48 results about "Rendering algorithms" patented technology

The invention discloses a lecturer video automatic generation method based on AI speech synthesis and animation driving. The method comprises the following steps: performing structured analysis on a PPT or a text script through an improved interior point method and an incremental shortest path algorithm; performing semantic grouping by applying a full-dynamic parallel single-link clustering algorithm and generating an enhanced script with an expressive mark; a CosyVoice technology is combined with a low-rank approximation method to generate a high-quality voice data stream; establishing a mapping relation between contents and action expressions through semantic analysis, and generating a complete action expression instruction set; and driving the digital human model by using the msueTalk technology, and generating a final lecturer teaching video through a parallel rendering algorithm. According to the invention, the method achieves the efficient and automatic generation of the education video, remarkably improves the content production efficiency, reduces the production cost, and guarantees the specialty and expressive force of the teaching video.

The invention relates to the technical field of industrial detection, and discloses a lithium niobate metasurface defect intelligent identification method based on reactive ion beam etching, which systematically solves the problem of scarcity of experimental data through physical driving simulation and a physics-based rendering algorithm, and provides large-scale field specific training data for a deep learning model. The bidirectional coupling simulation model generates defect morphology prediction data conforming to an etching dynamics law based on real process parameters and material parameters, the prediction data is converted into a synthetic image with real microscopic image statistical characteristics by a physical rendering algorithm, and the quality and consistency of the synthetic data are ensured by an automatic labeling and statistical verification process. According to the method, the number of samples of the ternary association database is multiplied, and extreme process conditions and rare defect modes which are difficult to obtain through experimental collection are covered.

The invention relates to the technical field of building geometric reconstruction and crack detection, in particular to a three-dimensional reconstruction and damage detection method for a building in an intelligent driving environment, and the method comprises the steps: collecting a building video in a road environment during intelligent driving, and reconstructing a sparse point cloud of the building through an SfM algorithm; inputting each frame of time sequence image into a pre-trained semantic segmentation model to generate a crack highlight image; according to the sparse point cloud, initializing a corresponding parameterized 3D Gaussian ellipsoid, and optimizing parameters of the 3D Gaussian ellipsoid; mapping the 3D Gaussian ellipsoid to a two-dimensional image plane through a camera model, and generating a rendered image through a differentiable rendering algorithm; and comparing the rendered image with the crack highlight image, iteratively calculating the overall loss, and transmitting the overall loss through a back propagation gradient until a preset convergence condition is reached, thereby obtaining a final 3D model which simultaneously expresses the overall structure of the building and crack details. According to the invention, accurate environment data can be provided for the automatic driving system.

The invention discloses a dynamic lightweight method for a digital BIM (Building Information Modeling) twin model of a combined air conditioning unit, and belongs to the technical field of BIM and digital twin. The method comprises the following steps: firstly, carrying out functional component semantic segmentation on a BIM model to generate a submodule set with labels; key performance features are extracted based on sensor data and CFD simulation results to generate a feature weight matrix; constructing a dynamic lightweight rule base to define a multi-level LOD template and a scene adaptation rule; fusing the feature weight and the real-time scene parameter through a fuzzy decision tree algorithm to output an LOD matching priority and a lightweight instruction set; and finally, extracting a target LOD data block according to an instruction set, and outputting a lightweight model by adopting a GPU accelerated rendering algorithm. The method solves the problems that the high-precision BIM model of the combined air conditioning unit is unsmooth in operation at a low-computing-power terminal and low in data transmission efficiency, and is suitable for a real-time operation and maintenance scene of a heating ventilation air conditioning system.

The invention discloses a hoisting equipment distribution visualization and monitoring method, and relates to the technical field of industrial hoisting safety, and the method comprises the following steps: generating a synchronously associated multi-source fusion data set; constructing a three-dimensional basic model of the hoisting operation area; dividing the three-dimensional dynamic model into a base layer, an equipment layer, a key component layer and a data layer by adopting a self-adaptive hierarchical rendering algorithm, adjusting the rendering precision of each layer according to a priority coefficient, and generating an equipment distribution visual interface; identifying an abnormal state of the equipment and positioning a fault position, and triggering a corresponding grading early warning signal according to an abnormal grade; and based on the abnormity type and the fault position corresponding to the early warning signal, a targeted disposal suggestion scheme is generated. The real-time monitoring, fault early warning and three-dimensional dynamic visualization of the hoisting equipment are realized by fusing space-time alignment, multi-source data synchronization, a digital twinning technology and a self-adaptive rendering algorithm, and the equipment safety and the operation efficiency are remarkably improved.

The invention relates to a rapid rendering method for an urban Beidou cubic grid. The method comprises the following steps: firstly, acquiring urban low-altitude geographic space data; constructing a multi-level grid system according to conflict detection precision and response requirements by using a Beidou cubic grid subdivision technology, and associating codes and airspace attributes with grid units; unloading the geometric generation task to a GPU for parallel calculation, and generating rendering adaptive grid geometric data; then developing a three-dimensional rapid rendering algorithm to realize real-time rendering of any region and precision grid; subsequently establishing a rendering result and airspace management cooperation mechanism to assist route planning, management and control and zoning; and finally, an algorithm optimization module is linked. The invention aims to solve the problems that the existing ground object 3D rendering adaptive city low-altitude field is insufficient, the conflict detection precision is not matched, the real-time performance is poor and the existing ground object 3D rendering adaptive city low-altitude field is disjointed with a low-altitude management service.

The invention discloses a penetrating supervision method and system based on market subject digital twinning, belongs to the technical field of market supervision and digital twinning, and aims to solve the technical problem of how to realize full-flow penetrating supervision of multi-source data and improve supervision accuracy and efficiency. Comprising the steps of performing data preprocessing on multi-source heterogeneous data, and constructing a unified data pool; performing 1: 1 three-dimensional modeling on the physical space of the market subject to obtain a digital twin model capable of real-time interaction; acquiring market subject state data in real time, synchronizing the state data to the digital twin model, checking the dynamic operation state of the market subject through the digital twin model, and marking an abnormal state; performing parallel computing on a large-scale scene through a parallel visual rendering algorithm, simulating a risk evolution path in combination with a simulation deduction and algorithm integration technology, and generating a risk level and early warning information; and generating a processing task based on the risk level and the early warning information, and feeding back a processing result to the digital twinborn model.

The invention relates to the technical field of medical simulation, in particular to an ablation needle puncture training auxiliary system, and aims to solve the problems of insufficient tactile feedback and low fineness when an existing ablation needle puncture training system simulates a needle tip to penetrate through multiple layers of biological tissues so as to improve the'needle feeling 'cognition and fine operation skills of doctors. The system is characterized by comprising a physical interaction module, a virtual simulation module, a central processing module and a visual presentation module, a high-fidelity virtual tissue model is constructed, real-time multi-mode tactile feedback is combined, and an accurate needle-tissue interaction mechanical simulation and multi-dimensional tactile rendering algorithm is adopted. By the adoption of the technical scheme, the problem of a touch gap is effectively solved, the authenticity and effectiveness of simulation training are improved, and the precision and safety of clinical puncture operation are remarkably improved.

This invention discloses a collaborative cloud rendering system, including a lighting and shadow rendering module. The module's direct lighting rendering unit performs rendering through direct lighting calculations, the shadow rendering unit performs rendering through shadow calculations, the ambient occlusion rendering unit performs rendering through ambient occlusion calculations, and the indirect lighting rendering unit performs rendering through indirect lighting calculations. The indirect lighting rendering unit runs on a cloud backend device. A hybrid rendering unit performs rendering through hybrid calculations and runs on a web frontend device. Depending on the performance of the web frontend device, each unit runs on either the web frontend device or the cloud backend device. Therefore, this invention proposes four key real-time lighting and shadow rendering algorithms for collaborative real-time dynamic lighting and shadow rendering in Web3D cloud rendering, optimized through analysis of various data such as algorithm frame rate, the operating efficiency of the device running the algorithm, and the lighting and shadow rendering results.

The invention relates to the technical field of virtual reality, in particular to an all-true virtual reality virtual-real fusion method and system, and millimeter-level virtual-real fusion is realized through light field reconstruction, algorithm dimension raising and space programming. Light field reconstruction adopts a 6500K spectrum matrix and a P0.9-P2.5 LED array to construct a three-order framework, and 0.1 mm positioning precision and 18KHDR rendering are realized in combination with a 200-point / cm < 2 > laser grid and a 1000Hz TOF array; in algorithm dimension raising, SE (3) manifold optimization is utilized to control the 6DoF pose error to be + / -0.03 degrees, 4K-18K super-division (the information amount is increased by 5.2 times) is achieved through a GAN network, and a material sensing engine simulates 87 types of surface characteristics (the reflectivity error is smaller than or equal to 0.8%); the distortion rate of space programming is reduced from 2.1% to 0.05% based on an SIGGRAPH2024 algorithm, 5.7 m < 2 > / second modeling is realized through three-level computing power scheduling, and delta Elt under 105lux is ensured through 1269 color block light field compensation; and 1.2. The system supports 3.2 TB / min VR content conversion, adapts to 23 types of special-shaped screens, and solves the problems of insufficient precision and poor environmental adaptability of traditional VR.

This invention discloses an artificial intelligence-based method, system, and storage medium for generating panoramic sound audio. The method includes the following steps: S1. Inputting the original audio signal and scene parameters, obtaining the time-domain-frequency domain features and spatial scene features of the original audio signal through a multimodal feature extraction algorithm, and weighted fusing the time-domain-frequency domain features and spatial scene features to obtain fused features; S2. Constructing the mathematical expression basis of the 3D spatial sound field model based on the spherical harmonic function, mapping the fused features to three-dimensional spatial coordinates through a sound field modeling algorithm to obtain the 3D spatial sound field model; S3. Using an adaptive rendering algorithm, converting the 3D spatial sound field model into multi-channel panoramic sound audio output according to the parameters of the target playback device. This invention achieves automated panoramic sound audio generation by inputting the original audio signal and scene parameters for feature extraction, fusion, and sound field modeling, and converting the output based on adaptive rendering. It has the advantages of automation, high efficiency, and high precision.

The application relates to the technical field of industrial detection, and discloses a lithium niobate metasurface defect intelligent identification method based on reactive ion beam etching, which systematically solves the problem of experimental data scarcity through physical driving simulation and a physical-based rendering algorithm, and provides large-scale field-specific training data for a deep learning model. A bidirectional coupling simulation model generates defect morphology prediction data conforming to etching dynamics rules based on real process parameters and material parameters, a physical rendering algorithm converts the prediction data into synthetic images with real microscopic image statistical characteristics, and an automatic labeling and statistical verification process ensures the quality and consistency of the synthetic data. The method expands the sample quantity of the ternary correlation database by several times, and covers extreme process conditions and rare defect modes that are difficult to obtain through experimental acquisition.

The invention discloses an audio transcoding method and device, equipment and a storage medium, and relates to the technical field of audio signal processing, and the method comprises the steps: carrying out the analysis of a target 2D audio and a target video, so as to obtain corresponding audio features and video features; determining a target scene type corresponding to the target video by using the audio features and the video features, and separating a plurality of independent object sounds from the target 2D audio; generating a space coordinate track corresponding to each independent object sound, and mapping each space coordinate track to a target sound channel position; and determining a target rendering algorithm based on the target scene type, and rendering each independent object sound in the target sound channel position by using the target rendering algorithm. Through the means of determining the target scene type by using the audio features, the video features and the target scene analysis model, the problem of poor audio and video adaptation degree is solved.

The application belongs to the technical field of three-dimensional reconstruction and machine learning, and discloses a generalizable neural radiance field reconstruction method based on patch extraction, which is based on a multi-view stereo vision and a volume rendering algorithm of a neural radiance field to implicitly learn a static three-dimensional scene, realize new view synthesis of a complex scene at any angle, and perform three-dimensional reconstruction of the scene. Firstly, two-dimensional image features of a source view and a target view are extracted, and a three-dimensional cost volume is constructed by using a plane scanning algorithm to distort the coordinate system of reference view features through homography transformation. Secondly, the cost volume constructed for the current scene is input into a three-dimensional convolutional neural network to obtain a neural encoding volume. The neural encoding volume is input into a multi-layer perception to regress volume density and color, so as to construct a neural radiance field. Then, the current neural radiance field is used for rendering to obtain a target view, patches of the target view and the reference view are extracted respectively, and reference view patch features and target view patch features are extracted by using a pre-trained VGG-16 low-level network to compare the features. Finally, the average absolute error is calculated to quantify the content feature difference between the target view patch and the reference view patch, and the difference is used as a regular term of an overall loss function to improve the rendering quality of the model. The application can be used for generalizable multi-view three-dimensional reconstruction, can enhance local features through an efficient training mechanism, and can improve the rendering capability for image details and object boundaries.

The invention discloses a composite distributed rendering method based on a depth map algorithm and a blocking algorithm, and relates to the technical field of image processing. The method comprises the following steps: dividing a scene image into a plurality of image areas, and partitioning the image areas based on a partitioning algorithm to obtain a plurality of first image blocks; based on the information complexity value corresponding to each first image block, classifying each first image block into a first type of image blocks and a second type of image blocks; based on a depth map algorithm, obtaining rendering priorities corresponding to the second-class image blocks; and on the basis of the rendering priority, independently rendering each second-class image block, and on the basis of a block rendering algorithm, carrying out block rendering on each first-class image block, and finally obtaining a rendered image. According to the method, the image blocks of the rendered image are classified according to the information complexity value, and different rendering modes are adopted for different image blocks, so that rendering resources are fully utilized, and the purpose of improving the rendering quality and efficiency is achieved.

Inverse rendering is important for training neural networks for generative artificial intelligence (AI), and it involves inverting the rendering process by taking an image and converting it into scene or model parameters that can be backpropagated through a network, helping to train the network to learn to generate models, materials, textures, etc. Because of the gradients required for this backpropagation, inverse rendering requires differentiable rendering algorithms. Current differentiable renderers are based on rasterization which make it difficult to learn scene properties depending on second order effects. The present disclosure provides closest silhouette queries for computing differential visibility, which can be used for inverse rendering.

The present disclosure relates to the technical field of audio processing technology, and in particular to an audio processing method, an audio processing apparatus, a computer-readable storage medium and a computer program product. The audio processing method includes: receiving IR information of audio data transmitted by a first renderer; determining, based on a sound field structure of the IR information, a corresponding rendering algorithm for the IR information; and processing, based on the corresponding rendering algorithm, the IR information by using a second renderer.

The application discloses a kind of real-time atomization video's augmented reality method, system and computer readable storage medium, method includes:1, augmented reality equipment gathers camera picture and coding to generate uplink video stream, and push stream to server;2, server is decoded to uplink video stream, shaping, and encapsulation is frame batch;3, call single view three-dimensional reconstruction model to frame batch, generate three-dimensional main scene and pixel level depth, according to pixel level depth calculation three-dimensional world coordinate;4, generate a atomization Gaussian particle for each pixel;5, 3DGS volume rendering algorithm will Gaussian scatter point of three-dimensional scene with atomization Gaussian particle be volume rendering and forward synthesis, obtain enhancement frame;6, to enhancement frame coding to generate downlink enhancement video stream and push stream to server;7, augmented reality equipment is pulled from server enhancement video stream and real-time decoding and display.The application has the advantages that augmented reality equipment end side burden is low, end-to-end time delay is small and atomization level is controllable.

The invention provides a display method and device, electronic equipment and a storage medium. The method comprises the steps that a video stream is acquired, a video layer is determined based on the video stream, and the video stream is used for presenting at least one animation obtained through pre-rendering; according to a layer set of the current frame, display content of a display screen of the wearable device in the current frame is determined, and the layer set comprises at least one dial plate layer and the video layer. On the basis, on one hand, the display of complicated contents of the dial plate can be realized on the premise of not depending on a real-time rendering algorithm, so that some relatively complicated animations can be applied to wearable equipment such as a smart watch on the premise of not deleting an animation algorithm; and on the other hand, the video layer and other layers are decoupled, so that the drawing processes of the video layer and the other layers do not interfere with each other, and the influence of the rendering bottleneck of the video layer on the refresh frame rate of the display screen is avoided.

The invention belongs to the field of image processing, and discloses an image transformation method based on Gaussian function parameterization representation, which comprises the following steps: S1, initializing a Gaussian function set according to a target image; s2, generating a rendered image by using an accumulation rendering algorithm; s3, training all parameters in the Gaussian function set by taking loss function convergence as a target to obtain an optimized Gaussian function set; s4, performing parameterized Gaussian representation on the target image by using the optimized Gaussian function set; s5, a Gaussian space transformation model is established, and image transformation is converted into updating of position and shape parameters in a Gaussian function set; and S6, using the updated Gaussian function set to regenerate a rendered image, namely a transformed image. According to the method, the target image is decomposed into a group of Gaussian basis functions, the image transformation operation is converted into linear algebraic operation of Gaussian basis function parameters through the Gaussian space transformation model, complex coordinate transformation and interpolation calculation can be avoided, and the calculation efficiency is effectively improved.

The invention provides a virtual modeling system and electronic equipment, and the virtual modeling system comprises a motion capture and collaborative synchronization module which is used for capturing gesture operation data of a first user under the condition that the first user logs in the virtual modeling system, synchronizing the gesture operation data to a second user in real time through a synchronous rendering algorithm, determining an operated virtual object according to the gesture operation data, and performing operation conflict detection on the first user and the second user; wherein the second user is another user who logs in the virtual modeling system; the physical feedback module is used for determining physical parameters of the operated virtual object; and the rendering and modeling module is used for performing virtual modeling based on the physical parameters and synchronizing the modeling state of the virtual modeling to the second user in real time through a distributed rendering algorithm.

The application provides a multi-device dynamic adaptation WebGL display optimization method and system, and belongs to the technical field of WebGL display. The method comprises the following steps: collecting display parameters of a target device to construct a device display characteristic vector; performing preprocessing on a texture image in a WebGL model; training and generating a mapping relationship library based on historical adaptation data, dynamically adapting the device characteristic display vector of the target device, obtaining required WebGL model running parameters, and dynamically adjusting a calculation graph of the preprocessed WebGL model; executing a vector graphics rendering algorithm on a fixed layout of the WebGL model, dynamically sampling and optimizing dynamic content of the WebGL model, and realizing multi-resolution adaptation; and monitoring the state of the target device in real time, pausing the rendering task of the WebGL model when detecting display mode switching, and re-collecting device parameters. The application realizes optimal visual effects and performance optimization on different devices.

The application relates to a map rendering method, device and equipment and a storage medium. The method comprises the following steps: loading data range current map element data; classifying the current map element data to obtain map element data of a first type of map element and / or map element data of a second type of map element in the current map element data; using a first rendering algorithm to render and display the map element data of the first type of map element in a display range buffer; and / or using a second rendering algorithm to render and display the map element data of the second type of map element in the display range buffer. The scheme provided by the application can make the display picture of the map flicker-free and improve the user experience.

The application discloses an X-ray image reconstruction method based on a parallel three-channel convolutional neural network, which comprises the following steps: (1) collecting multiple sets of training data, each set of training data comprising three X-ray projection images of a security inspection machine and a corresponding CT sequence, and marking in the image data; (2) constructing a generator model and a discriminator model; (3) locking the parameters of the discriminator model, training the parameters of the generator model, and designing a loss function of the first stage; (4) locking the parameters of the generator model, training the parameters of the discriminator model, and designing a loss function of the second stage; (5) iteratively training steps (3)-(4) until the parameters of the generator and the discriminator model reach local optimization; (6) inputting three X-ray projection images to be reconstructed into the trained generator model, and outputting a three-dimensional CT sequence; and performing visualization by using a volume rendering algorithm. The model of the application has a fast inference speed, is easy to deploy, and has high objective indexes and high subjective similarity in the generated reconstruction effect.

In some embodiments, a computing system accesses multiple input images of a specular object with a scene. The computing system encodes near-field interreflections of the scene on the specular object to obtain a first set of feature representations of the specular object in multiple viewing directions based on the multiple input images. The computing system encodes far-field reflections of the scene on the specular object to obtain a second set of feature representations in the multiple viewing directions based on the multiple input images. The computing system determines a set of specular color values for the specular object in the multiple viewing directions based on the first set of feature representations and the second set of feature representations using a multi-layer perceptron algorithm. The computing system renders the specular object representation at least based on the set of specular color values using a neural rendering algorithm.

The application discloses a scene reverse rendering method and device based on neural implicit expression, and the method is based on a signed distance field network, a neural radiation field network and a neural material field network to realize joint implicit expression of geometric information, illumination information and material information of a scene, and a ray tracing rendering algorithm based on Monte Carlo sampling is used to obtain a rendering result, a loss function is constructed according to the rendering result to optimize network parameters, and the signed distance field network, the neural radiation field network and the neural material field network capable of accurately predicting the geometric information, the illumination information and the material information are obtained. Meanwhile, the method also supports re-illumination editing and material editing, and achieves an augmented reality function.

The application discloses a kind of based on AI speech synthesis and animation driving's lecturer video automatic generation method, comprising: by improved inner point method and incremental shortest path algorithm to PPT or text script is structured and analyzed;Application full dynamic parallel single link clustering algorithm carries out semantic grouping and generates enhanced script with expressive mark;High-quality voice data stream is generated by using CosyVoice technology combined with low-rank approximation method;The mapping relationship between content and action expression is established by semantic analysis, and a complete set of action expression instructions is generated;The museTalk technology is used to drive the digital human model, and the final lecturer teaching video is generated by using parallel rendering algorithm.The application realizes the efficient and automatic generation of educational videos, significantly improves the content production efficiency, reduces the production cost, and ensures the professionalism and expressiveness of the teaching video.

The invention discloses a three-dimensional model loading optimization method and device and a storage medium, and relates to the technical field of three-dimensional graph rendering. The three-dimensional model loading optimization method comprises the following steps: performing compression processing on an obtained three-dimensional model based on a Draco compression library to obtain a compressed three-dimensional model; a MergeHelperHY optimization tool is constructed, small models in the compressed three-dimensional model are combined, and a combined three-dimensional model is obtained; and calling a rendering algorithm for the combined three-dimensional model to obtain a three-dimensional model loading result. The optimized model can have high loading speed and interaction response speed during page loading, the jamming phenomenon is effectively avoided, the user experience is improved, and it is ensured that good performance can be kept when more complex or larger three-dimensional models are processed.

The rendering model training method comprises the steps that an object in a first digital content scene is rendered through two different rendering modes, two rendered images which are different in rendering quality but the same in number are obtained, and the two rendered images are kept consistent in view angle; based on the two rendered images, performing fine tuning on the general AI image denoising model; when the rendering effect of the target model does not meet the requirement, the number of the two rendering images is updated, the general AI image denoising model continues to be finely adjusted, the number of the rendering images is configured by a user, and the number of the two updated rendering images is the same; the target model is a fine-tuned general AI image denoising model, and the target model is used for converting a low-quality rendered image into a high-quality rendered image; and when the rendering effect of the target model meets the requirement, fine tuning of the general AI image denoising model is terminated. In this way, the model after fine tuning can be combined with a traditional rendering algorithm subsequently, so that the requirement for high-quality and high-efficiency rendering scenes is met.

The invention relates to the technical field of large model front ends, in particular to a large model relation component display method and system based on angular, and the method comprises an angular packaging step of a topological graph component and a dynamic assembling step of a topological data source. The method has the beneficial effects that efficient visualization of a multi-level topological relation is realized by packaging a reusable topological component through Anglar and combining a self-defined node / edge rendering algorithm and a dynamic data assembly method, and the system readability and the user experience are improved.