76 results about "Visual structure" patented technology

The invention discloses a cross-modal interactive image restoration method fusing text semantic guidance and visual structure priori, which comprises the following steps of: firstly, acquiring natural language description input by a user and an image to be restored, and generating a semantic segmentation map of the image through a semantic segmentation model; encoding the text and image semantics by using a pre-trained cross-modal encoding model to obtain text and semantic features; guiding a semantic alignment attention module through Prompt to realize deep fusion of multi-modal semantic features and image space features; structural enhancement and regulation of image features are realized by constructing a text guide weight graph, performing element-level modulation on the text guide weight graph and the optimized semantic segmentation graph, constructing a cross-modal structure semantic feature graph and generating a structural modulation factor; a four-stage image restoration network is adopted, and a high-quality restoration image conforming to semantic guidance and structure prior is generated step by step. According to the method, the semantic consistency, the structural integrity and the visual reality sense of an image restoration result are improved.

A multi-feature three-dimensional reconstruction 3D Gaussian method based on laser vision comprises the steps that laser radar point cloud and camera images are aligned through space-time calibration, and a unified coordinate system is established; extracting geometric features by using point cloud data acquired by the Lidar point cloud, and initializing a Gaussian ellipsoid according to the Lidar point cloud; optimizing the brightness, the contrast ratio and the structural similarity of the rendered image and the real image by combining the mean absolute error L1 and the structural similarity SSIM; the curvatures of Gaussian ellipsoids of K-nearest neighbors are forced to be consistent, and long and short axes and line and surface features of the Gaussian ellipsoids are aligned to reduce geometric distortion; the distribution density of 3D Gaussian is dynamically adjusted through line / surface features and visual structure information extracted by Lidar, and balance between geometric detail enhancement and calculation efficiency is achieved. According to the method, the position, the scale and the rotation parameters of Gaussian are uniformly optimized, and the details and the calculation efficiency of the model are balanced while the consistency of the model structure is improved.

The invention belongs to the technical field of precision measurement and intelligent calculation, and particularly relates to a multi-mode large model driven inertial platform temperature field generation method. Comprising the following steps: (1) data acquisition and preprocessing: acquiring a structure appearance image and a multi-working-condition thermodynamic distribution diagram of an inertial platform, and performing space alignment by using affine transformation or perspective transformation; (2) descriptive text generation: inputting a multi-angle RGB image of the inertial platform, extracting spatial features through a pre-trained visual large model, and generating structured text description to describe the inertial platform and the environment thereof in the image; (3) constructing a diffusion model: inputting an inertial platform image and a descriptive text, respectively extracting visual structure features and semantic description features, and controlling the generation direction of a temperature field by utilizing the characteristic of gradual denoising of the diffusion model and combining text semantic guidance; and (4) model training and verification: selecting an optimizer and a learning rate scheduling strategy, and verifying the rationality of a model concern area.

The invention provides a front-end code evaluation method and system, electronic equipment and a program product, and the method comprises the steps: obtaining a front-end code, carrying out the compiling and rendering of the front-end code, and obtaining a generated front-end page and a target front-end page; performing code extraction processing on the generated front-end page and the target front-end page, and performing redundant node cleaning and code structure tree generation processing on extracted markup language codes to obtain a generated code tree and a target code tree; performing tree structure and element content error calculation processing on the generated code tree and the target code tree to obtain a content structure error; performing visual structure error calculation processing on the generated front-end page and the target front-end page to obtain a visual structure error; and according to the content structure error and the visual structure error, determining and obtaining an evaluation result for generating the front-end code. The scheme can improve the generation accuracy of the front-end code, and can be widely applied to the technical field of artificial intelligence.

The invention relates to the technical field of equipment asset management, in particular to an intelligent equipment asset management system and method based on a visual atlas, and the method comprises the steps: creating a visual structure comprising a floor father node and an equipment child node through a go.js engine, configuring differentiated visual identifiers for equipment in different states, completing the construction of a dynamic tree graph; establishing connection with an asset database, and obtaining equipment state update in real time; when the change of the equipment state is detected, the visual performance of the corresponding node is automatically updated; responding to a floor node click event, and executing unfolding / folding animation display subordinate equipment; receiving a search condition input by a user; highlighting matched nodes in the tree atlas, and automatically folding uncorrelated branches; when the equipment is selected, generating a time axis visual chart containing a state change time point; and the maintenance record document corresponding to each state change is displayed in an associated manner. And the management efficiency and the decision accuracy are improved.

A geometric structure aided visual localization method and a system implementing the method are provided. The method includes retrieving a 3D map of initialization location data, the 3D map being constrained by visual structures and geometric structures and modeled by a Gaussian mixture model of a set of Gaussian distributions with mapped landmarks; acquiring a series of real-time image frames by a camera of a mobile system; and for each real-time image frame: extracting local features from the real-time image frame; predicting a camera pose corresponding to the real-time image frame; creating a key frame by tracking the predicted camera pose in the 3D map; identifying temporally visible landmarks with respect to the created key frame; acquiring 2D-3D correspondences between the local features and the temporally visible landmarks; and localizing the mobile system by estimating a state of the real-time image frame based on the 2D-3D correspondences.

The utility model relates to a swimming suit technical field, specifically disclose a cover cup swimsuit with photo frame round hole structure, including cover cup cover and sponge cover cup, the surface of cover cup cover is equipped with hole body, cover cup cover covers on sponge cover cup, the front end face of sponge cover cup is outwardly convex through hole body, compared with the traditional collection of this technical scheme, and the process of saving position, the cover cup edition technology of wrinkle, this technical scheme replaces the collection of hole body cooperation cover cup protruding or wrinkle, from the effect, the surface has no trace of saving position or wrinkle, improves the aesthetic property, in addition, improves the three -dimensional penetrability of round hole, breaks through the planarization, the single limitation of traditional process, creates novel visual structure, product appearance has more discernibility and innovativeness, from the process, simplifies the complicated collection of saving position calculation, the operation such as wrinkle setting, reduces technical threshold and production difficulty, improves the edition efficiency and the finished product flatness, realizes multiple breakthroughs on structural rationality, appearance uniqueness, production convenience.

The invention discloses a transformation method of a traditional filter press, and relates to the technical field of solid-liquid separation equipment.The transformation method comprises the steps that a rack body of an original filter press is reserved; a pressing plate driving structure of the original filter press is reserved; the original pressing plate is reserved; a pulling plate driving structure and a pulling plate trolley in a pulling plate mechanism of an original filter press are reserved, and a visual structure and a distance measuring structure are arranged on the pulling plate trolley; a filter plate group of the original filter press is replaced by a filter pressing plate structure; a filter cloth monitoring structure is additionally arranged on a rack body of an original filter press; a sludge falling structure is additionally arranged on a rack body of an original filter press; the visual structure is used for collecting an image of the filter cloth assembly, the distance measuring structure is used for monitoring the position of the filter pressing plate, the filter cloth monitoring structure is used for monitoring the position of the filter cloth assembly, and the mud falling structure is used for separating a filter cake from the filter cloth assembly. The method is implemented on the basis of existing equipment, the structure of original matched equipment is not changed, the transformation cost is greatly reduced, and the method is suitable for intelligent upgrading of stock markets.

The application discloses a kind of based on machine vision's reinforced concrete silo construction safety monitoring system and method, related to the field of silo construction, solve the problem that existing reinforced concrete silo construction safety monitoring system has poor monitoring effect, including warehouse outside monitoring module: to be in visual structure monitoring period first to fifth visual monitoring plane periodic visual image analysis, and according to the analysis result obtains the structure visual monitoring deviation corresponding to target shallow circular silo, warehouse inside monitoring module: respectively to each warehouse inside monitoring area is damaged part monitoring, according to the monitoring result obtains the damaged area area ratio corresponding to each warehouse inside monitoring area, and according to this analysis obtains the warehouse inside area damage coefficient, monitoring early warning module: according to structure visual monitoring deviation and warehouse inside area damage coefficient, target shallow circular silo is carried out construction safety early warning, the application can provide guarantee for the construction safety of reinforced concrete silo.

The invention discloses an angle steel tower defect identification method based on a graph neural network. The method comprises the following steps: step 1, constructing an initial structure graph; 2, constructing visual structure joint features of dominant nodes to form a first node feature set; step 3, constructing label missing perceptual coding features of the hidden nodes to form a second node feature set; 4, performing feature estimation on the second node feature set by using a feature probability migration module to form a third node feature set; 5, correcting the third node feature set to obtain final hidden node features, and forming a fourth node feature set; 6, fusing the first node feature set and the fourth node feature set to generate a graph embedding vector set; step 7, performing defect type prediction on each node; and 8, outputting a defect identification result. According to the invention, the graph neural network and the feature probability migration module are combined to realize structural level defect intelligent identification of the unobservable area of the angle steel tower.

To improve work efficiency at maintenance time of a device to be diagnosed.SOLUTION: When having an operator recognize a component or a region where there is possibly occurrence of abnormality in a device being diagnosed that is obtained by analyzing sound data emanating from the device being diagnosed, the component or the region where there is possibly occurrence of abnormality is displayed by a display unit 24 by superimposing marks 70-73 at corresponding points in visual structure information 61 with which the structure of the device being diagnosed is visually graspable. A list of the components or regions where there is possibly occurrence of abnormality may be displayed by the display unit 24.SELECTED DRAWING: Figure 8

The invention discloses a data visualization narrative generation method, device and equipment based on paramagnetism, and a medium, and relates to the technical field of artificial intelligence. The method comprises the following steps: generating sentence-level parastyle features according to parastyle contents input by a user; generating corresponding plot information for the parastyle features; according to data input by a user, determining data feature information corresponding to the data feature name in the external features; generating visual structure information of features according to the data feature information; generating animation structure information of the parastyle features according to the parastyle features and the plot information, the data feature information and the visual structure information corresponding to the parastyle features; and obtaining to-be-rendered data based on the white features and the plot information, the data feature information, the visualization structure information and the animation structure information corresponding to the white features, and rendering the to-be-rendered data to generate a data visualization narrative picture. The visual narrative content with high personalization is generated according to the generation of the visual narrative driven by the user-defined parastaring.

The invention relates to an adhesive dispensing area acquisition method and device and an adhesive dispensing method and device, which are used for installing a functional assembly on a base through retention glue, the retention glue at least comprises a first colloid, and the method comprises the following steps: acquiring edge information of a visual structure of the base; acquiring boundary information of the cavity of the base based on the edge information of the visual structure; obtaining size information of the functional component; obtaining a target placement position of the functional component in the cavity according to the boundary information of the cavity and the size information of the functional component; and based on the boundary information of the cavity and the target placement position of the functional component in the cavity, obtaining a dispensing area of the first colloid. According to the dispensing area obtaining method and device and the dispensing method and device, the dispensing precision and the mounting reliability of the functional assembly are improved, the packaging consistency is ensured, and the product yield and the production stability are further improved.

The invention discloses a suspicious person monitoring method and system for bank security and protection, and relates to the related field of image processing, and the method comprises the steps that a multispectral imaging device collects a person movement image data stream to construct a three-dimensional behavior image data body; constructing a virtual behavior simulation environment, generating a multi-path standard behavior track set, and constructing a behavior inertia reference map; executing cross-time-domain optical flow consistency analysis and local visual structure stability calculation of the data volume to construct behavior trajectory disturbance field distribution, and generating a behavior inertia damage degree distribution matrix based on a difference mapping relation with the atlas; carrying out micro-scale visual enhancement processing, and constructing a micro-action tensity tensor by utilizing sub-pixel-level jitter evolution detection and skin table reflection dynamic fluctuation analysis; and inputting the tensity tensor and the damage degree distribution matrix into a visual risk time sequence evolution channel to generate a suspicious risk evolution curve. The technical problem of limited identification precision in existing suspicious person monitoring is solved, and the technical effect of improving the identification precision is achieved.

The invention discloses a lightweight binocular vision structure three-dimensional displacement and motion trail measurement method and system. The method comprises the following steps: performing binocular calibration on binocular equipment consisting of a master camera and a slave camera to obtain camera parameters; collecting a motion image of the to-be-detected structure with the single square composite target; processing the image to identify a positioning point locking target area and extract angular point coordinates; judging whether the target exceeds the field of view of the main camera according to the angular point coordinates, if so, switching the data of the slave camera, and fusing the data of the double cameras in the overlapping region; and calculating six-degree-of-freedom displacement based on camera parameters, angular point coordinates and switched data to obtain three-dimensional displacement and a motion track. The system comprises a binocular camera, a calibration module, an image processing module, a view field switching module, a calculation module and a display storage module. According to the method, the anti-interference performance and the arrangement efficiency are improved through the single-target design, the measurement range is expanded through switching of the field of view of the master and slave cameras, the efficiency is improved through lightweight calculation, the high-precision, large-range and dynamic measurement requirements can be met, and the method is suitable for engineering structure monitoring and precise manufacturing scenes.

The application provides a text branch-based AI conversation structure graph and a context dynamic assembly method. The method allows the user to open a new conversation unit as a text branch by designating the selected content as a context anchor from any conversation unit, thereby dynamically organizing the conversation into a visual structure graph with each unit as a node. The context dynamic assembly algorithm is based on the context anchor and the graph, intelligently evaluating the relevance of each historical conversation and the current input from the dimensions of semantic association, structural proximity, logical context, etc., and calculating the comprehensive weight and sorting for the most relevant content. Finally, the algorithm intelligently balances the fidelity and abstractness of the information at each level according to the sorting and the budget of the smallest semantic unit (Token), constructs a context load with optimal information density and dynamic fidelity, and provides it to the AI, significantly improving its understanding and response ability to the user's real intention in complex multi-threaded conversations.

The application relates to the technical field of data analysis, and discloses a webpage record authenticity verification method, device, equipment and medium, which comprises the following steps: obtaining a verification instruction triggered when a user browses a target webpage, extracting a webpage resource locator and an identity identifier in the verification instruction, constructing an index field, then traversing and reconstructing historical webpage content of the target webpage in a distributed evidence directory, and calculating a historical hash value. Meanwhile, real-time webpage content is obtained according to the webpage resource locator, text content, webpage layout and non-text elements are jointly analyzed through a multi-modal content perception engine, core main text, key metadata and visual structure fingerprints are extracted, data content features are generated, and a real-time content hash value is calculated. By comparing the consistency of the real-time hash and the historical hash, if the consistency is consistent, it is determined that the content has not been modified, otherwise it is determined that the content has been modified. The application improves the reliability of webpage record authenticity verification.

The invention provides a digital twinning-oriented campus energy consumption intelligent management method, which comprises the following steps of: projecting data into a three-dimensional campus model by adopting a space mapping algorithm according to an obtained energy consumption data set, and determining a distribution position of a high-energy-consumption area; if the determined high-energy-consumption area exceeds a preset threshold value, real-time data fluctuation is analyzed through an anomaly positioning algorithm, and a potential anomaly source is judged; obtaining the judged abnormal source information, and decomposing the energy consumption level from a macroscopic panorama to a microscopic building by adopting a multi-level analysis method to obtain a layered visual structure; for the obtained structure, rendering a dynamic thermodynamic diagram through a visual expression technology to obtain a visual spatial distribution view; extracting functional characteristic data from the obtained view, generating an energy-saving measure suggestion for the building position by adopting a decision tree algorithm, and determining an optimization path; and through the determined optimization path, the feedback cycle of the energy consumption data set is updated, and a continuous monitoring intelligent management framework is obtained.

The invention belongs to the technical field of wind turbine generator operation monitoring, and provides a wind power curve anomaly detection method, system and device, a medium and a program product, and the method comprises the steps: generating an image-text pair composed of a wind power curve image and an anomaly category text description; based on the image-text pairs, respectively extracting global power distribution features, graph topology features and text semantic features; integrating the global power distribution feature, the graph topology feature and the text semantic feature into a visual-structure comprehensive feature and a semantic-structure comprehensive feature through a parameterization fusion mechanism; and optimizing an alignment relationship between the visual-structure comprehensive features and the semantic-structure comprehensive features in a feature space by adopting a comparative learning method, and constructing a collaborative optimization model to realize anomaly detection. According to the method provided by the invention, the detection precision is improved through complementarity of the multi-modal information, efficient and reliable technical support is provided for intelligent anomaly detection of the wind turbine generator, and the method has practical application value.

The application discloses a continuous children's picture book image generation method based on dynamic memory enhancement. The application firstly performs multi-stage cascading cleaning on original children's picture book images to construct a high-quality image-text pair dataset; inserts a consistency enhancement low-rank adapter into a pre-trained text conditional image generation base model to construct a dual-source dynamic feature memory bank; performs frequency domain decoupling processing on latent feature maps corresponding to reference images and historical images, extracts local edge, texture details and local structure information, and combines a time sequence attenuation bias to perform memory enhancement attention fusion; constructs an automatic preference sample based on a multi-modal visual evaluation model, trains a preference alignment low-rank adapter, and obtains a continuous picture book image generation model based on preference alignment. The application can improve the role appearance consistency, page coherence and visual structure rationality in the continuous picture book image generation process, and reduce the occurrence probability of role drift, style mutation and local structure anomaly in long sequence generation.

The application discloses an optical illusion hidden image creation method based on a text-to-image large model. The method is as follows: 1) an initial feature z0 of a reference image x is extracted by using a pre-trained text-to-image large model; 2) a denoising diffusion implicit model (DDIM) is used to invert z0 based on a constructed inversion trajectory to obtain Gaussian noise; 3) the DDIM is used to sample to obtain a reconstruction feature based on a constructed reconstruction trajectory; 4) a generated trajectory is constructed, and the DDIM is used to sample a noise signal randomly sampled according to a standard Gaussian distribution based on the generated trajectory to obtain a latent space feature, wherein the generated trajectory is equal in length to the reconstruction trajectory, and each time step of the generated trajectory is guided by a target prompt text v; and 5) the text-to-image large model is used to decode to obtain a generated image that conforms to the semantics of v and has the visual structure of x.

The invention discloses a deep sea image instance segmentation system based on a lightweight hybrid visual structure, belongs to the field of instance segmentation, and aims to solve the problems that a model based on natural image training is poor in adaptability to degradation characteristics of deep sea images, low in feature extraction quality and affects a segmentation result. The system comprises a feature extraction module, an image feature fusion module, a dynamic feature fusion module, an instance segmentation module and an integrated weak supervision guidance module. The feature extraction module receives input original data frames including image data frames and video data frames; extracting a feature map based on the image data frame and the video data frame; the extraction strategy is used for adjusting a feature map of the pseudo mask received from the integrated weak supervision guiding module in real time; and the image feature fusion module receives the multi-scale feature map transmitted by the feature extraction module and obtains a fused feature map.

This application provides a method and system for optimizing the page structure of a presentation. The method includes: identifying visual elements in a presentation page, performing logical relationship analysis on the visual elements, and obtaining logical relationship information of the presentation page; mapping the visual structure information to the three-dimensional implicit field of the page to obtain three-dimensional voxels corresponding to the visual elements; calculating a visual focus heat map in the presentation page, and assigning the three-dimensional voxels corresponding to the target visual elements to visually sensitive areas; predicting a global optimal layout based on the positional relationship and first layout information of each visual element in combination with an information density energy function; optimizing the narrative path in the presentation based on the logical relationship information and the user's narrative preferences to obtain third layout information, so that the spatial arrangement relationship of the visual elements conforms to the narrative logic preferred by the user; and performing secondary layout of the presentation page based on the third layout information to optimize the page structure of the presentation page and improve page layout efficiency.

The application discloses a cross-domain cross-source data alignment method and system and electronic equipment. The method first inputs a plurality of sets of table data to be aligned. Then, key-value pairs in the data and positions of the key-value pairs in the table are extracted. Next, a data multi-modal representation model is used to generate vector expressions of the keys, values and visual positions. Semantic distances of the vector expressions from different data are calculated. Finally, the semantic distances between different data are evaluated to determine an alignment result. In addition to using pairing between keys, the application considers matching of values, enhancing matching of keys in the prior art. In addition to the representation in the text, the application fuses a table visual structure as part of semantic representation of the key-value pairs, breaking the limitation of the prior art that only uses single-modal information for matching.

The invention discloses a transient electromagnetic signal denoising method based on WTD-PC-ViT, and belongs to the field of geophysical electromagnetic data processing research, and the method comprises the steps: firstly carrying out the preliminary denoising of a noisy transient electromagnetic signal through employing an improved wavelet threshold denoising WTD method, and reducing the interference of a random noise signal on network training; then multi-scale feature extraction and time sequence modeling are carried out through a PC-ViT deep neural network, and self-adaptive suppression of residual random noise and enhancement of effective information are realized; according to the method, the time-frequency locality of wavelet analysis and the global feature capturing capability of a visual Transformer structure are fully combined, the signal-to-noise ratio of the denoised signal is remarkably improved, the output result is closer to an ideal real signal, and the method has high robustness and generalization capability so as to solve the problems that the signal-to-noise ratio of the late transient electromagnetic signal is low, and the signal-to-noise ratio of the late transient electromagnetic signal is low. And a traditional denoising method is difficult to effectively suppress random noise.

The invention discloses a document semantic restoration and stroke reconstruction method, system and device based on a condition control diffusion architecture and a medium, and belongs to the technical field of document image restoration, and the method comprises the following steps: extracting a visual structure condition, a semantic condition and a degradation condition of a degraded document image; respectively inputting the three types of conditions into corresponding encoders to obtain structure encoding features, semantic encoding features and degradation encoding features; inputting the three types of coding features into a diffusion model, respectively injecting the coding features at different network hierarchies through a layered injection mechanism, dynamically adjusting the weight according to the denoising progress, and generating a reconstructed image; performing coarse-grained restoration, fine-grained restoration and quality refining on the reconstructed image in sequence; semantic verification and visual evaluation are carried out on the repaired image, and a substandard region is positioned for key optimization; and outputting the optimized document image. According to the method, semantic understanding and stroke reconstruction are synchronously carried out through a multi-modal condition control mechanism and a layered injection strategy.

The utility model relates to a visual weighing and leakage detecting all-in-one machine which comprises a frame body, and a cover body is arranged on the front side of the top of the frame body. The visual structure is mounted on the right side of the frame body and used for visually detecting the appearance of the bottle; the weighing structure is mounted in the middle of the frame body and is used for performing standard detection on the weight of the bottle; and the side leakage structure is mounted on the left side of the frame body and is used for performing quality detection on sealing of the bottle. According to the utility model, visual detection, weight detection and side leakage detection are integrated, so that multiple detections of bottles can be automatically completed in a continuous production process, and the detection efficiency is remarkably improved; compared with a traditional single detection device, three detection functions are integrated, the number of devices needed on a production line is reduced, and therefore occupied space and production cost are reduced; according to the utility model, the functions of visual detection, weight detection and side leakage detection are integrated at the same time, so that the requirement of using a plurality of devices in series is reduced, and the production process and the production complexity are simplified.

The invention provides a fault diagnosis method and system for a numerical control machine tool, and the method comprises the steps: converting a complex relation between system element faults into a visual structure model through the knowledge of graph theory and a matrix tool and ISM; therefore, the fault correlation between the system elements can be determined, and the logic structure of mutual influence and dependence can be found out. And finally, structure and hierarchy analysis of a complex fault relationship is realized. In the work, the ISM-based system fault propagation model can directly and clearly display the fault propagation process and accurately position the fault source. According to the method, the key fault source and the key fault propagation path can be dynamically positioned. By neglecting the fault level deviation caused by component fault time accumulation, the limitation of the existing system is overcome. The model solves the problem of ignoring the time dependence of system component faults, and avoids the deviation caused by taking the component fault accumulation number as the fault level. Using only graph theory or data driven diagnostics may result in significant diagnostic errors. Therefore, the method has certain guiding significance in reducing the maintenance time, improving the system reliability and ensuring the safe operation of the machine tool system.

This invention discloses a multimodal cognitive fusion collaborative processing system and method based on AI large-scale model technology. The system includes a large-scale model semantic understanding module, a visual language model parsing module, and a real-time collaborative decision-making unit. The decision-making unit dynamically fuses textual semantic features and visual structural features through a gating mechanism, driving iterative reasoning between the two modules to output structured cognitive results. This invention reduces the illusion rate of key fields under complex layouts, effectively reducing localization errors, and enables zero-sample generalization without templates. Consistent iteration significantly shortens manual review time and greatly improves batch processing efficiency. It incorporates financial-grade security mechanisms such as encryption, auditing, and disaster recovery, providing a general foundation for highly reliable document understanding. The system adopts a microservice architecture, supports elastic deployment on cloud, edge, and client, has fault self-healing and fine-grained permission control capabilities, can be seamlessly integrated into existing business processes, maintains long-term self-evolution, is suitable for various high-compliance scenarios, and significantly reduces operational and legal risks.

This invention provides a method and system for early-stage rumor detection on social media based on the PGQH-TE model, belonging to the technical field of social media content security and information authenticity identification. It includes: encoding each early rumor event into a text modality and a grayscale image modality; extracting a global semantic prior vector from the text using a BERT model, and using this to guide an improved BiGRU network to extract text sequence features, an improved ResNet-18 network to extract visual structural features, and a specially designed parallel variational coding and bit-level adaptive fusion quantum hybrid network PVBF-QHN to extract quantum state features from the prior vector; projecting the text, visual, and quantum modal features into spatial point charges using a triangular electrostatic field Coulomb fusion method, and achieving lossless fusion of high-order cross-modal interactions based on electrostatic potential and potential gradient, thus realizing early-stage rumor detection on social media. This invention can achieve high-accuracy automatic identification of social media rumors in early-stage scenarios with highly sparse information.