38129 results about "Deep learning" patented technology

The invention provides an industrial environment monitoring and accident prediction method fusing multi-modal data, and relates to the technical field of data processing, and the method comprises the steps: carrying out the semantic collection and causal association preprocessing of multi-modal heterogeneous data collected in real time through constructing a dynamic industrial knowledge graph; a customized deep learning model is adopted to extract deep abstract features of each mode, and weak signals and potential risks are accurately represented and uncertainty is quantified; a high-fidelity digital twin model is utilized to drive a deep reinforcement learning algorithm, and dynamic optimization and verification are performed to generate a multi-level and multi-target preventive intervention strategy combination; an intervention strategy is executed through an edge-end-cloud three-layer collaborative intelligent architecture, and online learning and system sustainable evolution are realized by using a closed-loop data feedback mechanism. According to the method, the sensing and early warning capability of the early weak and complex abnormal state of the industrial environment can be remarkably improved, the accident evolution path is accurately predicted, and credible explanation is provided.

The invention relates to the technical field of equipment fault diagnosis, and discloses an equipment fault diagnosis and prediction method based on deep learning, and the method comprises the following steps: S1, collecting multi-modal data in real time through a plurality of sensors installed on equipment; s2, preprocessing the collected data; s3, constructing a hybrid deep learning model; s4, dynamic weighted fusion is performed on the features of different modal data by using an attention mechanism, and comprehensive feature representation is generated; s5, using the marked fault data and normal data to supervise and train the model; s6, inputting equipment operation data acquired in real time into the trained model, and judging the state of the equipment; and S7, generating a potential fault early warning signal based on a prediction result of the model. A piezoelectric vibration sensor and a thermal infrared imager are arranged on a motor bearing through vibration, temperature and sound sensors, vibration waveforms, thermal imaging slices and time-frequency diagrams are synchronously captured, and composite state characteristics such as mechanical wear and temperature anomaly of equipment are comprehensively reflected.

An AI-based Cybersecurity System and Method enable real-time detection, analysis, and mitigation of cyber threats within computing networks using adaptive artificial intelligence. The system continuously monitors network traffic, extracts behavioral and contextual attributes, and applies deep learning-based inference to identify anomalous activities indicating security breaches. The method integrates several computational units, including a network monitoring unit, feature extraction unit, artificial intelligence processor, contextual reasoning processor, and decision synthesis unit, to compute a composite risk index quantifying threat likelihood and severity. A classification processor categorizes detected threats into types such as ransomware, phishing, or unauthorized access, while a mitigation control processor initiates automated response actions to isolate compromised nodes and restore network integrity. An adaptive learning processor updates AI models using feedback from confirmed incidents. This provides a scalable, self-evolving cybersecurity framework that minimizes human intervention and enhances resilience against dynamic and zero-day threats.

The present invention provides a facial recognition system using deep learning methodologies while integrating privacy-preserving capabilities. This system employs convolutional neural networks (CNNs) to extract and classify facial features, ensuring high accuracy in recognition tasks. Moreover, the system addresses privacy concerns by incorporating techniques such as facial feature encryption and anonymization, thereby enhancing user privacy and data security. This invention is applicable across various domains, including security, surveillance, access control, and personalized services, where facial recognition is utilized while preserving individual privacy.

The invention belongs to the technical field of earthquake disaster scene recognition, and discloses an earthquake disaster scene recognition method based on deep learning. The method comprises the following specific steps: S1, data acquisition and preprocessing; S1.1, multi-source heterogeneous data acquisition and establishment of a comprehensive database containing seismic waveform data, surface deformation data, building structure data, geographic information data and historical disaster record data; through fusion of a 3D convolutional network, a graph attention mechanism, a space-time LSTM and an adaptive cross-modal attention fusion technology, combined modeling of a seismic waveform space-time evolution law, an earth surface deformation space distribution characteristic, a building group topology vulnerability and disaster chain time sequence association is realized, the characterization capability of a complex nonlinear disaster mode is effectively improved, and the method has the advantages of high adaptability and high reliability. And disaster assessment response time is shortened to a sub-second level through mixed precision quantification and edge computing deployment, and high recognition accuracy is still kept in a scene with strong noise and data missing in combination with a multi-task classifier and a physical constraint verification mechanism.

The invention discloses a Transform-based cross-modal fusion multi-modal emotion recognition method and device, which are used for solving the problems of modal isomerism, difficulty in time alignment and insufficient dynamic emotion modeling in a multi-modal emotion recognition task, and the method takes the accuracy and robustness of emotion recognition as performance evaluation indexes. Firstly, feature information of three modes of vision, voice and text is obtained, feature extraction is performed on each mode through a deep learning model, then features of different modes are fused by using a cross-mode Transform module, and a complex dependency relationship between the modes is dynamically modeled through a multi-head self-attention mechanism, so that more accurate emotion recognition is realized, and the emotion recognition efficiency is improved. And finally, performing emotion prediction on the fused features based on time sequence modeling and an emotion classification module. According to the method, the problems of modal isomerism, difficulty in time alignment and insufficient dynamic emotion modeling in multi-modal emotion recognition can be effectively solved.

A system and method for AI based user experience curation across multiple scenarios and finite time horizons of interest. The present invention integrates connectionist and symbolic AI techniques to generate coherent, relevant, and personalized content across various domains. The invention bridges the gap between connectionist AI and symbolic AI, enabling more adaptive, immersive, and engaging user experiences while prioritizing security and traceability and contextualization considerations behind recommendations or content generation. The platform furthers dynamic and tailored user interactions with digital systems across individual interactions, preferences, sequences and ongoing engagements across sessions by leveraging the power of analytics, deep learning, and AI to create truly intelligent, dynamic, and responsive user experiences enhanced with optimization and planning faculties.

The invention provides a marine ranch water quality parameter real-time correction and compensation method and system for a multi-source sensor, and relates to the technical field of multi-source sensors, and the method comprises the steps: constructing a double-layer edge computing network, and connecting a sensor through a micro-service architecture to collect water quality data. And carrying out data preprocessing in combination with wavelet transform. And establishing a sensor digital twinborn model, and calculating the real-time credibility. Establishing a multi-dimensional sensor association network, optimizing a weight coefficient by adopting federal learning, and establishing a self-evolution correction parameter matrix; and fusing the sensor data by using a multi-task deep learning model to generate an initial correction value. And calculating a theoretical reference value through a space-time sequence prediction model. A compensation coefficient is adaptively adjusted by adopting a fuzzy decision tree, hierarchical water quality parameter correction is realized, and a closed-loop self-optimization intelligent correction system is formed through verification of a digital twin model. The accuracy and reliability of marine ranch water quality monitoring data are effectively improved.

The invention relates to the field of semiconductor packaging material defect detection, in particular to a semiconductor packaging material defect detection method based on deep learning, which comprises the following steps: acquiring a surface image, and extracting a two-dimensional contour and a feature point set; preprocessing the image, and separating a packaging material main body area; constructing a two-dimensional defect identification model based on Transform, and outputting a two-dimensional detection result; scanning suspected and unknown defect areas to obtain three-dimensional point cloud data, and extracting geometric and texture features; fusing two-dimensional and three-dimensional data through a space-time alignment model; utilizing the multi-modal fusion model to output defect positions and types; and evaluating the defect importance based on the material node connectivity and the stress distribution, and generating a visual detection report. According to the invention, high-precision detection of semiconductor packaging material defects is realized, the defect identification rate, the positioning precision and the detection efficiency are improved through multi-modal data fusion and a deep learning model, and a visual report can be generated based on material structure quantification defect importance.

Systems and methods are provided for improving generative artificial intelligence (AI). Systems and methods can integrate more reliable data sources and enhance generative AI training and inference processes for complex tasks. The integration of real-time data and expert input can be included as crucial steps in aligning AI outputs with improved accuracy. Similarly, fine-tuning methodologies and augmentation algorithms can be used to focus on minimizing the occurrence of fabricated content, thereby significantly increasing the chances that the information generated is both current and credible.

The invention relates to the technical field of data analysis, in particular to an operation decision intelligent analysis method and system based on cross-domain data fusion. The method comprises the following steps: firstly, based on an enterprise multi-domain ontology knowledge base, performing entity identification and relation mapping on heterogeneous data from different business systems through a semantic mapping-based multi-source heterogeneous data dynamic fusion algorithm, and establishing a unified data model; then, a causal reasoning and deep learning fused hybrid intelligent decision engine is adopted to analyze and process the model; then, a multi-level causal relationship network among business variables is constructed through a causal relationship discovery algorithm by utilizing an analysis result of the hybrid intelligent decision engine, and an adaptive business scene analysis model based on reinforcement learning is used to dynamically adjust an analysis strategy according to business environment changes; generating a Pareto optimal decision scheme set through a multi-objective optimization algorithm, and outputting operation decision suggestions; according to the invention, the comprehensiveness and accuracy of intelligent analysis of enterprise operation decisions are improved.

The invention provides an intelligent ecological restoration system, method and device for a high and steep slope of a strip mine in an arid region. Comprising a data acquisition layer which realizes real-time acquisition of multi-dimensional environmental data through InSAR satellite remote sensing, a ground sensor network and unmanned aerial vehicle multispectral imaging; the transmission layer adopts LoRa and 5G hybrid networking; the platform layer is used for constructing a slope stability prediction and restoration scheme optimization platform based on a digital twinborn model and a deep learning algorithm; and the application layer is used for remotely controlling the repairing device through a mobile terminal and a Web terminal and monitoring the repairing progress in real time. The problems that a traditional restoration technology is poor in adaptability, low in vegetation survival rate, high in ecological restoration cost, insufficient in monitoring technology application, insufficient in monitoring feedback mechanism, insufficient in intelligence, long in ecological restoration period and the like are solved.

The invention discloses an image enhancement method and system in a complex coal mine environment, and relates to the technical field of image processing, and the method comprises the steps: carrying out the preprocessing of a collected coal mine image of a target region, and dividing the coal mine image into different semantic regions, including a bright region, a dark region and a dust shielding region, through a deep learning semantic segmentation model; according to semantic region characteristics, a differentiation enhancement strategy is made; a traditional Retinex model is improved, non-local mean filtering is introduced, and an illumination component and a reflection component are decomposed through pixel similarity matching. According to the method, the image is divided into the bright area, the dark area and the dust shielding area through the deep learning semantic segmentation model, differential enhancement strategies are formulated according to different area characteristics, detail distortion caused by global adjustment is avoided, local contrast suppression is adopted in the bright area, illumination compensation is enhanced in the dark area, and the image quality is improved. Noise diffusion of the dust shielding area is inhibited through edge preservation smoothing, the image quality of each area is remarkably improved, and it is ensured that image details in a complex coal mine environment are clear and visible.

A robotic surgical system includes a surgeon consol coupled to a patient consol, and the patient consol coupled to surgical instruments. A surgeon computer is coupled to or at the surgeon consol that is coupled to to one or more surgical instruments. A robotic surgery control system includes an artificial intelligence (AI) system with one or more deep learning algorithms. A feedback loop monitors and collects data from the one or more sensors. One or more cameras provide feedback to the robotic surgical system, and are configured to provide images of an anatomical object in at least a two dimensional (2D) arrangements of pixels / Deep learning algorithms of the AI system distinguish different anatomical objects from the images.

A small-size vehicle detection deep learning model based on feature fusion of multi-scale modules is provided, which solves the problem of small-size vehicle image detection. The model includes a Backbone network, a Neck layer and a Head network, wherein a C2f_DCNv3 module based on the combination of deformable convolution v3 (DCNv3) and a cross stage feature fusion (C2f) module and an SPPF_LSKA module based on the combination of a spatial pyramid pooling fast (SPPF) layer and a large separable kernel attention (LSKA) module are introduced into the Backbone network; a C2f_SCConv module based on the combination of spatial and channel reconstruction convolution (SCConv) and a C2f module is introduced into the Neck layer; and a multi-scale kernel detection (MSK_Detect) module is introduced into the Head network.

Systems and methods for protecting and fortifying machine learning engines, artificial intelligence (AI) engines, large language models, deep learning engines, reinforcement learning engines, and AI-based agentic units. An Offline Protection Unit analyzes characteristics of a Protected Engine, and performs offline fortification of the Protected Engine against attacks; by changing operational properties or operational parameters of the Protected Engine to reduce its vulnerability to attacks. An Online Protection Unit performs analysis of at least one of: (i) inputs that are intended to be inputs of the Protected Engine, (ii) outputs that are generated by the Protected Engine; and based on the analysis, dynamically performs online fortification of the Protected Engine against attacks; by dynamically changing operational properties or operational parameters of the Protected Engine to reduce its vulnerability to attacks.

The invention discloses a deep learning-based small target defect recognition model training method, relates to the technical field of defect recognition model training, and aims at meeting small defect detection requirements, starting with high-resolution diversified data construction and accurate labeling, highlighting weak targets through multi-scale feature fusion and spatial attention, and realizing high-resolution target defect recognition. A hard case scene is processed in cooperation with layer-by-layer screening and secondary intensified training, real-time iterative optimization is achieved through multi-model fusion and online dynamic adjustment and optimization, finally, multi-mode and time sequence dimensions are expanded to capture deeper and dynamic defect information, the missing detection and false detection rate is greatly reduced, and the detection efficiency is improved. The detection efficiency and adaptability of micron-sized defects under a complex process background are improved; furthermore, by means of multi-source data such as infrared, X-ray or 3D morphology and a time sequence modeling means, multiple dimensions are fused, and hidden or early cracks are brought into a detection and prediction range, so that a high-reliability and evolvable intelligent recognition system for the tiny target defects is constructed.

The invention discloses a dynamic visual content collaborative generation method based on a cognitive logic chain, and belongs to the technical field of visual content generation, and the method comprises the following steps: S1, user intention analysis and data input; s2, dynamically constructing a cognitive logic chain; s3, intelligent scheduling of the multi-modal generation module; s4, cross-modal content collaborative generation is carried out; s5, collaborative editing and real-time feedback are carried out; s6, iterative optimization of logic chain driving; s7, multi-dimensional quality evaluation: constructing an evaluation matrix containing semantic consistency, visual attraction and user participation degree, predicting a content propagation effect in combination with a deep learning model, and generating a quantitative improvement suggestion report; and S8, updating the self-adaptive knowledge reversely marking the cognitive logic chain according to the finally adopted content version, extracting a new association rule, and injecting the new association rule into the rule base. Through deep semantic analysis and dynamic logic chain construction, the system accurately captures a core creation target of a user and converts the core creation target into an executable visual strategy.

The invention belongs to the technical field of intelligent control of heating, ventilation and air conditioning systems, and particularly relates to an intelligent optimizing and energy-saving control method for a central air conditioner based on deep learning, which comprises the following steps of: acquiring operation data of a central air conditioning system in real time through an internet of things technology; the operation data comprises operation parameters of cold and heat source equipment, flow and lift parameters of a water pump, fan frequency parameters of a cooling tower, temperature and humidity data of an air conditioner terminal, environment temperature and humidity data, weather forecast data and the like. Through deep integration of Internet of Things perception, deep learning prediction and a multi-objective optimization technology, the limitation of a traditional control framework is broken through, meanwhile, accurate prediction of building cooling and heating loads is realized through construction of a hybrid deep learning model, an optimization objective of a full life cycle perspective is established in combination with an equipment performance degradation model, and the system performance is improved. A federal learning framework is innovatively introduced into region-level energy efficiency management, and the model generalization ability is improved on the premise of ensuring data privacy.

The invention discloses a building ground real-time settlement monitoring device and a use method thereof, and belongs to the field of building structure safety monitoring. The monitoring device comprises a hierarchical sensor network which is used for carrying out multi-time-scale real-time data acquisition and comprehensively obtaining deformation data and related environmental parameters of a building structure; the data processing and analyzing module is used for performing real-time processing and intelligent analysis on the acquired data, and identifying and classifying abnormal deformation characteristics of the building structure in time; the deep learning prediction module is used for quantitatively predicting the probability state and the evolution trend of building settlement by constructing a multi-scale time sequence prediction model; the multi-factor analysis module is used for carrying out coupling modeling and comprehensive analysis on the environmental factors, the structural characteristics and the abnormal evolution process so as to identify key influence factors and action mechanisms thereof; and the risk assessment and early warning module is used for performing grading assessment on the building settlement risk based on the prediction and analysis result and generating corresponding early warning information and decision support schemes.

The invention relates to the technical field of building information models, augmented reality, synchronous localization and map construction, and provides a deep learning-based SLAM-BIM augmented reality cooperative localization method and system, and the method comprises the steps: introducing a Transform time sequence feature extractor and a geometric relation graph, evaluating a dynamic distribution weight through combining with the confidence, achieving the cross-modal closed-loop detection, and obtaining an SLAM-BIM augmented reality cooperative localization result. A lightweight semantic segmentation network and a feature fusion module are utilized, a dense map with consistent geometric semantics is constructed, a space-time error propagation equation is constructed, online calibration is realized by means of BIM scale prior, an incremental fusion algorithm is designed, a global pose is optimized in combination with AR interaction, and the system fuses SLAM visual trajectory features and BIM semantic geometric features through a deep learning technology. According to the method, the problems that traditional SLAM accumulative errors are large and the BIM fusion precision is low are solved, robust positioning and map construction in a complex scene are achieved, the cooperation precision and real-time performance of SLAM and BIM are improved, and the method is suitable for AR scenes such as building construction and operation and maintenance.

The invention discloses a manipulator grabbing method based on deep learning target detection and image segmentation, and relates to the technical field of artificial intelligence and robotics.The manipulator grabbing method comprises the following steps that a scene image to be processed is collected, the image quality is improved through the multi-light-source fusion image enhancement technology, and recognition errors caused by uneven illumination are reduced; and inputting the enhanced image to a pre-trained deep learning model, executing a target detection task, and outputting an initial bounding box and a category label of the target object. According to the method, through multi-light-source image enhancement and high-precision image segmentation, the accuracy of target recognition and contour extraction is remarkably improved, and the capture failure rate caused by image misjudgment is reduced. And meanwhile, geometric consistency verification and a multi-factor grabbing scoring mechanism are introduced, dynamic screening and collision pre-detection are conducted on the paths, the grabbing stability and safety of the mechanical arm in the complex environment are effectively guaranteed, and the intelligence and robustness of the whole system are remarkably improved.

The invention relates to the technical field of electric vehicle battery charging detection, and discloses an intelligent charging pile system and method integrating real-time battery state detection. A data fusion module; a battery health state prediction module; a charging optimization control module; the fault early warning module is connected with the cloud platform and edge computing cooperative processing module; the method comprises the following steps: acquiring battery data through multiple sensors, and constructing a data frame; extracting electrical characteristics, and generating a battery state vector; a digital twin model is constructed, and health state prediction is carried out; formulating a dynamic charging power regulation strategy; comparing the health trend of the battery with an expected behavior, and generating a fault early warning signal; and uploading the data and the strategy to the cloud platform, and updating the control strategy. According to the invention, a multi-sensor data fusion technology is adopted, twin modeling and a deep learning algorithm are combined, the health state of the battery is monitored in real time, the health state of the battery is comprehensively evaluated, and the fault risk of the battery is accurately predicted.

An online testing and diagnosis method for vibration characteristics of blades of wind turbine is disclosed. Steps of testing and diagnosing blade vibration comprises: S1: installing vibration sensors at key positions of a blade, designing an adaptive data acquisition strategy, and automatically adjusting a sampling rate according to a vibration amplitude and environmental changes monitored in a real time; S2: extracting key features reflecting health status of the blade from massive data, and evaluating an impact of wind speed, temperature, and environmental factors on vibration characteristics; S3: designing a customized deep learning model for damages of the blade of a wind turbine, extracting a time sequence data and a vibration signal, identifying a damage among different types of damages and evaluating a damage degree; and S4: automatically adjusting a warning threshold based on a real-time data stream and a historical trend, and drafting a preventive maintenance plan.

The invention discloses a road traffic flow prediction method based on a space-time mixed attention network, and the method breaks through the limitation of a conventional time sequence model and a single deep learning architecture based on the systematic analysis of urban road traffic flow space-time heterogeneity, periodic non-stationarity and road network topological relevance, constructs the space-time mixed attention network, and achieves the prediction of road traffic flow. Spatial heterogeneous correlation of road network nodes is captured through a graph convolution network, dynamic time sequence evolution characteristics of traffic flow are modeled by adopting a hybrid architecture, a residual attention mechanism is introduced to realize layer-by-layer refining of multi-scale spatio-temporal characteristics, and the overall architecture of the method has remarkable advantages in the aspects of spatial topology modeling and time dynamic capture compared with a traditional model. Feature decoupling learning is carried out on multi-source heterogeneous data such as weather and events, adaptive integration of environment sensitive features is realized through a parameterized gating fusion strategy, and the prediction error fluctuation amplitude in an extreme weather scene is reduced by 34.8%.

The invention discloses a light industry supply chain multi-modal data fusion analysis method based on deep learning, and the method comprises the following steps: carrying out the cleaning and standardization processing of text, image, audio and video data collected in a supply chain environment, and constructing a standardized multi-modal data set; then, a special feature extraction network is adopted to generate each modal feature vector, and a feature incidence matrix is constructed through cross-modal correlation analysis; feature weights are dynamically adjusted in combination with a domain knowledge rule base, multi-modal feature interaction is achieved through a cross-modal attention fusion network, and unified fusion features are generated through a self-attention mechanism; and finally, constructing a supply chain decision model, and mapping the fusion feature into a supply chain state evaluation result and an optimization parameter. According to the method, knowledge rule constraint and a deep attention mechanism are fused, supply chain situation awareness precision and decision reliability can be effectively improved, and technical support is provided for intelligent management of the light industry supply chain.

The invention discloses an unmanned aerial vehicle electric power inspection image intelligent analysis method and system based on deep learning and multi-modal fusion and a medium thereof, and relates to the technical field of electric power equipment detection. The method comprises the following steps: planning an optimal inspection path by adopting an A * algorithm to realize multi-sensor synchronous data acquisition; adaptive histogram equalization and defogging processing are carried out on the visible light image, non-uniformity correction and temperature calibration are carried out on the infrared image, and filtering and registration are carried out on point cloud data; constructing a multi-scale feature fusion network based on improved VGGNet-16, and introducing deformable convolution and a cross-modal attention mechanism to realize multi-source data fusion; defect detection is carried out based on a three-level template library and a feature map cross-correlation algorithm, and the precision is improved in combination with non-maximum suppression and sub-pixel positioning; and finally generating a detection report containing defect types, positions and maintenance suggestions. According to the invention, the automation level and the detection precision of power inspection are obviously improved.

The invention relates to the technical field of cluster scheduling, and provides a GPU heterogeneous cluster scheduling method and system oriented to large model training and reasoning, which constructs a set of complete cluster scheduling system by integrating multi-source information such as hardware features, running states and historical task data and applying technologies such as a clustering algorithm, a fuzzy comprehensive evaluation method and reinforcement learning. Comprehensive, intelligent and dynamic management and scheduling of GPU cluster resources are realized, the cluster scheduling system can significantly improve the execution efficiency of GPU heterogeneous clusters in large model training and reasoning tasks, the resource utilization rate is improved, the energy consumption is reduced, and the stability and adaptability of the system are enhanced. And an efficient and reliable solution is provided for large-scale deep learning application.

The invention discloses a wind turbine generator hoisting construction tower drum operation system and a construction method thereof, and relates to the technical field of intelligent control. The problems that in the prior art, a static tension balance mechanism cannot restrain bending moment abrupt change, steel-concrete interface stress concentration causes microcrack propagation, the wave dynamic load compensation capacity is insufficient, and dynamic rigidity attenuation early warning is lacked are solved. Comprising a dynamic load prediction module, a multi-mode vibration suppression module, an offset compensation module and a digital twinborn decision module, a hoisting load is solved in real time through a multi-physics field coupling model and an improved time sequence deep learning algorithm, and interface crack propagation is suppressed in combination with traveling wave offset control and sweep frequency vibration. An improved Morison equation is adopted to drive a two-stage hydraulic servo to compensate a wave dynamic load, and a digital twin closed-loop correction mechanism is constructed based on a 5G URLLC protocol; the tower drum hoisting precision, the structural safety and the operation reliability under the complex working condition are remarkably improved.

The invention relates to an optimization method and device for sparse view angle three-dimensional Gaussian splash, and belongs to the technical field of three-dimensional reconstruction in computer vision, and the method comprises the steps: collecting a sparse view angle image; a multi-view stereoscopic vision model based on deep learning generates a geometrically consistent depth map for the sparse view image, converts the depth map into point clouds and fuses the point clouds to obtain dense point clouds; sampling dense point clouds by adopting voxel-guided farthest point sampling to obtain initialized point clouds, and constructing a three-dimensional Gaussian field; rendering the three-dimensional Gaussian field through an enhanced geometric renderer to obtain a rendering depth and a rendering normal; constructing a multi-level geometric regularization loss function, and optimizing the three-dimensional Gaussian field; and performing optimization adjustment on the three-dimensional Gaussian field based on a shape-scale constraint criterion and a two-stage adaptive opacity constraint strategy to obtain an optimized three-dimensional Gaussian field. According to the method, the problems of initialization failure, insufficient geometric supervision and element out-of-control of 3D Gaussian splashing under the sparse view angle are solved.