89results about How to "Accurate comparison" patented technology

The invention discloses an intelligent industrial chain construction method and system based on multi-dimensional qualification affirmation, and the method comprises the steps: constructing a multi-dimensional qualification evaluation model based on industrial characteristic data; obtaining the preprocessed multi-dimensional qualification data of the industry chain related enterprises; analyzing the multi-dimensional qualification data by using a multi-dimensional qualification evaluation model to obtain a multi-dimensional quantitative index of the qualification data of the industry chain related enterprises; based on the multi-dimensional quantitative indexes, a machine learning algorithm is adopted to carry out classification to obtain enterprise category distribution, and comprehensive scoring and sorting are carried out on industry chain related enterprises according to set weights to obtain a sorting result; interaction data, transaction intensity data and resource flow data among related enterprises of the industry chain are acquired; and constructing an intelligent industrial chain network diagram according to the data obtained in the steps. The method is helpful for understanding an industrial chain structure, identifying key enterprises and evaluating potential risks.

The invention discloses a roadbed compaction degree monitoring method and system based on airborne laser radar, and belongs to the technical field of roadbed compaction degree intelligent monitoring, the monitoring method comprises the following steps: using an unmanned aerial vehicle to collect roadbed data, and constructing a historical database and a navigation database; according to the timestamps of the data in the historical database and the navigation database, all the timestamps in the two databases are combined and subjected to duplicate removal, and a complete time axis is constructed; according to the three-dimensional point cloud data subjected to navigation deviation parameter correction, relevant parameters of the roadbed compactness are obtained through calculation, and a relevant parameter database is constructed; fitting the relevant parameter data in the relevant parameter database and the three-dimensional point cloud data in the historical database to obtain a roadbed compactness characterization parameter model; and finally, combined with two-dimensional image data in a historical database, generating change images of the roadbed compactness in different time periods. According to the invention, intelligent monitoring of the roadbed compactness can be effectively realized.

The invention relates to a landslide motion analysis method based on unmanned aerial vehicle remote sensing aerial survey and three-dimensional geological modeling, which comprises the following steps: acquiring high-resolution topographic data of a landslide hidden danger area by utilizing a laser scanner technology carried by an unmanned aerial vehicle, and calculating landslide instability failure ranges and stability coefficients under different scenes by utilizing finite element Geostudio software; and taking the stability coefficient as a reduction coefficient of a rock-soil body strength parameter, then determining the rock-soil body strength parameter under different scenes in discrete element PFC software, and realizing three-dimensional failure motion process analysis of the landslide under different scenes by using discrete elements. According to the invention, through PFC software, landslide occurrence scales under different rainfall and earthquake scenes and motion characteristics of landslides of different scales after instability can be compared; large deformation under the fluid-solid coupling condition can be effectively simulated, rock-soil body strength parameters under different rainfall (or earthquake) scenes can be determined, and the landslide three-dimensional motion process under different scenes can be carried out.

An astronavigation material database system belongs to the technical field of astronavigation material databases and comprises a basic platform layer, an application service layer and a presentation layer. The public service layer comprises a public service plate; the application function layer comprises but is not limited to a material management plate, a material data application plate and a material big data analysis plate. The invention discloses an astronavigation material data system, which comprises a word standard specification, establishes an astronavigation material database, can perform big data analysis according to a computer software technology, a network technology and a digital technology, and can realize intelligent recommendation and intelligent knowledge management of astronavigation materials.

The invention belongs to the technical field of discrete intelligent manufacturing, and particularly relates to a discrete intelligent manufacturing method and system.The discrete intelligent manufacturing method comprises the following specific steps that product information in order products is split to form a bill of materials, the received order information is stored based on ERP software, and according to data in the bill of materials, the order information is stored according to the data in the bill of materials; computer aided design software is adopted to create a three-dimensional detailed model of the product, parts of the three-dimensional detailed model of the product are split, and the production quantity and time nodes of each part are determined; through the cooperation of the structure, the materials required by the product orders are split and listed according to the number of the product orders, so that workers can purchase the materials required by the products more conveniently, and through the mutual cooperation of the ERP system and the MES system, according to the materials listed on the material list, the materials required by the products can be purchased more conveniently. According to the method, the production steps are split while the number and the time node of the parts processed by the materials are determined, so that the parts are produced in batches in the production process.

The invention provides a multi-dimensional comparison method and device for PDF documents and electronic equipment, and the method comprises the steps: dividing a first PDF document and a second PDF document which are to be compared into a plurality of object regions; analyzing the plurality of object areas to obtain object contents to be compared; aiming at each piece of first object content in the first PDF document, determining that the first object content corresponds to second object content to be compared in a second PDF document; comparing the first object content with the second object content to determine difference information; synchronously marking the difference type and the index information of the difference information in the first PDF document and the second PDF document. According to the method and the device, differences of contents of multiple dimensions of texts, images, tables and formulas in the two PDF documents can be comprehensively, accurately and automatically compared and marked, so that the accuracy and the efficiency of document processing are improved, and the requirements of complex application scenes are met.

The invention relates to the field of video inspection, in particular to an active video inspection system based on Internet of Things driving, which comprises a current analysis module, an image input module, an image input detection module, an image output detection module and an audio detection module. According to the method, the current of the conference room equipment is monitored, abnormal equipment is found in time, rapid fault diagnosis is achieved, the effectiveness of the video stream is effectively judged by obtaining and processing the video stream and analyzing the grayscale image, it is ensured that the input picture is always in an available state, the blank screen phenomenon and picture distortion are avoided, and the user experience is improved. The image output detection module verifies the display effect, ensures the accuracy of an output picture and determines an abnormal area when the output picture is abnormal, the audio detection module plays a test audio and collects feedback, the sound receiving effect of a microphone is effectively evaluated, the conference room equipment is automatically and actively inspected, the equipment state is continuously monitored and evaluated, and the sound receiving efficiency is improved. Potential faults can be found and prevented in time, and the sudden fault risk of equipment is remarkably reduced.

The present invention discloses a roadbed compaction monitoring method and system based on airborne laser radar, belonging to the technical field of intelligent monitoring of roadbed compaction. The monitoring method includes using a drone to collect roadbed data, and constructing a historical database and a navigation database; according to the timestamps of the data in the historical database and the navigation database, all timestamps in the two databases are merged, duplicated, and constructed to obtain a complete time axis; according to the three-dimensional point cloud data corrected by the navigation deviation parameters, the roadbed compaction related parameters are calculated, and a related parameter database is constructed; the related parameter data in the related parameter database and the three-dimensional point cloud data in the historical database are fitted to obtain a roadbed compaction characterization parameter model; finally, combined with the two-dimensional image data in the historical database, the roadbed compaction change image under different time periods is generated. The present invention can effectively realize the intelligent monitoring of roadbed compaction.

The invention discloses a vernier sensitization vacuum pressure sensor based on optical fiber end face interference, and relates to the technical field of film sensors. The problem of low measurement precision caused by electromagnetic interference of a traditional capacitive film pressure sensor is solved. An interference cavity formed between the lower end face of the center optical fiber of the separated vernier assembly and the pressure sensing diaphragm serves as a sensing cavity for pressure detection, the sensing cavity (an optical fiber Fabry-Perot structure) formed by the pressure sensing diaphragm and the center optical fiber is adopted to replace a capacitor structure, optical waves are used for measurement, electric signals are not transmitted in the optical fiber, and the measurement precision is high. Therefore, the cable has excellent anti-electromagnetic interference and long-distance transmission capabilities. The pressure sensor is mainly used for pressure measurement.

The invention discloses an intelligent measuring and reporting device and method for insect pest conditions of vegetables in a facility, and the method comprises the steps: carrying out the color trapping of pests through employing a sticky card according to the color taxis principle of the pests in the daytime, and trapping the pests through employing light according to the phototaxis principle of the pests at night; shooting the insect sticking plate in different time periods to obtain image information, and uploading the image information to a cloud analysis server through a network for image recognition and judgment, data statistics and comprehensive analysis to obtain an insect condition result analysis report; the obtained insect condition result analysis report is issued to a terminal in real time for a user to obtain, and technical support is provided for subsequent insect pest prevention and control.

The invention relates to the technical field of model construction, in particular to a construction method and system for an OLED efficiency roll-off prediction model, and the method comprises the following steps: extracting key performance indexes, including brightness, pixel density, temperature and voltage parameters, based on collected OLED equipment performance correlation data, evaluating the correlation of parameters in the aspect of predicting OLED efficiency roll-off, and calculating the performance of the OLED equipment; and a data processing flow is adjusted, and a key performance index set is constructed. According to the invention, the key performance indexes are extracted and optimized by using the collected OLED equipment performance related data, so that the model can reflect the efficiency roll-off of the OLED more accurately, integrates new use data and analyzes the time sequence and the use frequency, thereby improving the recognition capability of the OLED service life influence factors, and improving the recognition efficiency of the OLED. And the stability of the model under various test conditions is improved through iterative testing, so that the reliability of a prediction result is optimized, and the efficiency of data analysis and processing is optimized by collecting and processing multi-source data in real time and adopting a continuous feedback mechanism.

The invention discloses a project progress verification system based on a BIM and a video fusion technology, and the system comprises a BIM model loading module which is used for obtaining a BIM model file and loading a BIM model; the engineering video acquisition module acquires video stream data of a construction site through preset engineering video acquisition equipment; the fusion management module fuses the acquired video stream data with the BIM model to obtain a fusion model; the feature extraction module determines a key area position which is marked on the BIM model in advance, extracts model data on the key area position and performs model feature extraction; and the progress verification module performs data matching on the model features extracted from all the key area positions according to a preset feature library, determines the construction progress of all the key area positions according to a matching result, and generates a project progress verification result. According to the method, the engineering progress can be automatically, efficiently and accurately compared, and the condition of the engineering progress can be more intuitively displayed.

The invention relates to a ship gear identification method and device based on target tracking and re-identification, and a medium. The method comprises the following steps: obtaining ship information; target identification is carried out based on YOLO5s introduced with an extended Kalman filtering algorithm, a ship trajectory is tracked from ship entry, and a ship detection frame of a trajectory tracking result and ship basic information are superposed on a ship real-time video; if the ship trajectory is lost due to shielding in the ship trajectory tracking process, determining a lost target trajectory based on a Deep OC-SORT algorithm and a Redis re-recognition technology; in the ship trajectory tracking process, if it is recognized that the ship does not move for continuous preset time, it is judged that ship gear stopping is completed, the ship stopping position is obtained, and a ship actual gear diagram is formed; and judging whether the ship is scheduled according to the ship scheduling graph, the actual scheduling graph and the basic information of the ship, and if the ship is not scheduled, giving an alarm. Compared with the prior art, the method has the advantages that accurate positioning of the ship position can be achieved, and the like.

The invention discloses an intelligent homework correcting method based on a syntax tree and semantic analysis, and relates to the technical field of intelligent teaching, according to an education application scene, a dictionary suitable for the education scene is adopted to improve the accuracy of word segmentation and part-of-speech recognition, meanwhile, text cleaning and text formatting are carried out, the subsequent correcting accuracy is improved, and the teaching efficiency is improved. A syntax tree of formula analysis is adopted, simpler, more convenient and more accurate formula comparison is performed, a semantic SWRL consistency representation method is adopted, whether user answers and standard answers are equivalent or not can be more accurately compared, and homework correction methods of text cleaning, character string comparison, formula semantic comparison, text semantic comparison and rule semantic comparison are combined, so that the homework correction efficiency is improved. According to the method, the correction accuracy can be considered under the condition that the efficiency is improved, the problems that a traditional character string comparison method is not sensitive to the sequence of characters and misjudgment is likely to be caused are solved, meanwhile, the uncertainty in a semantic similarity comparison method is greatly reduced, and a more accurate correction result is obtained.

A general assembly design system and a method for controlling general assembly design information of a spacecraft, wherein a basic information reading module in the system obtains basic information of equipment from an IDS system; a cabin panel assembly is taken as a general assembly design unit, and the general assembly design unit is opened in PRO / E; a general assembly information setting module traverses the PRO / E model tree, identifies all equipment models and bracket models in the general assembly design unit and creates corresponding general assembly parameters; the identified equipment model is matched with the read basic information to determine the basic information of all equipment models; at the same time, the input bracket model basic information is received, and the general assembly parameter value is determined according to all the basic information, and the general assembly parameter value is assigned to the corresponding equipment model and bracket model and stored; the general assembly information reading module reads the general assembly parameter value from the equipment model and bracket model opened by PRO / E, and displays it in a preset form; and the general assembly information output module outputs the general assembly parameters and parameter values ​​stored in the general assembly information setting module to the general assembly information database for storage.

The invention provides a Parkinson's disease gait comparative analysis method based on a time sequence characteristic curve, and the method comprises the steps: obtaining video data containing gait information which comprises gait information before and after a user uses a medicine; performing posture recognition on a human body in the video data, and extracting position information of key body parts; calculating gait features according to the extracted key point information, and constructing a time sequence curve according to changes of the features along with time; performing data processing on the extracted gait features by using a digital filter; local frequency and amplitude are integrated into distance measurement by using a frequency amplitude and amplitude enhanced dynamic time warping method to quantify the difference of gait movement; obtaining a curve difference distance; metrics are calculated based on cumulative frequency differences and amplitude differences in the optimal regular path. The method has remarkable advantages in the aspects of accuracy, time sensitivity, robustness, adaptability, comprehensive evaluation ability, quantitative analysis, clinical application potential and the like.

The invention discloses a handwriting recognition and judgment method and system based on knowledge transfer learning, and relates to the field of handwriting recognition. Establishing a rectangular plane coordinate system, calculating a curvature and an average curvature, comparing the curvature and the average curvature with a preset threshold value, judging stroke types, and counting linear strokes and the total stroke number; calculating a vector included angle and an angle average value; the total handwriting length, the average acceleration and the handwriting generation rate are obtained in combination with real handwriting sample data, and then a comprehensive index is calculated; respectively calculating an average pressure value and a pressure variance, further calculating a comprehensive pressure index, and calculating a handwriting feature index and a writer identity authenticity score according to the comprehensive pressure index; calculating a score average value and a score standard deviation of the identity authenticity of the writer, and obtaining a handwriting authenticity index threshold value; calculating an identity authenticity score of the to-be-judged writer, and comparing the score with a handwriting authenticity index threshold to judge the authenticity of the handwriting; according to the method, handwriting recognition and authenticity judgment can be more accurately carried out.

The invention discloses a cognitive potential evaluation method and system based on a reinforcement learning calculation model. The method comprises the following steps: acquiring a cognitive evaluation task combination based on a preset task library; cognitive evaluation data of the user for each cognitive evaluation task is obtained, and the cognitive evaluation data at least comprises specific options of the user for completing the cognitive evaluation task and score results; on the basis of a reinforcement learning calculation model, the learning rate of each cognitive evaluation task is calculated; converting the learning rate of each cognitive evaluation task into a norm score; the norm scores of all the cognitive evaluation tasks are subjected to weighted average to obtain comprehensive evaluation scores of the user, and different comprehensive evaluation scores correspond to different cognitive potentials. According to the method, learning modeling can be carried out on the performance of the user in a certain cognitive field through the reinforcement learning calculation model, and the cognitive potential of the user in the certain cognitive field is effectively evaluated.

The invention relates to the technical field of plate machining control, and particularly discloses a plate combined machining control system and method.The system is provided with a plate cutting machining judgment module, a plate punch forming machining judgment module, a plate drilling and tapping machining judgment module and a control center; according to the method, the plates are controlled to be cooperatively machined through multiple technologies of cutting, punch forming and drilling and tapping, the machining reasonability of a certain program can be judged based on the machining processes of cutting, punch forming and drilling and tapping of the machined plates through multi-dimensional data analysis, the accurate machining completion condition of the machined plates is obtained, and the machining efficiency of the machined plates is improved. The composite machining technology is more accurate, the waiting time between the technologies can be shortened, the production efficiency of the plate is improved, meanwhile, the machining accuracy of parts on the plate in the machining technology can be improved, and the utilization efficiency of the plate is improved to the maximum extent.

The invention is suitable for the technical field of image data processing, and provides a defect detection method and device of a roller printing machine, the roller printing machine and a storage medium. The defect detection method of the roller printing machine comprises the steps that a to-be-recognized image collected by a camera and target printing image data are obtained; extracting a plurality of target pixel value regions in the target printing image data; calculating first midpoints of the plurality of target pixel value regions, and calculating a first distance between the plurality of first midpoints; according to the pixel value range corresponding to the target pixel value region, extracting to-be-recognized pixel regions corresponding to the plurality of pixel value ranges in the to-be-recognized image; and according to the first distance and the second distance, determining whether the printed matter in the to-be-identified image has printing defects. According to the technical scheme, the efficiency and precision of printing defect detection are effectively improved, manual intervention can be reduced, and the detection requirements of different printed matters can be met.

The invention provides a media screen content comparison method and system based on video features and similarity calculation, and the method comprises the steps: obtaining a video image, and carrying out the preprocessing of the video image; carrying out key frame extraction on the preprocessed video image to obtain a key frame sequence; extracting an SURF (Speeded Up Robust Feature) with scale invariance corresponding to each frame from the key frame sequence; constructing a K-D tree according to the SURF features of the video feature points, and performing two-stage feature matching based on the K-D tree and RANSAC; according to a feature matching result, performing statistical index calculation on a matching point number and then outputting a similarity index frame by frame; and according to the similarity indexes, comparing the similarities between the frames of the video images, and formulating corresponding early warning schemes according to different similarity levels. According to the invention, the comprehensive evaluation model is established through the video content comparison technology, real-time monitoring and early warning are realized, the inconsistency between the playing content and the preset content can be found in time, the propagation of error information is avoided, and the green and healthy urban light environment is ensured.

The invention relates to an automatic generation method, system and device for a clinical image reference map and a medium, and the method comprises the steps: a doctor determines the anatomical positioning of a required reference map according to a browsed image, and extracts the image examination category, examination items and basic information of a patient of the patient, and the relative positions of the layer where the anatomical positioning is located in all scanning layers; the method comprises the following steps: matching same image examination items and categories in a hospital image diagnosis report database, matching basic information of a patient, examination equipment and the like according to a priority set by a doctor, determining a retrieval keyword according to anatomical positioning, determining a retrieval structured medical image report statement based on the retrieval keyword, and querying the hospital image report database. Matching the report statement and judging the report as a normal report; and calling the report image, matching a reference atlas layer based on the relative position of the anatomical positioning layer in all scanning layers, automatically generating a reference atlas, and displaying the reference atlas to a physician for reference. According to the invention, accurate and abundant imaging reference images can be provided for imaging diagnosis doctors, and the working efficiency can be improved.

The utility model discloses an infusion soft bag production line and a soft bag aluminum film three-stage code identification and detection device, which are characterized in that a shooting unit is arranged on a film conveying mechanism, an aluminum film subjected to code spraying and printing operation is shot, a code spraying and printing picture is obtained and output, and then the code spraying and printing picture is obtained through an identification unit; and finally, after the character comparison unit obtains the aluminum film recognition result and the aluminum film recognition result is consistent with the expected aluminum film character content, the aluminum film is judged to be the aluminum film with qualified printing, otherwise, the aluminum film is judged to be the aluminum film with unqualified printing, and after the aluminum film subjected to code spraying and printing is cut into a specified shape by a film cutting knife, the aluminum film is subjected to code spraying and printing. The cut aluminum film is taken out of the film conveying mechanism through the film taking unit and then placed at a designated position, the problems of missing spraying, wrong spraying and insufficient spraying of the three-stage code of the aluminum film are solved through automatic photographing and rapid recognition and comparison, accurate code spraying is achieved, and the workload and the working intensity are reduced.

The invention belongs to the field of video playing, and provides a coaxial playing and comparing method based on a video file, which comprises the following steps: S1, video preprocessing; s2, carrying out coaxial alignment treatment; s3, synchronously playing the video; s4, area selection and dynamic comparison; s5, performing real-time comparison display; s6, dynamic analysis and intelligent assistance; through frame rate synchronous processing and a coaxial alignment algorithm, different videos can play the same frame number at the same time point, and key frames are accurately aligned, so that the video comparison efficiency is greatly improved; a user does not need to manually adjust the playing progress and the alignment mode of the video, so that time and energy are saved; according to the method, multiple feature extraction and matching algorithms are adopted, the feature matching accuracy of the video is improved, an affine transformation matrix is calculated through multiple sets of feature point pairs, an average value is taken, the accuracy of the transformation matrix is further improved, and pictures of different videos can be overlapped and compared more accurately.

The invention discloses a calligraphy stroke feature extraction method and system based on a fixed-size grid. The method relates to the technical field of stroke feature extraction, and comprises the following steps of calligraphy stroke image preprocessing and grid division, grid-based stroke space distribution feature extraction, size feature extraction, grid-based stroke direction feature extraction and feature fusion and output. The method comprises the following steps: S1, normalizing a binary image of calligraphy strokes to a fixed canvas and dividing the binary image into grids; s2, calculating the stroke density of each grid and constructing a density feature vector; S3, refining the image to obtain a skeleton, calculating a direction field and generating a direction feature vector; and finally, the two vectors are connected in series and fused, and a joint feature vector is output for a subsequent task, so that the calligraphy stroke feature extraction accuracy is improved, and the problem of low calligraphy stroke feature extraction accuracy caused by insufficient understanding of dynamic essence, complex form and high-level semantics of calligraphy art in the prior art is solved.

The embodiments of the present application belong to the field of image detection technology in artificial intelligence, and relate to a handwriting verification method, device, computer equipment and storage medium based on twin neural networks. In addition, the present application also relates to blockchain technology, and the user's standard handwriting image and imitation handwriting image can be stored in the blockchain. After obtaining the imitation handwriting image imitated for the standard handwriting image, the present application extracts the standard handwriting features and the imitation handwriting features of the standard handwriting image and the imitation handwriting image respectively, and finally calculates the handwriting similarity of the standard handwriting features and the imitation handwriting features to determine the handwriting similarity as a confirmation method for verifying whether the handwriting is consistent. The present application can quickly and accurately compare whether the two pictures of given two handwritten handwriting pictures are the same handwriting, and has higher robustness and accuracy. At the same time, the feature extraction operation has fewer model parameters and higher processing efficiency under the condition that the accuracy does not change much.

The invention relates to the technical field of plastic molds, in particular to a plastic mold processing control method and system, and the method comprises the steps: carrying out three-dimensional scanning on a plastic mold; acquiring three-dimensional data of the plastic mold, and comparing the three-dimensional data with a preset value of the plastic mold; the electric spark assembly is used for cutting the plastic mold according to a preset value; according to the plastic mold machining control method and system, an efficient, precise and reliable solution is provided for precise machining of the plastic mold by integrating the advanced three-dimensional scanning technology and the electric spark machining technology, the machining precision and efficiency are improved through the method and system, the machining quality is optimized, and the machining cost is reduced. And the environmental influence is reduced, the processing reliability is enhanced, the cost is saved, and the product competitiveness is improved.

The embodiment of the present invention discloses a sample data expansion method, device, electronic device and storage medium, which relates to the field of digital imaging and deep learning technology, wherein the method comprises: obtaining digital microscopic imaging and annotation of a sample data set as a training set, designing an initial data expansion model, obtaining an output image according to the initial data expansion model, the initial data expansion model comprises a feature extraction model, a feature generation model and a feature discrimination model, using the feature discrimination model to compare the input image and the output image to obtain a loss value, optimizing the feature generation model according to the loss value, using the training set to iteratively optimize the initial data expansion model until the set conditions are met, obtaining a trained data expansion model, and using the data expansion model to expand the sample data set. The present invention solves the problem that the prior art can only perform model training on a small sample data set, is prone to overfitting, has poor generalization ability, and is difficult to achieve good results.

The utility model provides an auxiliary device for impurity removal of molten aluminum. The auxiliary device comprises a molten aluminum storage assembly, comprising a supporting frame assembly and a molten aluminum measuring assembly, the supporting frame assembly is fixedly connected with the side end of a molten aluminum storage assembly, and the molten aluminum measuring assembly is installed on the supporting frame assembly; wherein the molten aluminum storage assembly comprises a molten aluminum storage barrel, an upper mounting plate and a liquid discharging pipe, the upper end of the molten aluminum storage barrel is open, the upper mounting plate is hinged to the opening of the upper end of the molten aluminum storage barrel, the liquid discharging pipe is fixedly communicated with the lower end of the molten aluminum storage barrel, and a switch valve is fixedly installed on the liquid discharging pipe; the molten aluminum measuring assembly comprises a molten aluminum measuring box, and the upper end of the molten aluminum measuring box is open. When the aluminum liquid impurity removal device is used for carrying out impurity removal research on the aluminum liquid, the aluminum liquid can be quantitatively obtained, so that the aluminum liquid amount before and after impurity removal can be accurately compared, the aluminum liquid impurity removal amount can be accurately obtained, and the accuracy of the aluminum liquid impurity removal research is improved.

The application discloses a kind of fast moving face recognition method and system based on multi-frame image enhancement, comprising the following steps: through imaging device acquisition multi-frame image sequence, obtain multi-frame image after pre-processing;The multi-frame image after pre-processing is input into the improved MonST3R network, carries out spatial alignment and structure analysis, obtains the image after alignment and three-dimensional structure information;The above result is input into CRNet network, separates and enhances multi-scale feature, and obtains enhanced feature;Enhanced feature is input into space-time attention fusion module, and multi-frame feature representation is obtained by weighted fusion;Fusion feature is input into identification network, and dimension reduction is carried out, and face embedding feature vector is obtained;Similarity comparison is carried out between embedding feature and database template, and recognition result and confidence prompt are output.The application can realize efficient, accurate and resource-friendly face recognition under dynamic complex environment.