37 results about "Membrane computing" patented technology

The invention provides a storage cluster fault diagnosis method and system based on a spiking neural P calculation model, and the method comprises the steps: collecting operation indexes containing static data and dynamic data from a monitoring system of a storage cluster, defining a neuron for each operation index after preprocessing through the construction and training of the spiking neural P calculation model, and carrying out the fault diagnosis of the storage cluster. Initializing an initial state value of each neuron according to the collected data; and performing state updating and reasoning by using pulse signal propagation and fuzzy reasoning through a trained spiking neural membrane calculation model, judging a fault according to a state value of a neuron, and when the state value of a certain neuron exceeds a set fault threshold value, considering that an index has a fault or a potential fault, and identifying a specific fault type in the storage cluster through state combination of the plurality of neurons. According to the invention, efficient fault diagnosis of the distributed storage system is realized.

The invention provides a verification code verification method and device, equipment, a medium and a program product, and aims to solve the problem that an automatic script in the related technology cannot realize automatic testing of verification of a verification code of a rotary sliding block. The method comprises the steps of performing separation processing on a rotating main body and a target pattern in a verification code image to obtain a main body binary mask corresponding to the rotating main body and a target binary mask corresponding to the target pattern; based on the main body binary mask and the target binary mask, a target angle difference between the rotating main body and the target pattern is calculated, and the target angle difference is used for indicating the rotating main body to rotate to a rotating angle matched with the target pattern; generating a rotation track based on the target angle difference by using a slow motion function, wherein the rotation track comprises random disturbance and a random pause point; and controlling the rotating main body to rotate based on the rotating track to obtain a verification result. According to the invention, automatic testing of verification of the rotary sliding block verification code can be realized.

The invention provides a wild animal buffer zone delimiting method and system for a habitat, and relates to the technical field of ecological environment division, and the method comprises the steps: obtaining habitat information data; processing the habitat information data based on a quantum hybrid optimization algorithm and a preset multi-fractal coupling model, and generating a habitat suitability distribution mode; performing multi-scale landscape connectivity analysis according to the suitability distribution mode to obtain a connectivity topological graph of the habitat space; performing membrane calculation ecological network model construction and coupling on the connectivity topological graph to obtain an initial wildlife buffer zone delimiting scheme; generating an optimized buffer band delimiting scheme through a random programming and preset robust optimization hybrid strategy; and q-order fuzzy measure and Choquet integral are adopted to carry out collaborative evaluation, and a wild animal buffer zone intelligent delimiting scheme is obtained. According to the method, the complex trade-off relationship among multiple targets is objectively quantified, and the optimal balance of the scheme in all aspects is ensured.

The invention relates to a method for grid reconstruction based on a skeleton-driven pipeline net-shaped structure, and the method comprises the steps: extracting a skeleton from a point cloud of the pipeline net-shaped structure based on an improved Laplace shrinkage algorithm, and obtaining a skeleton point set and a skeleton curve; calculating the two-dimensional radius of a skeleton point on an image plane by using a segmentation mask obtained by fusing the multi-view image and the point cloud, and then mapping the two-dimensional radius to a real three-dimensional radius by using camera parameters in combination with the depth information of the skeleton point; generating a surface model of the pipeline net-shaped structure according to the skeleton of the pipeline net-shaped structure and the real three-dimensional radius information; the improved Laplacian contraction algorithm specifically comprises the following steps: sensing local density distribution characteristics of point cloud data in real time, so that the Laplacian contraction weight is dynamically inversely proportional to the local point cloud density. Compared with the prior art, the method has the advantages of improving the integrity and accuracy of the three-dimensional reconstruction model of the pipeline network structure and the like.

The invention requests to protect an automatic laboratory mouse wound area tracking method and system based on image recognition, images of a wound area are regularly acquired through a multispectral imaging device, parameters such as ambient light and shooting distance are synchronously recorded in the acquisition process to ensure data consistency, the images are preprocessed by using a distributed computing system, and the data consistency is ensured. Inputting the preprocessed image into a deep segmentation model trained based on a large-scale data set to accurately identify and output a binarized contour mask of a wound region, calculating wound geometric features based on the mask, and dynamically generating a quantitative report including an area change rate and a healing trend coefficient; all time sequence data are stored in a time sequence database and used for constructing a complete wound healing track map, and visualization and trend analysis of the healing process are supported. According to the invention, high-flux, automatic and quantitative monitoring of the wound healing process of the laboratory mouse is realized, and objectivity and analysis efficiency of animal experiment data are remarkably improved.

The application discloses a kind of building contour regularization-oriented polygon generation methods for remote sensing, it is related to remote sensing image recognition technical field, including: through the image instance segmentation model of training completion, real-time high-resolution remote sensing image is carried out sliding window prediction, generates candidate building instance mask, and according to candidate building instance mask calculates center point coordinate and mask area;According to mask area, generate adaptive center point distance threshold, and combine center point coordinate and mask area to exclude duplicate candidate building instance mask;Obtain elevation data that is spatially aligned with real-time high-resolution remote sensing image, carry out spatial superposition calculation to non-duplicate candidate building instance mask and elevation data, obtain elevation statistical characteristics;The application realizes the deduplication processing of candidate building instance mask, and under the unified spatial reference, combine elevation data to complete building floor value calculation and regularized building polygon generation.

A device may receive, from a robotic device, time-of-flight sensor data that includes a point cloud derived from depth images of an environment with a floor and one or more obstacles, and may shift the depth images by a number of pixels to generate shifted depth images. The device may subtract the shifted depth images from the depth images to generate final images, and may calculate Euclidean distances associated with the final images. The device may generate masks for the final images associated with Euclidean distances that are less than a search radius, and may calculate a covariance matrix based on the masks. The device may calculate eigenvectors for the covariance matrix, and may generate an occupancy grid for the environment based on the eigenvectors for the covariance matrix.

The invention discloses a method and a system for determining phenotypic parameters of soybean pods. The method comprises the following steps: constructing a synthetic data set based on prior knowledge, and generating a branch-pod automatic combination and a label for the synthetic data set; establishing a pod phenotype extraction model on the basis of an improved Swin Transform backbone network and a non-modal segmentation network structure, and performing soybean pod non-modal segmentation on the basis of the pod phenotype extraction model to obtain a non-modal mask; the accurate phenotype is automatically extracted on the basis of a non-modal mask, the pod number is calculated, soybean pod phenotype parameters are determined, and the soybean pod phenotype parameters comprise pod length, pod width and area parameters calculated on the basis of a complete mask. The invention further discloses a corresponding system, electronic equipment and a computer readable storage medium.

This invention discloses a knowledge distillation method for remote sensing target detection based on feature-separated attention, comprising: extracting feature attention maps from the feature maps output by the teacher network and the student network respectively; separating the foreground and background regions of the feature maps, and calculating a foreground attention mask using the feature attention map of the teacher network and a background attention mask using the feature attention map of the student network; calculating the L2 distillation loss using the foreground and background attention masks; and transferring the knowledge from the teacher network to the student network based on the L2 distillation loss. This invention can effectively select the region to be distilled, improve distillation efficiency, and improve the detection accuracy of the final lightweight target detection network without changing the student network structure or increasing computational cost.

The invention discloses a proton exchange membrane fuel cell model parameter identification method based on a membrane calculation dynamic membrane structure, and belongs to the field of fuel cell optimization. The method comprises the following steps: 1) establishing a PEMFC mathematical model; 2) taking the minimum absolute error integral of a stack output voltage experiment value and a model value as a target optimization parameter; 3) combining a differential evolution operator with exchange, selection and termination rules of membrane calculation to construct a dynamic membrane calculation model (DMC); 4) optimizing PEMFC parameters by using DMC, and calculating an output voltage error sum of squares; 5) obtaining an optimal parameter combination enabling the objective function to be minimum through DMC iterative search, and comparing and verifying with a classical algorithm; and 6) applying the optimized parameters to an actual PEMFC model. Through the dynamic membrane calculation optimization strategy, the parameter identification precision is remarkably improved, and the model output voltage is highly matched with experimental data.

The invention provides an inclined electric power signboard identification method and system based on deep learning. The method comprises the following steps: taking a to-be-detected electric power signboard image as an input image; carrying out multi-direction and multi-scale filtering to extract key points so as to generate a space mask; extracting multi-scale features from an input image, dynamically weighting and fusing the multi-scale features and associating a spatial mask, and generating a binary segmentation mask; calculating a mass center position of the binarized segmentation mask as an original point position of polar coordinates, performing polar coordinate transformation on the input image, performing multi-direction multi-scale filtering on the obtained polar coordinate domain image, calculating a multi-scale feature descriptor, and further obtaining an RMLG descriptor; and carrying out tilt correction on the input image based on the centroid position and the multi-scale feature descriptor, and carrying out character prediction on the image subjected to tilt correction in combination with the RMLG descriptor to obtain a character prediction sequence corresponding to the input image. According to the invention, identification of the inclined power equipment signboard in a complex scene can be realized.

This invention discloses an intelligent evaluation method and system for the teaching process based on learning behavior analysis, belonging to the field of intelligent evaluation technology for the teaching process. The method includes: acquiring classroom videos to extract student eye region images and establishing a mapping between time blocks and teaching segments; extracting specular and dark nucleus proxy masks and calculating the symmetrical pupil-occluded specular arc index; calculating the evidence reliability weight and eye region visibility evidence value accordingly; constructing a classroom knowledge graph containing evidence and artifact nodes, and writing the above parameters into the graph; retrieving the evidence set of the target segment, using the reliability weight to weight and aggregate the visibility evidence value to obtain an attention score and generate a traceability link. This invention effectively suppresses polarization screen reflection interference by quantifying physical artifacts and utilizing graph aggregation, achieving interpretability and traceability of the evaluation results.

The invention discloses a photovoltaic cell high-precision modeling method fused with a leader mechanism P system. The method comprises the following specific steps: 1) collecting actual working current and working voltage of photovoltaic cell operation; 2) analyzing a photovoltaic cell principle, establishing a photovoltaic cell model, and taking a root-mean-square error of estimated output and actual output of the model as a target function; 3) combining the leader mechanism of the chimpanzee optimization algorithm with a membrane structure, a rewriting rule, a selection rule, a communication rule and an object set of a standard P system (membrane calculation model) to form a P system optimization algorithm (LPOA) fused with the leader mechanism; 4) minimizing the objective function through an LPOA optimization algorithm, identifying unknown parameters of the photovoltaic cell model, obtaining optimal parameters and forming a mathematical model; 5) performing current and power data fitting according to the obtained optimal parameters and models; and 6) applying other algorithms to the photovoltaic cell model to obtain an optimal parameter combination and an optimal error value, and comparing the optimal parameter combination and the optimal error value with the LPOA.

The invention discloses an automatic component segmentation and identification method suitable for a bridge point cloud, and the method comprises the steps: extracting a multi-view visible three-dimensional grid from the bridge point cloud, rendering the multi-view visible three-dimensional grid into a two-dimensional video frame, and achieving the dimension reduction of the three-dimensional point cloud; obtaining a segmentation mask of each component under each visual angle by using the visual large models SAM and SAM2; projecting segmentation results under diagonal view angles, matching the segmentation results of the same members on the two sides based on a segmentation member clustering center nearest pairing algorithm, calculating scores of index points according to segmentation masks of all view angles, and realizing member-level segmentation of the bridge point cloud; a segmentation result is optimized based on an error point iterative repair algorithm; and realizing the classification and identification of the bridge components by using the multi-modal large model CLIP. According to the method, only the point cloud data of the bridge point cloud needs to be collected for analysis processing, component-level segmentation of the bridge point cloud can be rapidly and automatically achieved, component categories are classified based on the segmentation result, and automatic instance segmentation of the bridge point cloud is achieved.

This application provides a mask layout correction method, system, medium, program product, and terminal based on a combination of membrane computing and the DBSCAN algorithm. The method involves obtaining an initial layout, performing preliminary clustering on the initial layout using a preset clustering method to obtain several partitions and their corresponding density and contour features; obtaining initial density and contour parameters for DBSCAN clustering; iteratively optimizing the initial clustering parameters of the DBSCAN clustering using a preset membrane computing algorithm based on the density and contour features of each partition to obtain the optimal clustering parameters for each partition; performing DBSCAN clustering on each partition based on the corresponding optimal clustering parameters to obtain the core region, boundary region, and sparse region of the initial layout; and correcting the core region, boundary region, and sparse region based on a preset correction strategy to obtain the corrected layout of the initial layout. This improves the accuracy of mask layout correction.

The invention discloses a feature screening method, a feature screening device and a feature screening system. The feature screening method comprises the steps of performing feature extraction on a target image through a deep learning model to obtain a target feature; for each prompt image, performing feature extraction on the prompt image through a deep learning model to obtain a first prompt feature; calculating cosine similarity based on the target feature, the first prompt feature and the corresponding prompt mask; calculating classification scores based on the target features, the first prompt features and the corresponding prompt masks through a deep learning model; determining an attention weight according to the cosine similarity and the classification score to screen the first prompt feature to obtain a second prompt feature; and performing feature fusion on the plurality of second prompt features corresponding to the plurality of prompt images to obtain a fused prompt feature. Therefore, extra reasoning duration and reasoning video memory are not increased, redundant prompt feature information can be eliminated, and a high-precision segmentation result is kept.

The application provides a FY4B remote sensing single-channel super-resolution reconstruction method based on ESRGAN, which comprises the following steps: analyzing, uniformly cutting, scale constraining and quality detecting the resolution data of FY4B and FY3D to obtain pretreatment data; generating a maximum effective pixel intersection mask according to the pretreatment data; calculating the effective pixel ratio and performing patch screening according to the maximum effective pixel intersection mask to obtain training data; constructing an ESRGAN network structure and performing data reconstruction on the training data to obtain a predicted image. Through the establishment of paired samples of the same day, the uniform cutting alignment and scale constraint, the patch screening and mask loss calculation based on the maximum intersection of effective pixels, and the reconstruction strategy combining pixels, high frequencies and adversarial learning, the FY4B single-channel data is realized from low resolution to high resolution in detail enhancement and structure reconstruction.

The application claims an image recognition-based automatic tracking method and system for wound area of experimental mice, which regularly collects images of the wound area through a multispectral imaging device, and synchronously records parameters such as environmental light and shooting distance during the collection process to ensure data consistency, pre-processes the images using a distributed computing system, inputs the pre-processed images into a deep segmentation model trained based on a large data set to accurately identify and output a binary contour mask of the wound area, calculates wound geometric features based on the mask, dynamically generates a quantitative report including an area change rate and a healing trend coefficient, and stores all time series data in a time series database to construct a complete wound healing trajectory atlas, and supports visualization and trend analysis of the healing process. The application realizes high-throughput and automatic quantitative monitoring of the wound healing process of experimental mice, and significantly improves the objectivity and analysis efficiency of animal experiment data.

The invention relates to the technical field of integrated circuits, and discloses a frequency domain segmentation collaborative gradient driven inverse lithography method, system and device and a medium, the method comprises the following steps: obtaining a target layout image of an integrated circuit, initializing the target layout image by using initialization parameters, and carrying out multi-scale feature weighted summation to obtain a multi-scale fusion mask; performing frequency domain mapping on the multi-scale fusion mask to obtain a frequency domain component, generating a candidate region according to the frequency domain component, and performing low-frequency fusion to obtain an optimal candidate region; smoothing and mapping the optimal candidate area to obtain a mapping mask, and calculating wafer images corresponding to the mapping mask in the maximum process window and the minimum process window; calculating a loss function according to the wafer image, and performing gradient optimization on the initialization parameter according to the loss function to obtain an optimized mask parameter; and performing iterative optimization on the optimized mask parameters to obtain optimal mask parameters, and calculating an optimal mask image by using the optimal mask parameters.

The application discloses an ore crushing control method and system, electronic equipment and a readable storage medium. The method inputs a multi-scale feature map fused into a feature mapping layer to obtain a multi-scale fusion feature tensor; inputs the multi-scale fusion feature tensor into a candidate region generation layer to determine a target candidate region; inputs the target candidate region into an instance segmentation layer to determine a target instance segmentation mask; calculates a particle equivalent diameter based on the target instance segmentation mask; determines a medium crushing discharge particle size and a fine crushing discharge particle size under a P80 particle size index according to the particle equivalent diameter; calculates a material instantaneous conveying flow based on a continuous time sequence point cloud volume integration model; performs time integration on the material instantaneous conveying flow to obtain a fine crushing discharge throughput; and performs hierarchical control on ore crushing through the medium crushing discharge particle size, the fine crushing discharge particle size, the fine crushing discharge throughput and a fine crushing machine running power. The application can improve the ore crushing efficiency and the qualified rate of the crushing products.

The invention belongs to the technical field of remote sensing data intelligent classification, and discloses a remote sensing image classification method based on membrane calculation, PSO particle swarm and BP neural network, which uses an organization type P system in membrane calculation as a calculation framework, and uses a speed-displacement model of the PSO particle swarm algorithm as an evolution rule of an object in a basic membrane. Meanwhile, sharing and co-evolution of information between objects are achieved through a transfer rule of a membrane, finally, a global optimal object is stored in the environment, then the global optimal object obtained in the environment serves as input of the optimized BP neural network, finally, the BP neural network is trained, and the optimal object is obtained. And inputting remote sensing image data into the trained BP neural network to obtain a final remote sensing image classification result. The remote sensing image classification efficiency and precision are improved, and rapid classification of the remote sensing images is realized.

The invention discloses a mechanical overall design method for a multifunctional intelligent control LED street lamp, and the method comprises the following steps: S1, dividing an LED street lamp into a plurality of functional modules, and forming a physical structure configuration result of the functional modules; s2, generating a membrane structure mapping model based on a physical structure configuration result of the functional module; s3, forming an intra-membrane control rule configuration set based on the membrane structure mapping model; s4, collecting external environment data, and generating a perception input data set; s5, inputting the sensing input data set into the environment sensing membrane unit to form a state object set; s6, inputting the state object set into the functional membrane unit, and generating a control behavior instruction set; s7, setting membrane structure expansion conditions, and generating an updated membrane structure mapping model; and S8, loading the membrane structure mapping model to an edge control unit, and executing reasoning and control instruction issuing. According to the invention, a membrane calculation structure mapping and rule reasoning mechanism is adopted to realize multifunctional intelligent control of the LED street lamp.

The invention relates to a disordered grabbing method based on deep learning and 3D point cloud, and the method comprises the steps: processing a gray image through a deep learning instance segmentation model, and obtaining an accurate pixel-level mask of each to-be-grabbed object; a pixel-level mask is mapped to the 3D point cloud data, and an independent 3D point cloud of each object to be grabbed is segmented; based on the optimized object point cloud, the grabbing pose of each to-be-grabbed object is calculated, and the grabbing pose comprises a grabbing target point and a grabbing pose vector; and on the basis of the grabbing poses and the pixel-level masks, the space shielding relation between the to-be-grabbed objects is calculated, the grabbing sequence of the robot is planned according to the space shielding relation, and the objects which are higher in position and free of shielding are preferentially selected according to the grabbing sequence. According to the method, the modern sensor technology and a data processing algorithm are combined, accurate calculation of the grabbing pose is achieved, and the disordered grabbing quality and efficiency are remarkably improved.

The invention relates to the crossing field of grain storage intelligent monitoring and digital twinning technologies, and provides a grain storage multi-field coupling digital twinning-oriented temperature field prediction method and system and a storage medium, and the method comprises the following steps: S1, modeling and discretization: building a two-dimensional granary section under a unified geometric and regular grid; s2, performing multi-working-condition simulation and unified analysis rearrangement, and performing transient simulation on multi-physical-property and multi-boundary-parameter working conditions by adopting CFD (Computational Fluid Dynamics); and S3, geometric priori construction: generating a fluid domain mask and a grain pile domain mask, and calculating a signed distance field as geometric priori. Scalar conditions such as external temperature, time phase, boundary parameters and the like are mapped into channel-level scaling-offset operators through affine transformation, fine-grained regulation and control of a feature space are achieved, compared with a simple splicing method, the boundary neighborhood prediction precision is improved, the self-adaptive influence radius and assimilation intensity are designed for double domains, and the method has the advantages of being high in adaptability and high in adaptability. The boundary layer temperature correction precision is improved, and the observation information utilization efficiency is improved.

This invention discloses a method and system for segmenting welded areas based on semantic segmentation networks and depth geometric constraints, belonging to the field of welded area extraction technology. The method includes the following steps: acquiring RGB and depth images of the workpiece to be welded; acquiring rectangular detection boxes and a preliminary mask for marking the welded areas; constructing an energy model; using the energy model to perform global pixel-level optimization on the preliminary mask to determine the mask boundary, obtaining an optimized mask; calculating the gradient field, normal direction, and local curvature of the depth image in the boundary neighborhood, interpolating the depth variation curve along the normal direction to obtain an interpolation curve; and using the interpolation curve to perform depth interpolation processing on the optimized mask to obtain the three-dimensional boundary information of the weld. The method provided by this invention simultaneously considers feature representation, geometric alignment, and engineering usability, realizing a complete boundary inference process from coarse network segmentation to fine depth geometric localization.

The application discloses a wavefront coding microscopic system depth of field expansion joint optimization method based on deep learning, and belongs to the field of microscopic medical imaging. The application comprises the following steps: collecting clear focus images for neural network model training and testing; calculating defocus phases according to corresponding optical system parameters, and simulating defocus blur of the clear focus images; using a learnable aberration correction mask matrix, updating data to correct defocus aberration in the training process; calculating a joint point spread function through the defocus correction mask to generate a defocus correction image; using the defocus correction image and the clear focus image pair, and a point spread function quality and image quality evaluation loss function to train a generative model to generate a high-quality optical image. The application optimizes a mask free surface type, introduces an optical physics mechanism to fuse an image similarity evaluation function to form a model gradient back transmission loss function, ensures the effectiveness of wavefront coding aberration correction, and improves the imaging quality and robustness.

The application provides a habitat wild animal buffer zone delineation method and system, relates to the technical field of ecological environment division, and comprises the following steps: acquiring habitat information data; processing the habitat information data based on a quantum hybrid optimization algorithm and a preset multi-fractal coupling model to generate a habitat suitability distribution pattern; performing multi-scale landscape connectivity analysis according to the suitability distribution pattern to obtain a connectivity topology atlas of the habitat space; constructing and coupling a membrane calculation ecological network model for the connectivity topology atlas to obtain an initial wild animal buffer zone delineation scheme; generating an optimized buffer zone delineation scheme through a random programming and a preset robust optimization hybrid strategy; and performing collaborative evaluation by using a q-order fuzzy measure and a Choquet integral to obtain an intelligent wild animal buffer zone delineation scheme. The application objectively quantifies the complex trade-off relationship among multiple targets and ensures the optimal balance of the scheme in all aspects.

The application discloses a kind of component automated segmentation identification method suitable for bridge point cloud, the method includes: extracting multi-view visible three-dimensional grid to bridge point cloud and rendering as two-dimensional video frame, realize the dimensionality reduction of three-dimensional point cloud;Using visual large model SAM and SAM2 obtains each component segmentation mask under each view angle;The segmentation result under diagonal view angle is projected, based on the nearest pairing algorithm of segmentation component cluster center, the segmentation result of same component on two sides is matched, according to the score of index point calculated from each view angle segmentation mask, realize the component level segmentation of bridge point cloud;Based on error point iterative repair algorithm, the segmentation result is optimized;Using multimodal large model CLIP realizes the classification identification of bridge component.The application only needs to collect the point cloud data of bridge point cloud for analysis and processing, can quickly and automatically realize the component level segmentation of bridge point cloud, based on the classification of component class based on segmentation result, realize the automated instance segmentation of bridge point cloud.

The application provides a method, device and storage medium for table structure recognition of three-line tables, and the method comprises the following steps: preprocessing a picture to obtain a binary image, performing horizontal erosion and horizontal expansion on the binary image after inversion to obtain a horizontal table line image; performing pixel logical OR on the binary image and the horizontal table line image to obtain a table line free image, performing horizontal erosion and vertical erosion on the table line free image respectively, and merging the results to obtain an initial cell mask; traversing the horizontal table line, re-segmenting the cell to obtain a final cell mask, calculating and obtaining information of all cells in the final cell mask, and obtaining a table structure representing the three-line table; the application improves the accuracy of three-line table structure recognition of various types, effectively reduces the problems of cell missed detection and multiple detection, the result contains complete and specific information, and is beneficial to subsequent table structure restoration and the like; a non-data driven method is adopted, a large amount of data is not required, and the calculation amount is small; the table structure of the three-line table can be more accurately and universally recognized.