute the multimedia teaching method.

[0043] In an embodiment of the present invention, a three-dimensional demonstration content adaptive rendering architecture is sent to a multimedia device terminal to execute a multimedia teaching method. First, a high-speed data transmission channel is established between the multimedia device terminal and the rendering control terminal, the TCP protocol is used for data encapsulation, the data transmission buffer size is set to 256KB, all parameter data in the three-dimensional demonstration content adaptive rendering architecture are serialized, the rendering priority table, the bandwidth allocation table and the buffer frame depth table are packaged item by item in a binary stream format, a CRC32 check code is calculated for the packaged data stream, and the CRC32 check code is attached to the end of the data stream to ensure the integrity of the transmitted data. A reception confirmation mechanism is set during the data transmission process, and each transmission Input a 4KB data block and wait for the device terminal to return an ACK confirmation signal. Retransmit the data block that has not been confirmed due to timeout up to 3 times. After the transmission is completed, the data stream is unpacked and buffered on the multimedia device terminal side. The unpacked rendering architecture data is mapped to the GPU rendering scheduling unit using a memory mapping buffer. The size of the memory mapping area is dynamically allocated based on the total number of bytes of the rendering task, and the minimum allocation unit is set to 64KB. During the unpacking process, the structure of each table is verified to ensure that the content of the data table is intact. The rendering instruction sequence in the buffer is loaded at high speed by DMA. After completion, the GPU rendering task initialization instruction is triggered to start the adaptive rendering process of the three-dimensional demonstration content under the multimedia teaching method.

[0044] Step S32 includes the following steps: Step S321: performing adaptive frame rate synchronization processing according to the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data to obtain adaptive frame rate synchronization data; Step S322: performing view structure rendering resolution matching on the adaptive frame rate synchronization data according to the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data, thereby obtaining view structure rendering resolution matching data; Step S323: performing a reset frame buffer logic design based on the adaptive frame rate synchronization data and the view structure rendering resolution matching data, and generating a reset frame buffer design logic; Step S324: Adaptive rendering architecture design is performed through adaptive frame rate synchronization data, perspective structure rendering resolution matching data and resetting frame buffer design logic, thereby obtaining a three-dimensional presentation content adaptive rendering architecture.

[0045] In an embodiment of the present invention, adaptive frame rate synchronization is performed based on nonlinear frame skipping intensity regression data and rendering failure degree fitting normalization data. First, the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data are aligned at the frame level. Interpolation correction is performed on data sequences of different time scales. The interpolation algorithm uses linear interpolation, and the interpolation interval is set to 16.67ms, corresponding to a base frequency of 60 frames per second. After the time alignment is completed, a weighted average of the frame skipping intensity and the rendering failure degree is performed on each frame of data, with the weight coefficient set to 0. 6 and the failure degree weight 0.4, and perform sliding window filtering on the obtained weighted average sequence. The filter window length is set to 5 frames, the sliding step is set to 1 frame, and the filtered sequence is divided into frame rate adjustment intervals. The interval thresholds are set to 0.3 and 0.7. When the weighted average value is less than 0.3, the target frame rate is set to 30fps. When the weighted average value is greater than 0.7, the target frame rate is set to 60fps. The remaining intervals are set to 45fps. For each frame, the frame rendering cycle duration and rendering instruction scheduling delay are calculated according to the current target frame rate, and finally the adaptive frame rate synchronization data is formed.

[0046] In an embodiment of performing perspective structure rendering resolution matching on adaptive frame rate synchronization data based on nonlinear frame skipping intensity regression data and rendering failure degree fitting normalized data, first, the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalized data are jointly spatially characterized. During the division process, the visible area is spatially gridded according to the spatial coordinate system of the scene three-dimensional model. The grid resolution is set to 64×64 pixel area units. The frame skipping intensity weighted value and the failure degree weighted value are calculated for each grid unit. The weighted average method is used to calculate the joint complexity score of each grid unit. The joint complexity interval division threshold is set to 0.4 and 0.8, corresponding to low complexity areas, medium complexity areas and high complexity areas respectively. Different rendering resolution levels are set for different complexity areas. Low complexity areas are rendered at 0.5 times the original resolution, medium complexity areas are rendered at 0.75 times, and high complexity areas are rendered at full resolution. The spatial continuity of the rendering resolution level data is optimized, and the boundaries of blocks of different levels are graded and smoothed using a 3×3 grid neighborhood sliding window method to avoid visual discontinuity caused by sudden changes in rendering resolution. The smoothed resolution level distribution data is time-aligned with the adaptive frame rate synchronization data to generate perspective structure rendering resolution matching data. In an embodiment of the reset frame buffer logic design based on the adaptive frame rate synchronization data and the perspective structure rendering resolution matching data, the target frame rate sequence in the adaptive frame rate synchronization data is first divided into interval segments, and an independent buffer strategy is designed for the interval where the continuous frame rate is maintained and the length is greater than 10 frames. The buffer depth is set to 8 frames for the target frame rate interval of 30fps, the buffer depth is set to 6 frames for the 45fps interval, and the buffer depth is set to 4 frames for the 60fps interval. The buffer size is calculated according to the rendering resolution and color depth of each frame. Assuming that the single frame resolution is 1920×1080 pixels and the color depth is 32 bits, the buffer storage requirement is approximately 7.9MB per frame. According to the perspective structure The rendering resolution matching data sets differentiated buffer partitions for areas of different complexity. The storage space allocation for the buffer in the low-resolution area is reduced according to the reduction ratio. The buffer space reserved area is set to 10% of the total buffer capacity for dynamic overflow control. The read and write order of the entire buffer is logically optimized, and a ring buffer design is adopted. The frame read and write positions are managed by a ring pointer. The pointer step is controlled by the target frame rate. After each frame is rendered, the pointer moves forward one buffer unit. The maximum read and write delay is controlled within 16ms. An integrity check is performed on the buffer logic structure to ensure the buffer consistency of each frame rendering task at the target frame rate and target resolution, and finally generate the reset frame buffer design logic.In an embodiment of an adaptive rendering architecture design using adaptive frame rate synchronization data, perspective structure rendering resolution matching data, and frame buffer reset design logic, first, parameters of the three types of data are fused, and a one-to-one mapping relationship is established for different data types using a multidimensional array indexing method. The index table includes frame sequence number, region number, target frame rate, target resolution level, and buffer allocation unit number. Frame-by-frame task allocation calculation is performed on the mapping table, and a time slice allocation method is used to control the rendering time of different priority regions. The total time slice length of each frame is set according to the target frame rate. A maximum time slice length threshold is set for the rendering task of each region. The time slice length threshold is set to 2ms for low-complexity blocks, 4ms for medium-complexity blocks, and 6ms for high-complexity blocks. The rendering task queue is task-level pipeline rearranged, and a processing order of high complexity first and low complexity later is executed. Buffer management instructions and resolution reset instructions are added to the rendering instruction stream. A logical consistency check is performed on the output data structure of the rendering architecture, and the final three-dimensional presentation content adaptive rendering architecture is output.

[0047] The present invention also provides a multimedia teaching system for executing the multimedia teaching method described above, the multimedia teaching system comprising: The content structure refinement module is used to obtain multimedia interactive courseware; perform content structure refinement annotation on the multimedia interactive courseware to obtain three-dimensional demonstration content structure refinement data; The rendering failure degree fitting module is used to quantify the frame skipping jitter of spatial interactive behavior on the 3D presentation content structure refinement data to generate nonlinear frame skipping intensity regression data; based on the nonlinear frame skipping intensity regression data, the polygon rendering failure degree is fitted on the 3D presentation content structure refinement data to obtain the rendering failure degree fitting data; The content adaptive rendering architecture module is used to design an adaptive rendering architecture based on the rendering failure degree fitting data, thereby obtaining a three-dimensional demonstration content adaptive rendering architecture; and sending the three-dimensional demonstration content adaptive rendering architecture to the multimedia device terminal to execute the multimedia teaching method.

[0048] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A multimedia teaching method, characterized in that: The following steps are involved: Step S1: Acquire multimedia interactive courseware; perform content structure refinement annotation on the multimedia interactive courseware to obtain three-dimensional demonstration content structure refinement data; Step S2: quantifying the spatial interaction behavior frame skipping jitter on the 3D presentation content structure refinement data to generate nonlinear frame skipping intensity regression data; Based on the nonlinear frame skipping intensity regression data, polygon rendering failure degree fitting is performed on the 3D presentation content structure refinement data, thereby obtaining rendering failure degree fitting data; Step S3: designing an adaptive rendering architecture based on the rendering failure degree fitting data, thereby obtaining an adaptive rendering architecture for 3D presentation content; The adaptive rendering architecture of the three-dimensional demonstration content is sent to the multimedia device terminal to execute the multimedia teaching method.

2. The multimedia teaching method according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Obtain multimedia interactive courseware; Step S12: Verify the content accuracy of the multimedia interactive courseware and generate multimedia interactive verification courseware; Step S13: extracting the three-dimensional demonstration content in the multimedia interactive verification courseware; Step S14: performing content structure refinement annotation on the three-dimensional demonstration content to obtain three-dimensional demonstration content structure refinement data.

3. The multimedia teaching method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: Obtain basic performance attributes of the multimedia device; Step S22: quantifying spatial interaction behavior frame skipping and jitter on the 3D presentation content structure refinement data according to the basic performance attributes of the multimedia device to obtain spatial interaction behavior frame skipping and jitter data; Step S23: performing nonlinear regression analysis on the frame skipping intensity of the spatial interaction behavior frame skipping jitter data to generate nonlinear frame skipping intensity regression data; Step S24: performing polygon rendering failure degree fitting on the three-dimensional presentation content structure refinement data based on the nonlinear frame skipping intensity regression data, thereby obtaining rendering failure degree fitting data.

4. The multimedia teaching method according to claim 3, characterized in that: Step S22 includes the following steps: Step S221: simulating and calculating a GPU performance load instability fluctuation curve in the basic performance attributes of the multimedia device; Step S222: Evaluate the dynamic GPU core computing power requirement range for content structure rotation / scaling / slicing in the 3D presentation content structure refinement data; Step S223: performing dynamic demand vertex transformation geometry increment analysis on the dynamic GPU core computing power demand interval of content structure rotation / scaling / slicing to obtain dynamic demand vertex transformation geometry increment data; Step S224: Calculating the computing power constraint ratio of the dynamically required vertex transformation geometry increment data according to the GPU performance load instability fluctuation curve to generate the GPU computing power constraint ratio; Step S225: quantify the frame skipping jitter of the spatial interaction behavior according to the GPU computing power constraint ratio to obtain the frame skipping jitter data of the spatial interaction behavior.

5. The multimedia teaching method according to claim 3, characterized in that: Step S23 includes the following steps: Step S231: performing disorder jitter intensity analysis on the frame skipping jitter data of the spatial interaction behavior to obtain the frame skipping disorder jitter intensity; Step S232: performing a logarithmic transformation of the jitter intensity time series variance of the spatial interaction behavior frame skipping jitter data according to the frame skipping disorder jitter intensity to obtain jitter intensity time series variance distribution characteristic data; Step S233: performing skewness distribution coupling on the jitter intensity time series variance distribution characteristic data to obtain jitter intensity variance skewness coupling data; Step S234: performing frame skipping intensity nonlinear regression analysis on the jitter intensity variance skewness coupling data to generate nonlinear frame skipping intensity regression data.

6. The multimedia teaching method according to claim 3, characterized in that: Step S24 includes the following steps: Step S241: performing polygon mesh topology analysis on the 3D presentation content structure refinement data to obtain content structure polygon mesh data; Step S242: Calculating the frame displacement mutation ratio on the nonlinear frame skipping intensity regression data, thereby obtaining the frame skipping intensity frame displacement mutation ratio; Step S243: performing structural polygon angular aliasing fuzzy analysis on the content structure polygon mesh data according to the frame skip strength and frame displacement mutation ratio to obtain structural angular aliasing fuzzy data; Step S244: performing interactive motion blur quantization based on the structural angular aliasing blur data and the frame skipping intensity / frame displacement mutation ratio to obtain interactive motion blur quantization data; Step S245: performing polygon rendering failure degree fitting based on the interactive operation motion blur quantization data, thereby obtaining rendering failure degree fitting data.

7. The multimedia teaching method according to claim 6, characterized in that: Step S244 includes the following steps: Based on the frame skipping intensity and frame displacement mutation ratio, a fracture interpolation process is performed between adjacent frames to obtain frame skipping fracture interpolation data between adjacent frames; The probability density of the displacement vector is calculated according to the frame skipping intensity and the frame displacement mutation ratio, thereby obtaining the probability density of the frame skipping displacement vector; Performing random process decomposition of the structural polygonal angular sawtooth shape on the content structure polygonal mesh data according to the probability density of the frame skip displacement vector and the frame skip fracture interpolation data between adjacent frames to obtain random process decomposition data of the sawtooth shape; The sharp sawtooth distribution data of the sawtooth morphology random process decomposition is identified by displacement mutation, and the sharp sawtooth distribution data is obtained; According to the sharp sawtooth distribution data, the fuzzy analysis of the structural polygonal edges and corners is performed to obtain the fuzzy data of the structural polygonal edges and corners.

8. The multimedia teaching method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: normalizing the rendering failure degree fitting data to obtain rendering failure degree fitting normalized data; Step S32: designing an adaptive rendering architecture based on the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data, thereby obtaining an adaptive rendering architecture for 3D presentation content; Step S33: Sending the three-dimensional demonstration content adaptive rendering architecture to the multimedia device terminal to execute the multimedia teaching method.

9. The multimedia teaching method according to claim 7, characterized in that: Step S32 includes the following steps: Step S321: performing adaptive frame rate synchronization processing according to the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data to obtain adaptive frame rate synchronization data; Step S322: performing view structure rendering resolution matching on the adaptive frame rate synchronization data according to the nonlinear frame skipping intensity regression data and the rendering failure degree fitting normalization data, thereby obtaining view structure rendering resolution matching data; Step S323: performing a reset frame buffer logic design based on the adaptive frame rate synchronization data and the view structure rendering resolution matching data, and generating a reset frame buffer design logic; Step S324: Adaptive rendering architecture design is performed through adaptive frame rate synchronization data, perspective structure rendering resolution matching data and resetting frame buffer design logic, thereby obtaining a three-dimensional presentation content adaptive rendering architecture.

10. A multimedia teaching system, characterized in that: For executing the multimedia teaching method according to claim 1, the multimedia teaching system comprises: The content structure refinement module is used to obtain multimedia interactive courseware; perform content structure refinement annotation on the multimedia interactive courseware to obtain three-dimensional demonstration content structure refinement data; The rendering failure degree fitting module is used to quantify the frame skipping jitter of spatial interactive behavior on the 3D presentation content structure refinement data to generate nonlinear frame skipping intensity regression data; based on the nonlinear frame skipping intensity regression data, the polygon rendering failure degree is fitted on the 3D presentation content structure refinement data to obtain the rendering failure degree fitting data; The content adaptive rendering architecture module is used to design an adaptive rendering architecture based on the rendering failure degree fitting data, thereby obtaining a three-dimensional demonstration content adaptive rendering architecture; and sending the three-dimensional demonstration content adaptive rendering architecture to the multimedia device terminal to execute the multimedia teaching method.

Images