36results about How to "Precision Fusion" patented technology

This invention discloses a data synchronization method and system based on front-end orchestration management, belonging to the field of data synchronization. The method includes the following steps: selecting the data table and field information to be monitored through the front-end interface, generating configuration information, and orchestrating the synchronization process logic according to business needs; storing the configuration information in real time to the configuration center to generate real-time monitoring rules; capturing changed data of the data table and field information, encapsulating it into standard data units and sending it to a message queue; preprocessing the standard data units in the message queue; the synchronization adaptation module matching the corresponding driver type according to the synchronization process logic to generate a synchronization instruction set, and executing data synchronization operations based on the preprocessing results; performing full-link monitoring of the synchronization process, and performing task lifecycle management based on the lifecycle rules configured on the front end. This invention improves synchronization efficiency through dynamic threshold verification, multi-source field fusion weighting, and dynamic encryption strategies, and the full-link monitoring further enhances front-end manageability.

The application discloses a liquid droplet fusion microfluidic method based on a liquid droplet fusion microfluidic device regulated by focused surface acoustic waves, and the device comprises an interdigital transducer, two focusing arc electrodes arranged on the interdigital transducer, and a microfluid channel system bonded on the upper part of the interdigital transducer. The focusing arc electrodes are matched with the microfluid channel system, and the microfluid channel system is provided with a first collection outlet joint, a first dispersed phase inlet joint, a continuous phase inlet joint, a second dispersed phase inlet joint and a second collection outlet joint. The focusing surface acoustic waves generated by the interdigital transducer and the liquid droplet microfluid channel structure are used to flexibly realize the fusion control of liquid droplets with different sizes in the microfluid channel. The symmetrical structure is adopted to enhance the durability and repeatability of the microfluid device. The method is not only suitable for the fusion control between microdroplets, but also suitable for the fusion control between microbubbles and between microdroplets and microbubbles. The surface acoustic waves are non-contact and have good biological compatibility, and thus meet the needs of the liquid droplet fusion technology in the fields of biochemistry and medical treatment.

This application provides a map generation method and apparatus based on high-resolution remote sensing imagery. Its purpose is to achieve accurate matching and fusion of small patches and semantically compatible large patches through comprehensive similarity calculation of spectral, texture, and shape three-dimensional features, ensuring the accuracy of the generated map of the target area. Specifically, the map generation method based on high-resolution remote sensing imagery includes: acquiring a plot raster map based on remote sensing imagery of the target area; converting the plot raster map into an initial vector polygon map to obtain vector patches; calculating the comprehensive similarity between small patches and any large patch based on spectral, texture, and shape features in the feature data of the vector patches, wherein the large patch is a patch whose minimum distance to the small patch is within a preset distance threshold; fusing the large patch corresponding to the maximum comprehensive similarity with the small patch to obtain a plot vector file of the target area, and using the plot vector file to generate a map of the target area.

The invention provides a long focal length optical projection assembly which is composed of a diaphragm and a plurality of lenses from an object side to an image side, the plurality of lenses are a first lens with positive focal power, a second lens with negative focal power, a third lens with positive focal power, a fourth lens with positive focal power and a fifth lens with negative focal power, the object side surface of the first lens is an aspheric surface, and the object side surface of the second lens is an aspheric surface. The second lens is a biconcave lens, the third lens is a meniscus lens with the concave surface facing the object side, the surfaces of the two sides of each of the fourth lens and the fifth lens are aspheric surfaces, through specific focal power distribution and by using the aspheric lenses, distortion is controlled to be at the minimum level (lt, 0.2%), and it is ensured that a projected image and a real scene are accurately fused. And meanwhile, different material combinations are adopted in related lenses to realize visible light full-band apochromatic, so that color ghosting is eliminated, and the reduction degree of image display is improved.

The application relates to the technical field of image processing, and particularly discloses an image registration and fusion method and device, a storage medium and computer equipment. The method comprises the following steps: acquiring PET scanning data and CT scanning data; determining a PET field of view angle corresponding to the PET scanning data and a CT field of view angle corresponding to the CT scanning data; performing three-dimensional image reconstruction on the CT scanning data according to the CT field of view angle to obtain a target CT image, and performing three-dimensional image reconstruction on the PET scanning data according to the target CT image and the PET field of view angle to obtain a target PET image; and performing registration and fusion processing on the target CT image and the target PET image to obtain a target fusion image. The application can more accurately and efficiently integrate PET images and CT images, and can effectively improve the registration effect and overall quality of the PET / CT fusion image.

The invention discloses a many-to-many unmanned aerial vehicle interception method based on a neural network, and the method comprises the steps: constructing a threat function according to the position, height, course and speed of the unmanned aerial vehicles of a friend and a foe for the unmanned aerial vehicles in a many-to-many combat scene, constructing a threat matrix through the training weight of the neural network, and distributing an optimal interception target for the unmanned aerial vehicles of our part; and many-to-many unmanned aerial vehicle interception is realized. According to the method, the threat matrix based on neural network weight training is used to realize maximization of interception advantages.

This invention discloses a method, apparatus, device, and medium for morphological operations based on local density maps, relating to the field of image processing technology. The method includes: calculating a local density map based on a target image; using the local density map to characterize the local grayscale distribution features of each pixel; assigning structural element parameters of different sizes to different regions based on the local density map; performing morphological operations on the target image using structural element parameters of different sizes to obtain multiple morphological operation results; and fusing the multiple morphological operation results based on the local density map to obtain the final morphological operation result. This invention accurately perceives local image features through local density maps, adaptively assigns structural element parameters of different sizes to different regions, and organically combines the processing advantages of different parameters through a fusion strategy. While maintaining high computational efficiency, it effectively balances detail preservation and noise suppression, significantly improving the processing quality of morphological operations.

The application discloses a kind of based on risk assessment and RBA* cargo ship path planning method and system, the cargo ship path planning method based on risk assessment and RBA* of the present application includes the following steps: obtaining the marine environment data of the target area of cargo ship navigation;Risk assessment is carried out to the marine environment data to obtain the risk assessment result of the target area of cargo ship navigation;According to the starting point and end point in the space grid point of the target area of cargo ship navigation, based on risk assessment result, using the RBA* algorithm of fusion risk carries out path planning, to obtain the planning path between starting point and end point.The present application aims at accurately fusing meteorological marine elements, dynamically adapting complex marine environment, realizing safety-efficiency collaborative optimization of marine cargo ship path planning.

The specification discloses an emotion recognition model training method and an emotion recognition method and device, and a storage medium. The first importance weight, the second importance weight and the fusion weight are used to determine a sample emotion result corresponding to a sample voice. A loss function is constructed by using a standard emotion result of the sample voice and the sample emotion result, and an emotion recognition model is trained by using the loss function, so that the emotion recognition model adjusts parameters until the emotion recognition model converges. The converged emotion recognition model accurately fuses the features of the audio, the features of the text and the features of the interaction between the audio and the text.

According to the RGB-D saliency target detection method based on mixed feature cross fusion, through a mixed feature cross fusion module, a model can fully utilize complementary information between an RGB image and a depth image, and efficient fusion of features is achieved. The module enhances the interaction between RGB and depth features through a self-attention mechanism, so that the connection between different modal features is better modeled and captured, and the problem of feature distortion under the condition of poor depth image quality is solved. Besides, the invention provides a specific training strategy and an optimization scheme, and the model can rapidly converge on a large-scale data set by adopting the combination of an MAE pre-training model and a ViT model, and obtains excellent performance in a saliency target detection task.

The invention relates to the technical field of computer 3D modeling, and discloses an urban three-dimensional terrain rapid modeling method and system based on AI point cloud semantic segmentation, and the method comprises the steps: carrying out the high-quality modeling of a feature object three-dimensional model in a target urban terrain according to an urban terrain semantic point cloud fusion preprocessing data set in cooperation with a point cloud three-dimensional modeling platform; precise modeling of the urban terrain in a finite point cloud data state is realized based on semantic segmentation and point cloud fusion; urban terrain three-dimensional model image information is accurately collected through a point cloud three-dimensional modeling platform and screenshot software, and urban terrain three-dimensional model modeling defect intelligent identification is carried out in combination with an AI algorithm and urban terrain feature object standard modeling defect image data; and based on the urban terrain three-dimensional model data and in cooperation with a point cloud three-dimensional modeling platform, performing dynamic and efficient repair and reconstruction processing on modeling defects of the urban terrain three-dimensional model, and realizing accurate modeling repair and reconstruction in an urban terrain finite point cloud data state based on model repair.

The application relates to the technical field of target detection, in particular to a high-altitude electric power operation violation identification method, which discloses the following steps: collecting high-definition images and low-quality images of an electric power operation site by using a drone, and making an electric power site high-definition data set; constructing an image enhancement network based on a local information enhancement module and a global information enhancement module; building a violation identification model based on an improved YOLOv8; selecting a suitable optimizer and a loss function; training and optimizing the constructed model. The optimal model after training and optimization is converted into an RKNN model in the front end, and is deployed to the front end of the drone. Real-time detection is realized by relying on the camera in the front end, and whether the operation personnel violate the rules is judged by the judgment unit according to the detection result. The application provides a high-altitude electric power operation violation identification method, which considers the construction of an operation data set, the enhancement of low-quality images, and simultaneously proposes a violation identification network model based on an improved YOLOv8, so that the accuracy of detection is maximally improved.

The application relates to the technical field of computer vision and artificial intelligence, and provides a remote sensing image target detection method based on an improved YOLOv9s, which comprises the following steps: acquiring a remote sensing image and performing a pretreatment operation; introducing a kernel selection feature fusion (KSFF) module into a neck network of a YOLOv9s model, and replacing at least one Concat structure in the original model; adding a cross-space multi-scale attention (CSMA) module in the neck network; replacing at least one SPPELAN structure in the original model with a parallel pooling feature modulation (PPFM) module in the neck network; and training, testing and evaluating the improved YOLOv9s network. Through the comprehensive application of network structure improvement, data pretreatment optimization, a semi-supervised learning strategy and an advanced training method, the improved YOLOv9s model has higher detection precision in a remote sensing image target detection task.

This application discloses a high-temperature induction furnace, comprising: a cabinet, a feeding assembly, and a configuration assembly. A vacuum chamber is installed on the upper outer side of the cabinet, and a door is hinged to the front outer side of the vacuum chamber. A gas filling and discharging assembly is arranged on the left side of the vacuum chamber. A multi-hole viewing window is installed through the upper front side of the vacuum chamber. From left to right, the feeding assembly, an infrared temperature measurement system, and a radiation baffle are arranged sequentially on the upper rear side of the vacuum chamber. This application belongs to the field of metal material preparation technology. The purpose of this application is to solve the problem that the existing technology has significant functional limitations and cannot achieve auxiliary functions such as suction casting, slag analysis, and smelting distillation through configuration adjustment. The achieved technical effect is that it facilitates the assembly of a structure that can be adjusted to achieve auxiliary functions such as suction casting, slag analysis, and smelting distillation on the upper outer side of the multi-hole viewing window, thereby facilitating the meeting of complex material preparation and experimental needs and reducing functional limitations.

The present application relates to a kind of synthesis method and electronic equipment of coronary artery CCTA image, comprising: obtaining the image of paired NCCT and CCTA, after pretreatment, obtain high consistency image training pair;LDM model suitable for adapting coronary artery image synthesis is constructed, the LDM model includes frozen variational autoencoder, the denoising U-Net of trainable based on cross attention mechanism, and decoder, the denoising U-Net integrates ResNet2D module, Transformer2D module and multi-scale cross attention layer;After pre-processing, the NCCT image to be processed is input into the LDM model trained, and the latent representation of CCTA image is generated by the reverse diffusion denoising process in latent space, and finally the high-resolution Syn-CCTA image is output by decoder.Compared with prior art, the present application has the advantages of no contrast agent to realize coronary artery CCTA image with auxiliary diagnostic value.

This invention relates to the field of visual inspection technology and discloses a paper roll ridge detection system and method. The invention includes: irradiating the surface of the paper roll with a first laser to obtain a first wavelength image, and alternately irradiating the surface of the paper roll with a second laser to obtain a second wavelength image; performing geometric registration on the first wavelength image and the second wavelength image to obtain a first registered image and a second registered image; performing scattering intensity difference operation on the first registered image and the second registered image to obtain a ridge enhancement feature map; and performing cross-attention recognition on the ridge enhancement feature map to obtain a ridge position mask and height prediction results. This invention can automatically optimize detection parameters for different types of paper, solving the problem of poor adaptability caused by fixed threshold settings in existing technologies, and has higher detection accuracy and robustness compared to single-task methods.

The invention provides a distributed multi-target fusion tracking method and device based on time calibration, and relates to the technical field of sensors. The distributed multi-target fusion tracking method based on time calibration comprises the following steps: on the basis of a generalized covariance cross fusion criterion, constructing a posterior probability density expression of a time calibration parameter by using a Gaussian mixture approximation technology and a probability hypothesis density filter; selecting a mean value of the Gaussian component with the maximum weight from a plurality of Gaussian components of the posterior probability density expression as an estimated value of the time calibration parameter; and based on a generalized covariance cross fusion criterion, fusing the first multi-target probability density function and the calibrated second multi-target probability density function to obtain a multi-target state estimation value of the tracking target. According to the invention, data from a plurality of sensor nodes can be efficiently and accurately fused in a distributed multi-target tracking scene so as to realize accurate positioning of a tracking target.

The invention discloses a short-term thermal load prediction method and system based on time-frequency interaction, and belongs to the technical field of heat supply load prediction, and the method comprises the steps: obtaining thermal load sequence data of a heat supply system, and carrying out the normalization of the data; decomposing the normalized thermal load sequence data into a plurality of IMF components by using a variational mode decomposition algorithm; constructing a time-frequency interaction prediction model; inputting each IMF component into a time-frequency interaction prediction model, that is, each IMF component correspondingly obtains a multi-scale reconstruction feature and a bidirectional time sequence feature through a wavelet convolution model and a BiLSTM model; a time sequence context dynamic retrieval mechanism is established, and joint representation features corresponding to IMF components are obtained; and the combined representation features of all the IMF components are summed after passing through a full connection layer and reverse normalization in sequence, and a final heat load prediction value of the heat supply system is obtained. According to the invention, the thermal load prediction accuracy and robustness of the heat supply system can be improved.

The application discloses a kind of deep fake video detection methods based on multidimensional feature collaborative modeling, belong to computer vision field, including the following steps: using face recognition technology to the complete face image is preprocessed, obtain the image of face local area;Respectively extract the frequency domain feature and spatial domain feature of face local area;Adopt position perception dual-domain fusion module to the extracted spatial domain feature and frequency domain feature are adaptively weighted modulation and nonlinear fusion;Adjust the contribution degree of local area in local-global feature fusion process, obtain the final global fusion feature;Automatic weighting balances the loss of each task, realizes the prediction classification to fake video.The application has adopted the multidimensional feature fusion architecture of local and global, spatial domain and frequency domain, effectively improves the detection precision of deep fake video.

The application relates to the computer technical field, and in particular to a billiards interaction method and device based on AR and real-time trajectory analysis, which comprises the following steps: collecting global image data of a billiards table and spatial position data of a billiards table surface in real time through a camera and a laser radar, and performing noise reduction, format conversion and time axis alignment processing on the collected data; based on the processed real-time data, combining a preset physical model and an LSTM neural network, predicting a main ball trajectory, a target ball collision path and a hole entry probability, and generating an optimal ball hitting route adapted to the skill level of a player; based on the optimal ball hitting route, generating an AR guide line, projecting the AR guide line to the real billiards table surface, and adjusting the projection coordinates of the AR guide line; and outputting a multi-modal interaction feedback according to the mode selected by the player, real-time ball hitting operation behavior and ball movement results. The application helps to realize high-precision trajectory prediction and multi-modal interaction feedback.

The application discloses a kind of agricultural digital twin construction method and system based on data fusion, it is related to agricultural platform technical field.The application constructs farmland three-dimensional grid data model and initializes environment benchmark data;Constructing farmer digital business entity;According to agricultural operation instruction, the farmer skill weight parameter is mapped to the corresponding farmland grid, and the farmland process state data that fuses environment benchmark and operation weight is generated;Respond to the harvesting event of agricultural product, extract the corresponding grid data set from farmland process state data, and instantiate to generate agricultural product digital object;Finally, farmland model, farmer entity and agricultural product object are encapsulated by data inheritance relationship, and agricultural digital twin is output.It realizes the fusion and two-way feedback of farmland, farmer and agricultural product, achieves the whole link accurate traceability from production to product, solves the problem of isolated modeling of each element of agricultural digital twin in the prior art, lack of data correlation and feedback mechanism.

The invention discloses an intelligent travel guide method and system based on multi-modal fusion, and relates to the field of artificial intelligence. The method comprises the following steps: acquiring and inputting text modal data into a lightweight intention recognition model to obtain an intention probability vector; obtaining the intention probability vector and the intention modal mapping matrix, performing weight calculation processing according to the intention probability vector and the intention modal mapping matrix to obtain a query intention weight, a recognition intention weight and a planning intention weight, and performing correction normalization processing on the query intention weight, the recognition intention weight and the planning intention weight to obtain a final fusion weight; performing feature extraction on the voice modal data, the visual modal data and the text modal data, and performing linear weighted fusion processing on the voice features, the visual features and the text features according to the final fusion weight; and inputting the final fusion vector and the intention probability vector into a tourism task model to obtain a tour guide result. According to the method, adaptive fusion is used as a nerve to enable multiple kinds of perception to work cooperatively, so that the intention of the user can be identified more accurately.

This invention relates to the field of data processing, and more particularly to a human-computer interaction method for an intelligent simulated baby robot based on user behavior pattern recognition. The method includes: real-time acquisition and preprocessing of user behavior data; extraction of preliminary feature data from the preprocessed behavior data and dimensionality reduction processing to obtain dimensionality-reduced feature data; intelligent fusion processing of the dimensionality-reduced feature data using an adaptive spatiotemporal feedback embedded multidimensional data fusion algorithm to obtain comprehensive behavior data and identify the user's current needs; prediction of user behavior based on the comprehensive behavior data; merging the user's current needs with the predicted user behavior results to obtain a comprehensive demand result; and calculation of the optimal response delay time based on the comprehensive demand result to realize the interaction between the intelligent simulated baby robot and the user. This method solves the technical problem of inaccurate and incomplete processing of user behavior data by traditional robots, leading to poor adaptability and stability in interaction with the simulated baby robot.

The application relates to the technical field of electric correlation parameter measurement, and provides a booster station high-voltage equipment temperature field online monitoring method and system. The method comprises the following steps: investigating high-voltage power supply equipment in a booster station, and determining the installation positions of multiple sets of infrared dual-view thermal imaging devices; starting the multiple sets of infrared dual-view thermal imaging devices, synchronously collecting visible light images and infrared thermal images of the high-voltage power supply equipment in respective monitoring areas, fusing and displaying the infrared thermal images and the visible light images in the same field of view to obtain dual-view images; acquiring key point temperature values and temperature field distributions of the high-voltage power supply equipment in the respective monitoring areas in real time and online through the infrared dual-view thermal imaging devices, monitoring the collection areas corresponding to the temperature field distributions in real time, tracking and displaying the highest temperature value, the lowest temperature value and the average temperature value in the collection areas, and realizing online and real-time monitoring of the temperature field of the high-voltage equipment in the booster station.

The invention relates to the technical field of image recognition, and discloses a coal conveying belt foreign matter detection method and system based on a visible light image, and the method comprises the steps: carrying out the smooth denoising of the visible light image of a coal conveying belt operation region, obtaining a standard image, carrying out the multi-stage enhancement, obtaining a contour enhancement and texture suppression image, and carrying out the fusion of a fusion feature; semantic segmentation is carried out on the standard image to obtain a segmentation mask, a three-dimensional point cloud is obtained through binocular vision, and statistical projection is carried out to obtain a region surface normal distribution diagram; coupling and evaluating the fusion features, the mask shape features and the geometric features of the normal distribution map, and generating a comprehensive foreign matter detection report of the coal conveying belt operation area; the efficiency of detecting the foreign matters on the coal conveying belt based on the visible light image can be improved.

This invention relates to the field of artificial intelligence technology and discloses an AI-driven multimodal data analysis method and system. The system includes a data acquisition module for setting up acquisition nodes and terminals, acquiring multimodal data and device operating parameters in real time, and converting the formats; a multimodal fusion module for constructing a cross-modal semantic association model, processing spatiotemporal characteristics, fusing representational features, and associating simulations; a feature extraction module for constructing image, text, and speech-related extraction algorithms to obtain corresponding features; a comprehensive analysis module for standardizing features, performing association calculations, and comparing them with preset thresholds to analyze semantic consistency; and a hierarchical processing module for combining contextual calculations and analysis when semantic inconsistencies occur, performing secondary comparisons to analyze the target scene performance. This system and method can effectively process multimodal data, improve the accuracy and efficiency of semantic analysis, and are applicable to various scenarios.

The invention discloses a vehicle-road collaborative perception method and system based on feature quantitative compression and semantic segmentation guidance, and belongs to the technical field of intelligent traffic systems. Importing the shallow high-resolution features into a feature compression branch to execute dimension simplification and vehicle-road communication adaptive low-bit non-uniform quantization coding to generate compression features, and importing the deep high-semantic features into a semantic segmentation branch to generate scene semantic masks including vehicle, pedestrian and road categories to generate semantic masks; the vehicle-mounted terminal decompresses the received compression features to obtain decompression features; and at the vehicle-mounted terminal, pixel-level dynamic weight matching fusion is carried out on the decompression features and the semantic masks, enhanced roadside sensing features are generated, the sensing features acquired by the vehicle terminal are fused, 3D target detection is executed, and an accurate detection result is output. According to the method, the vehicle-road communication data volume can be effectively reduced, the perception robustness of the shielded and beyond-visual-range target is improved, and engineering cooperation of the communication efficiency and the detection precision is realized.

The application discloses a rotary drum sail lift and drag load intelligent prediction model based on multi-source data fusion, and particularly relates to the field of fluid mechanics and flow control. The model adopts a seven-layer architecture of "input layer->multi-source data preprocessing module->multi-branch feature coding module->cross-modal attention fusion module->physical information constraint module->spatiotemporal feature extraction module->output layer". The end-to-end hybrid intelligent prediction model is designed for the strong non-uniform flow characteristics of the rotary drum sail. The core idea is "data-driven as the main, physical constraint as the auxiliary, and multi-source data dynamic fusion". The benchmark data set after numerical simulation calibration and the experimental measurement data set are fused to realize the deep fusion of multi-source data, and meanwhile, the physical rationality of the prediction result and the full-working-condition generalization ability are ensured.

The application discloses a probiotic immune regulation function prediction system based on cross attention fusion, relates to the technical field of bioinformatics and artificial intelligence, and changes the integration mode of multi-modal data through a cross attention fusion module, dynamically establishes deep bidirectional correlation between features through two cascaded cross attention sublayers, the first sublayer establishes a connection between metabolome features and a genome, focuses on gene information most relevant to a current metabolic state, and the second sublayer reversely inquires about metabolome context that has been corrected by gene information, the bidirectional and iterative attention mechanism simulates network relationships that genes encode proteins, proteins catalyze metabolites, and metabolites feedback regulate gene expression, a gating fusion unit adaptively learns contribution weights of two information flows, so that the model can capture key gene-metabolite synergistic pathways driving immune regulation functions, and lays a solid foundation for high-precision prediction.

The invention relates to a three-dimensional model and DEM fusion method and device based on multi-source image matching, and the method can effectively solve the problems of displacement, floating or subsidence during the data superposition of a three-dimensional model and a three-dimensional earth platform, and achieves the fusion of a three-dimensional model and a DEM through the matching of the pixel points of a satellite image and an aerial image in a three-dimensional geographic engine. A digital elevation model is utilized to calculate accurate three-dimensional geodetic coordinates and elevation values, a high-precision image control point file is provided for three-dimensional modeling software, the image control point file enables the generated three-dimensional model to have high absolute geographic coordinate precision, and therefore it is guaranteed that the three-dimensional model is accurately fused with existing geographic data in a three-dimensional geographic engine. According to the method, the geographic space accuracy of the three-dimensional model can be remarkably improved while the high detail expressive force of the aerial image is kept, and the method is particularly suitable for an application scene where a local fine model and a large-range geographic background need to be precisely integrated.