106 results about "Graphic recognition" patented technology

The invention relates to the technical field of graph recognition, in particular to a ship type recognition method and system based on transfer learning, and the method comprises the following steps: carrying out the combined preprocessing of a visible light ship image and a synthetic aperture radar image, and separating a ship inherent feature layer and an environmental noise feature layer through an adversarial domain decoupling network; generating an interference-removed cross-domain feature matrix; enhancing space-time consistency features related to the ship motion mode, and outputting an enhanced feature matrix after enhancement; and S2, constructing a differentiable decision tree classifier based on the enhanced feature matrix in S2, selecting a category with the highest probability as a final recognition result to be output, comparing the confidence coefficient of the final recognition result with a preset threshold, and if the confidence coefficient is lower than the preset threshold, triggering incremental learning. The adaptive capacity to complex sea conditions, ship type diversity and shielding conditions is improved, the error recognition rate and the attack resisting success rate are remarkably reduced, and the reliability of the system in the open environment is effectively improved.

The invention discloses an AI intelligent image-text situation content accurate layout full-marketing generation method, and particularly relates to the technical field of image-text recognition. The method comprises the following steps: performing semantic analysis on a text input by a user, extracting a deliberate graph item, reverse semantics, an auxiliary emotion item and a tonality keyword, and constructing a multi-semantic field group; the field groups are injected into a collaborative generation engine composed of an image generation sub-model and a semantic prediction sub-model, and the semantic prediction sub-model predicts graph attribute tags and guides the image generation sub-model to conduct regional attention intervention; after the image is generated, the matching degree of the image content and the field group structure is evaluated through a structure-level semantic consistency judgment mechanism, and regeneration is triggered if the image content and the field group structure are inconsistent; finally, the image content with the aligned structure is output and subjected to image-text combination typesetting with the original text, and image-text situation content meeting the multi-platform marketing application requirement is generated. According to the method, the semantic accuracy, the structural consistency and the visual expression quality are remarkably improved while the generation efficiency is ensured.

The invention discloses a graphic processing engine development method and system based on virtual simulation, and relates to the technical field of computer graphics and virtual reality, and the method comprises the steps: collecting three-dimensional point cloud data, carrying out the convolution operation through a convolutional neural network algorithm, extracting a three-dimensional space feature matrix, and constructing a graphic recognition model. Inputting original data of a to-be-processed graph into the graph recognition model, outputting a graph parameter set, matching a virtual scene model in a preset scene library, and adjusting parameters of the model; establishing a physical coordinate system according to the adjusted model, converting a to-be-processed graph into grid units through a grid division algorithm, and calculating displacement data of the grid units; and reading displacement data, generating a graphic processing instruction sequence and sending the graphic processing instruction sequence to a graphic processor, and outputting an imaging result by superposing texture information and a ray tracing effect on the grid units. According to the method, a complete processing chain from data acquisition to rendering output is established, and the automation level and accuracy of graphic processing are improved.

The invention relates to the field of drawing recognition, in particular to a PDF drawing data extraction method and system based on intelligent recognition. Comprising the following steps: reading an internal structure of a PDF engineering drawing to obtain a native text stream, a vector path and a grating image; identifying the native text flow through a shunt preprocessing framework to form structured text data; rendering the vector path and the grating image to obtain a background image; analyzing the structured text data by utilizing the intelligent recognition model through the character recognition and extraction sub-model, obtaining drawing metadata and recording the position, and obtaining a character recognition result; analyzing the background image through a graphic element recognition and classification sub-model, recognizing and classifying component elements, and obtaining a graphic recognition result; and performing fusion according to the visual space corresponding relation to form a drawing analysis result. According to the method, the adaptive capacity of engineering drawings with various sources and different qualities is improved through the shunting preprocessing framework and the intelligent identification model.

The invention relates to the field of damaged vision recovery of artificial retinas, and particularly discloses a preparation method of a neuromorphic visual interface for recovering the pattern recognition capability of a blind animal. The interface is composed of a photo-thermal assembly, a photoelectric assembly and an electrode array. The photo-thermal assembly is dispersed in a continuous medium, can rapidly and transversely transfer a thermal signal, and is used for repairing the rapid extraction process of an animal visual pathway on a graphic contour; the electrode arrays are located on the upper surface and the lower surface of the photoelectric assembly, can vertically transmit electric signals in a delayed mode and correspond to the follow-up pattern detail machining process of a visual path. The interface also reproduces a central-peripheral activation-inhibition effect formed by retinal horizontal / bipolar cell transverse / longitudinal signal transmission, and realizes multi-channel signal connection with a visual center. The technology is mature, processing is simple, an artificial system which is highly matched with animal optic nerve signals and functions is developed, the artificial system is different from existing light magical vision, and recovery of the pattern recognition function is achieved.

The embodiment of the invention provides a chip layout image processing method and device, electronic equipment and a storage medium, and the method comprises the steps: determining a basic structure type corresponding to a target graph according to the line segment position information in an obtained chip layout image; and determining a refined structure type corresponding to the target graph according to the line segment position information and the line segment angle information. And performing rule inspection on the target graph based on standard structure parameters preset for the refined structure type. The defect that only a graph parallel to the transverse axis direction or the longitudinal axis direction of a two-dimensional rectangular plane coordinate system or a graph with a specific bevel edge angle can be recognized at the present stage is effectively overcome and solved, and then the inspection precision and the inspection efficiency of design rule inspection or mask rule inspection are remarkably improved. The practicability and reliability of the rule checking process and the subsequent image adjusting process for the target graph are improved to a certain extent, so that the performance and the yield of the finally produced chip are powerfully guaranteed.

The present invention provides a graphic recognition method and obstacle avoidance system for intelligent agricultural robots, which relates to the technical field of image pattern recognition. By introducing polarization imaging, the present invention effectively suppresses water droplet highlights, avoids misjudging optical artifacts as diseases, and significantly improves recognition accuracy. For low-confidence targets, the original micro-area thermal excitation physical review mechanism can confirm their true physical properties online, solving the problem of recognition uncertainty; in addition, the dynamic risk volume constructed based on multi-dimensional information, combined with the adaptive speed control of path risk foresight, realizes differentiated intelligent obstacle avoidance for rigid and flexible obstacles. While ensuring safety, this method greatly improves the robot's travel efficiency and smoothness in complex environments.

The invention discloses a standard part batch replacement method, device and equipment and a storage medium, and the method comprises the steps: loading a three-dimensional model of a clamp, carrying out the pattern recognition of a replaced standard part, and extracting part features; traversing the clamp three-dimensional model to identify a standard component needing to be replaced, and adding a reasonable interference part identification mark; capturing assembly surface features and executing standard part batch replacement operation according to the matching relationship; traversing the replaced clamp three-dimensional model, and generating a standard part batch replacement report according to the interference condition; and correcting the standard part batch replacement report, and storing the reasonable part interference identification marks in a database. By automatically recognizing the features of the replaced standard parts, screening the standard parts needing to be replaced in the clamp three-dimensional model, conducting batch replacement through feature matching and storing the reasonable interference recognition marks in the database for calling, the workload of manually marking the replaced standard parts and the reasonable interference marks is reduced, the misoperation rate is reduced, and the work efficiency is improved. The standard part batch replacement efficiency is improved.

The invention relates to the field of pattern recognition signal alarm devices, in particular to an intelligent switch cabinet fault signal alarm device and method based on pattern recognition, and the device comprises a temperature detection array which is used for continuously detecting and obtaining the temperature data of a plurality of regions in a switch cabinet; the infrared scanning module is used for starting infrared scanning to confirm a temperature abnormal area when the temperature data is abnormal, and obtaining coordinates of the temperature abnormal area; the photographing module is used for moving to the temperature abnormal area for photographing based on the temperature abnormal area coordinates to obtain an abnormal image; and the alarm module is used for identifying the fault type in the abnormal image to obtain alarm data and uploading the alarm data to the upper computer to give an alarm. According to the invention, the type of the fault of the switch cabinet can be quickly found and positioned by identifying the graph in the switch cabinet, so that maintenance personnel can timely feed back and process the fault, and the working efficiency is improved.

The invention relates to the technical field of computer recognition, in particular to a tactical graph recognition method and system on a football field based on tactical relevancy, and the method comprises the steps: building a coordinate system based on the football field, and collecting the position data of a player and any moving target of a ball in each frame; obtaining spatial distance relevancy, speed acceleration relevancy and historical tactical relevancy to generate a relevancy matrix; a weight matrix is obtained, and the tactical relevancy of any two players is determined in combination with a relevancy matrix; constructing a tactical correlation graph, defining the tactical correlation degree as the weight of a corresponding edge in the tactical correlation graph, selecting any vertex in the edge with the maximum weight as a starting point to obtain a plurality of polygons, screening, and combining the screened polygons to obtain a tactical graph of any frame; the spatial distance relevancy, the speed acceleration relevancy and the historical tactical relevancy of every two players are analyzed, the tactical relevancy of the players is comprehensively determined and visually presented in a tactical graph, and the accuracy is higher.

The invention discloses an early diagnosis and treatment system for children with autism based on facial representation multi-modal fusion. A system collects a face video when a child watches a specific video (7 minutes, 08 seconds), first 2000 frames of images are extracted as input, expression data are extracted by adopting an improved contrast language-image pre-training model, sight line data are obtained in combination with GazeMobileNet, and the expression data are displayed on a display screen. Time sequence feature extraction and classification are carried out through a multi-mode fusion network (including a preliminary memory module, a key extraction module and a hidden learning module) of a multi-valve memory mechanism. In the intervention stage, the system provides intervention games such as whack-a-mole, pattern recognition and the like to train reaction, cognition, coordination and language abilities, and an evaluation report is generated according to training data such as reaction time and the like. The system integrates data acquisition, feature learning, multi-modal classification and intervention training, improves the diagnosis accuracy and the intervention effect, and has the characteristics of simplicity and convenience in operation and high universality.

The invention relates to the field of electric power systems, and discloses a relay protection setting value automatic checking system and method based on artificial intelligence. The method comprises the following steps of: inputting a constant value sheet into a constant value sheet table, carrying out region correction and pre-segmentation processing on the input constant value sheet picture by using an OCR (Optical Character Recognition) pattern recognition technology and a BMHS (Binary Matrix Hazardous Substances) improved algorithm, recognizing characters in the constant value sheet table one by one by using a TableMaster algorithm, and carrying out correlation comparison analysis on a recognition result and a preset dictionary, so as to correct the recognition result. The method comprises the following steps: recombining a corrected identification result into a complete fixed value sheet document, performing item-by-item comparative analysis on the recombined fixed value sheet document and fixed value items in a fixed value sheet published in a system, so as to automatically check the difference problem of the fixed value sheet, generate a check report, and display the check information of the fixed value sheet to a user, thereby improving the user experience. Checking errors can be reduced, and the efficiency and accuracy of constant value checking can be improved.

The invention provides a wafer defect measuring method and related equipment. The wafer defect measuring method comprises the following steps: acquiring wafer images of a wafer at a plurality of positions; identifying a graph in each wafer image through a graph identification model, and obtaining a corresponding positioning graph from the graph of each wafer image; positioning the wafer based on each positioning graph, and obtaining a defect image of the wafer; recognizing a graph in the defect image through a graph recognition model, and obtaining a target graph from the graph of the defect image; and measuring the wafer defect in the defect image based on the target graph. According to the wafer defect measurement method, the accuracy of pattern recognition and the accuracy of defect positioning can be improved, missing detection and false detection are further prevented, and the detection speed and the detection precision can be both considered. In addition, when the pattern recognition model is trained, only the pattern needs to be used for training, the data acquisition cost can be reduced, the data balance is improved, and the interpretability and generalization ability of the model are improved.

This disclosure relates to a hand-drawn graphic recognition method, apparatus, and system, and a computer-readable storage medium. The hand-drawn graphic recognition method includes: receiving an input of a hand-drawn graphic, the input including a stroke set, the stroke set including at least one stroke, each stroke including an information sequence of a plurality of trajectory points, information of each trajectory point including coordinates and writing time of the trajectory point; determining whether the stroke set forms a closed graphic according to coordinate data of the input stroke; and in the case where the stroke set forms a closed graphic, recognizing a category of the closed graphic based on the coordinate data of the stroke.

The present invention relates to the technical field of graphic recognition, and specifically relates to a ship type recognition method and system based on transfer learning, including the following steps: jointly preprocess visible light ship images and synthetic aperture radar images, separate the ship inherent feature layer and the environmental noise feature layer through an adversarial domain decoupling network, and generate a cross-domain feature matrix free of interference; strengthen the spatio-temporal consistency features related to the ship motion mode, and output an enhanced enhanced feature matrix; construct a differentiable decision tree classifier based on the enhanced feature matrix of S2, select the category with the highest probability as the final recognition result for output, compare the confidence level of the final recognition result with a preset threshold, and if it is lower than the preset threshold, trigger incremental learning. The present invention improves the adaptability to complex sea conditions, ship type diversity and occlusion conditions, significantly reduces the misrecognition rate and the success rate of adversarial attacks, and effectively improves the reliability of the system in an open environment.

The invention provides an intelligent graph recognition method and device based on collaborative drawing, equipment and a storage medium. The method is applied to the technical field of computers, and comprises the following steps: acquiring a canvas for collaborative drawing, performing pattern recognition processing on a pattern in the canvas for collaborative drawing, determining a boundary of the pattern, and performing segmentation processing on the pattern according to the boundary of the pattern to obtain at least one pattern and a pattern block corresponding to the pattern; the canvas for collaborative drawing is a drawing result obtained by collaborative drawing of a plurality of users; performing intelligent pattern recognition processing on the pattern block, and determining a visual component corresponding to a pattern in the pattern block; pixel point pickup processing is carried out on the graphic block, and each pixel point of the graphic block is traversed to obtain a pixel color value of the graphic block; and the data board is generated according to the visual component corresponding to the graph in the graph block and the pixel color value of the graph block, so that the identification accuracy is improved, and the development efficiency is improved.

The present invention discloses a graphic recognition method, a device, a storage medium and an electronic device. Among them, the method includes: obtaining a screenshot of a candidate graphic currently drawn on an operation interface and a set of drawing trajectory points of the candidate graphic; determining a first-level graphic category to which the candidate graphic belongs based on the screenshot; identifying a second-level graphic category from sub-categories of the first-level graphic category based on the set of drawing trajectory points; and displaying a target graphic fitted according to the second-level graphic category in the operation interface. The present invention solves the technical problem of low recognition accuracy caused by a single recognition dimension.

The invention relates to a vector structure data generation method and device, equipment, a medium and a program product. The method comprises the steps of obtaining a target house type image; performing pattern recognition processing on the target house type image to obtain pattern recognition data; performing text recognition processing on the target house type image to obtain text recognition data; and generating vector structure data corresponding to the target house type image according to the pattern recognition data and the text recognition data. By adopting the method, the accuracy of the vector structure data can be improved.

The invention provides a drawing processing method and device for a digital ship electrical design platform, and the method comprises the steps: determining a selected target object from a target drawing in response to a first selection operation for the target drawing displayed in a graphical user interface of drawing design software, the target object comprises a target cable segment and target equipment blocks connected with two ends of a target cable; wherein different types of drawings correspond to different identification rules; identifying attribute information of the target object based on a target identification rule which corresponds to a pre-configured pattern identification mode and is matched with the type of the target drawing; and based on the preset table template matched with the type of the target drawing, performing structured processing on the attribute information of the target object, generating a target structured table, and displaying the target structured table in the graphical user interface, thereby realizing specialized recognition of the electrical system diagram in the ship industry, and improving the recognition efficiency and the recognition precision.

The invention relates to the technical field of vehicle-mounted assembly, and provides a vehicle-mounted backlight assembly method and equipment, aiming at assembly materials with normal functions and partial appearance flaws, a virtual tolerance zone is preset to limit a fluctuation range of a product contour, and when the assembly materials which do not conform to an incoming material management and control standard are detected, the virtual tolerance zone is preset to limit the fluctuation range of the product contour. Image optimization is carried out to reduce the defect detection sensitivity, the similarity and the actual tolerance zone are further calculated through pattern recognition to ensure that the optimized product image is closer to the theoretical center, finally, the optimized image is judged again, if yes, fitting assembly is carried out, and if not, image optimization is carried out again. And closed-loop detection is realized, so that normal fitting of assembly materials with normal functions and weak appearance defects is ensured, the yield is improved, and the cost advantage is expanded.

The application discloses a kind of based on multi-task learning's light hand-drawn figure recognition and parameter regression method and system, including hand-drawn track point sequence is preprocessed into standardization binary image, high-level semantic feature is extracted by feature extraction network, and class and parameter are output in parallel by classification branch and multi-parameter regression branch, call special regression subnetwork refinement regression for pentagon, pentacle, finally output shape type and geometric parameter according to JSON protocol.Model uses two-stage training, mask MSE regression is done to polygon, curve, circular class in joint stage, independent standard MSE regression is done to pentagon, pentacle in special stage, combined with weight cross-entropy classification loss and multi-branch selective back propagation strategy, the application realizes light, efficient hand-drawn figure real-time identification and regular, applicable to mobile terminal image editing class application.

The invention relates to the technical field of intelligent desks, and discloses an intelligent desk for guarding health of teenagers, the intelligent desk comprises a supporting device, a display device, a lighting device, a sensor device and a control device, the display device, the lighting device and the control device are all located at the top of the supporting device, and the sensor device is located in the supporting device. The panel camera can scan facial expressions of students, the facial expressions are analyzed through the mainboard CPU, the emotional state of the teenagers and whether the teenagers feel fatigue, tension or anxiety can be evaluated, and the information is very helpful for timely finding and correcting bad habits of the teenagers and providing psychological support. According to the utility model, students' homework can be scanned through the image-text recognition camera, the homework can be checked and answered through the mainboard CPU, so that the learning efficiency of the students can be effectively improved, and the display screen can be externally connected with other learning equipment through the USB connector, so that various learning contents can be provided for the students.

The embodiment of the invention provides a positioning method and device based on multiple DM codes, an autonomous mobile device and a storage medium, and relates to the technical field of mobile robots. The method comprises the steps of obtaining a first layer image, a second layer image and a third layer image obtained by performing continuous downsampling on a to-be-detected image; determining an actual position and an actual angle of at least one DM code in the to-be-detected image according to the first layer image, the second layer image and the third layer image; based on the actual position and the actual angle of the at least one DM code, determining the actual position and the actual angle of a graphic identification code comprising the at least one DM code; and positioning is carried out based on the actual position and the actual angle of the graphic identification code, so that the navigation positioning efficiency, accuracy and robustness based on the multi-DM code are improved.

The invention belongs to the technical field of pattern recognition and garden engineering monitoring, and particularly discloses a garden landscape tree construction progress monitoring and recognition system which comprises a data acquisition layer, a data transmission layer, a system platform layer and an application layer. The data acquisition layer is used for acquiring construction site environment parameters and tree state data in real time; the data transmission layer transmits data of the data acquisition layer to the system platform layer by adopting a network communication technology; the system platform layer is used for data storage, analysis and visualization; the application layer is used for providing mobile terminal application and decision support functions; the construction progress recognition accuracy is improved, the advanced image recognition algorithm and the multi-sensor data fusion technology are adopted, the defects of a traditional color segmentation method are overcome, the construction progress can be recognized more accurately, personal errors are reduced, the tree growth environment is monitored comprehensively, and soil humidity, illumination, temperature and other environmental factors are monitored in real time through multiple sensors.

The invention relates to the technical field of image recognition and pattern recognition, and discloses a fusion tracking method based on AI large model video analysis and RFID positioning, which comprises the following steps: when a target enters a visual non-overlapping blind area, according to a phase gradient change trend of a radio frequency identification signal sequence acquired by a radio frequency signal acquisition unit, acquiring a target image according to a phase gradient change trend of the radio frequency identification signal sequence; generating a probability weight map representing the spatial position of the target; projecting the image to an image coordinate system, and generating an airspace attention weight adjustment mask; according to the method, the mapping relation between the radio-frequency signal phase gradient and the image neuron response weight is established, the spatial topology constraint is introduced in the feature extraction stage, structural contraction of the feature space is achieved, and the target handover is completed by utilizing the mask to modulate the depth feature map and extracting candidate target feature vectors according to the modulated depth feature map. And the target binding stability in the cross-monitoring-area tracking process is effectively enhanced.

The invention discloses a plane random placement image analysis processing system and a processing method thereof, and the system is suitable for a scanning device which is composed of a movable carrying platform, a scanning camera, an adjusting mechanism and a fixing mechanism. Comprising an image acquisition module, an image splicing module, an image preprocessing module, a target segmentation module, a shape recognition module, a pattern distribution module and an output module, and all the modules are called in sequence to form a closed-loop processing flow. The method comprises multiple steps of image acquisition, splicing, preprocessing, segmentation, identification, pattern distribution and result output, image frames are acquired by adopting a snakelike trajectory, and splicing fusion, accurate identification and task output of random image targets are realized in combination with strategies of displacement estimation, multi-scale fusion, brightness compensation, feature matching and the like. The method can be widely applied to scenes such as pattern recognition, pattern matching and batch output, and has the advantages of high automation degree of the processing flow, high recognition accuracy, high target placement randomness and the like.

The present invention is a process and structure of a ferroelectric neural network cell with self-learning capability and enhanced character recognition length. The ferroelectric neural network cell is a content addressable memory (CAM) structure (or FeCAM for short) based on a ferroelectric field-effect transistor (FeFET), and the FeCAM is a 1N1P-FeCAM including an N-type FeFET and a P-type FeFET. Therefore, the present invention reduces the number of devices to minimize cell area, and each cell consists of two complementary FeFETs to achieve a density that is 33% higher than that of Ternary CAM (TCAM). H2 plasma treatment (HPT) is effective in enhancing symmetry between ION1 and ION2 to increase the number of mismatch cells (Max HD, MHD). Therefore, the present invention is useful to graphic recognition and can be applied to various fields of artificial intelligence.

A 3D printing device and its laser positioning accuracy calibration method are disclosed. The calibration method includes: placing a standard plate with several graphics on a working plane; setting an imaging system at the laser incident end of a galvanometer; keeping the laser off and controlling the galvanometer deflection to scan any position A of each graphic; acquiring an image of the standard plate observed under the reflection of the galvanometer at position A through the imaging system; performing image processing on the standard image; extracting the coordinate values ​​of the positioning patterns at position A of the corresponding graphics using a graphic recognition algorithm; forming an actual matrix with the coordinate values ​​of the positioning patterns corresponding to all graphics in the array; calculating the deviation between the coordinate values ​​of each positioning pattern in the actual matrix and the corresponding target coordinate values ​​in the target matrix; and generating a galvanometer accuracy calibration table using a calibration algorithm. This invention's laser positioning accuracy calibration method for the 3D printing device improves accuracy and is simple to implement.

The present invention relates to a method for extracting a molecular structural formula, a computer-readable medium, and a computing device. The method for extracting a molecular structural formula according to the present invention includes: a target detection step of detecting a molecular structural formula in an information source file and segmenting out a molecular structural formula image including the molecular structural formula from the information source file; a graphic recognition step of performing graphic recognition on the molecular structural formula image to obtain a character part and a skeleton part of the molecular structural formula; a feature point marking step of marking feature points in the skeleton part as first type feature points or second type feature points, the skeleton part including one or more line segments, and the feature points including endpoints of the one or more line segments; and a connection relationship inference step of using a Markov logic network to infer connection relationships between first type feature points, between second type feature points, and between first type feature points and second type feature points according to a predefined inference formula.