374 results about "Feature description" patented technology

The invention belongs to the technical field of three-dimensional measurement, and discloses a 3D laser line scanning system based on a point cloud positioning algorithm and a control method. The method comprises the following steps: acquiring environment initial point cloud data and scanning target information; carrying out feature extraction, constructing an environment feature description model, carrying out three-dimensional space partitioning processing, carrying out regional reflectivity feature analysis, and constructing a space reflection feature model; adaptive optimization configuration is carried out on the laser scanning parameters, and a scanning strategy is generated; scanning path dynamic planning is carried out on the scanning target information, scanning density distribution calculation is carried out, and a scanning path is generated; performing laser line scanning control by using the scanning path and the scanning strategy, and performing scanning data quality real-time monitoring to obtain scanning quality evaluation data; scanning parameter iterative optimization is carried out on the scanning quality evaluation data, point cloud data real-time fusion reconstruction is carried out, a high-precision three-dimensional model is generated, the modeling precision is improved, and meanwhile the system resource utilization efficiency is optimized.

The invention provides a visible light and infrared fusion-based river channel fish banning ship monitoring method and system, and relates to the technical field of computer vision, and the method comprises the steps: firstly obtaining a visible light image sequence and an infrared image sequence of a river channel monitoring region; ship visual feature extraction and thermal feature extraction are carried out on the visible light image sequence and the infrared image sequence respectively to obtain a visible light ship feature set and an infrared ship feature set, and cross-source feature fusion processing is carried out on the two feature sets to generate a cross-source fusion feature set; and calling a pre-trained fish forbidding ship identification model to analyze the cross-source fusion feature set, generating a ship monitoring result including the ship type, the position change track and the suspected fish forbidding behavior identifier, and finally generating fish forbidding early warning information including the real-time position and behavior feature description of the suspected fish forbidding ship based on the ship monitoring result. And measures can be quickly taken to stop the fishing forbidding behavior.

The invention relates to the technical field of data analysis, provides an intelligent factory automatic monitoring method and system based on knowledge base enhancement, and realizes more accurate anomaly analysis and more effective process adjustment of an intelligent factory. The method comprises the steps of performing knowledge enhancement fusion processing on an obtained real-time monitoring data set of an intelligent factory through a pre-constructed process knowledge base and a pre-constructed monitoring rule base, and generating a process knowledge graph; performing abnormal mode recognition processing on the process knowledge graph based on a semantic matching strategy, extracting feature description of an abnormal event and a semantic association path with a historical monitoring text, and generating an abnormal mode analysis result containing abnormal root cause inference; according to the abnormal mode analysis result and the dynamic incidence relation in the process knowledge graph, an automatic monitoring report containing root cause priority ranking and optimization operation guidance is generated, and the automatic monitoring report is fed back to the intelligent factory control terminal to trigger process adjustment operation.

The invention relates to an underwater topographic measurement method based on laser radar and visual fusion, which comprises the following steps: synchronously acquiring underwater point cloud, images and physical environment sensing data, and endowing a unified space-time label to realize the space-time consistency of multi-source data; according to the method, the data quality of different modes is improved by means of preprocessing, denoising, scale normalization, feature enhancement and the like, an environment interference weight matrix is constructed in combination with an environment sensing model, feature extraction and matching algorithm parameters are adaptively adjusted according to different environment states, and multi-scale feature description, spatial consistency and physical constraint criteria are established, so that the multi-modal data quality is improved. According to the scheme, high-reliability feature matching between the point cloud and the image is achieved, finally, through environment-driven iterative optimization and fusion, the robustness and precision of space registration are improved, high-precision multi-modal data automatic registration can be stably achieved in the underwater dynamic environment, adaptability is high, and environment perception and space measurement quality is effectively improved.

The invention discloses a two-stage infrared and visible light image registration method, system and device. The method comprises the following steps of image preprocessing, contour extraction, angular point detection, feature description and matching, affine transformation estimation, multi-scale optical flow estimation, optical flow constraint and loss function, reverse resampling and fusion and error evaluation. According to the invention, rough registration is carried out by using contour angular point features, so that a preliminary alignment result can be quickly obtained; refined alignment is carried out in combination with an unsupervised optical flow network, and sub-pixel-level registration precision is achieved. The contour angular points are based on shape information of an image target, are natural and are not influenced by spectral differences, and the matching stability is enhanced through main direction and angle features. The unsupervised depth optical flow model estimates a pixel displacement field by learning consistency characteristics of an input image, and does not need to depend on annotation data. The combination can effectively eliminate the difference between infrared light and visible light, and improves the robustness and adaptability of registration.

The invention relates to a point cloud registration method based on improved ISS-TOLDI features in combination with ICP, and belongs to the technical field of laser point cloud application. The method comprises the steps of obtaining a source point cloud and a target point cloud, and performing preprocessing; extracting feature points from the point cloud data by using an internal shape descriptor ISS algorithm; performing feature description on the extracted feature point set by using an improved TOLDI algorithm; performing coarse registration on the point cloud by using a sampling consistency initial registration algorithm; and performing fine registration on the coarsely registered point cloud data by using an iterative closest point ICP algorithm to complete point cloud registration. According to the method, the TOLDI algorithm is improved, so that the calculation complexity is reduced, the feature integrity is ensured, and the point cloud registration precision is higher.

The invention relates to the technical field of unmanned aerial vehicle geological survey, and discloses a path planning method and system for unmanned aerial vehicle geological survey, and the method comprises the steps: obtaining the image, spectrum and topographic data of a remote mountainous area through a multispectral sensor and a laser radar, and recognizing potential vein abnormal points through data fusion and geological feature analysis; and constructing a geological interestingness weight map and planning a flight path. Dense sampling flights are performed in a high interestingness region, high resolution data are processed in real-time to identify key reconnaissance targets, a weight map is updated, and a flight trajectory is dynamically adjusted. Through the closed-loop self-adaptive exploration process, accurate positioning and feature description of the hidden vein are realized, the mineral exploration efficiency is improved, the labor cost and the safety risk are reduced, and an innovative solution is provided for mineral resource exploration in remote areas.

The invention relates to the technical field of augmented reality application, and discloses an augmented reality system and method for immersive experience. The method comprises the following steps: collecting dynamic spatial data of a physical environment in real time, and synchronously obtaining an interactive behavior data stream of a user terminal; performing environment semantic modeling on the dynamic spatial data to generate multi-level feature description of a physical environment, and determining a rendering priority sequence of virtual-real superposition content according to the multi-level feature description; extracting behavior pattern characteristics in historical interaction data of the user, performing immersion correlation analysis on the behavior pattern characteristics to obtain environment cognitive preference parameters of the current user, and generating a virtual-real interaction event sequence in combination with the environment cognitive preference parameters and the interaction behavior data flow; and dynamically compensating the virtual-real synchronization error in the immersive experience process according to the rendering priority sequence and the virtual-real interaction event sequence. According to the method, the accuracy and naturalness of virtual-real fusion can be improved, the personalized requirements of the user are met, and the overall immersive experience effect is enhanced.

The invention provides a satellite fault diagnosis method and system based on multi-source information fusion. The satellite fault diagnosis method comprises the steps that firstly, a multi-source monitoring data set of platform system state data, load system working data and measurement and control link interaction data is generated in the satellite testing process based on an emergency fast response satellite testing system; then performing multi-source information association extraction processing on the multi-source monitoring data set to generate an association feature set reflecting a coupling relationship between systems; then inputting the associated feature set into a pre-trained satellite fault diagnosis model for fault mode analysis processing to obtain a fault feature description containing a fault trigger node and an influence link; a preset fault knowledge base is matched based on the fault feature description, and a fault diagnosis result containing a fault type identifier and a propagation path is generated; and finally, generating a satellite fault handling strategy according to a fault diagnosis result. The method can comprehensively and accurately diagnose the satellite fault and improve the satellite reliability.

According to the movie and television play table book extraction method and device, the storage medium and the computer equipment provided by the invention, after an audio and video file of a movie and television play is split into a video file and an audio file, feature recognition is performed on the video file to obtain a subtitle text, speaker face information and a video understanding text; performing voice understanding on the audio file to obtain a voice transcription text and a voice understanding text; wherein the voice transcription text can be corrected into the standard transcription text with high accuracy through the subtitle text. Therefore, based on the face information of the speaker, the line segment of each speaker in the standard transcriptional text and the audio and video file is aligned, so that speaker information with accurate segmentation and semantic coherence can be obtained; and then, through combination with a character side-writing text generated by side-writing analysis on the speaker based on the video, the voice understanding text and the speaker information, table book information is constructed, and related feature description of the character can be covered on the basis of containing the line content, so that the content and depth of the table book are enriched.

The invention discloses a remote sensing fine-grained ship detection method for multi-scale context feature enhancement, and the method comprises the steps: enabling an output feature of a backbone network to pass through a deep separable convolution layer through a constructed attention feature enhancement module, carrying out the blocking processing, obtaining a local feature, and carrying out the attention enhancement of feature description through a channel. Then, through a constructed multi-scale context feature fusion module, extracting context information of ships with different appearance scales from multi-scale features output by the feature pyramid through hole convolution operation with different expansion rates; after target ship edge information is captured through a constructed feature refining fusion module, the target ship edge information is fused with output features of a context feature fusion module, the identification capability of a network model on ship details and structural features is improved, and the problem that an existing method cannot achieve self-adaptive multi-scale feature enhancement and fusion, and the ship detail and structural feature recognition efficiency is improved is solved. Therefore, the problem of loss of detail features beneficial to positioning and classification is solved.

Embodiments of the present application provide a fingerprint anti-counterfeiting neural network training method, a fingerprint anti-counterfeiting method and device, comprising: obtaining multiple sets of training data, each set of training data including: first native domain data, second native domain data and third native domain data; using the multiple sets of training data to train an initial classification network to obtain a target classification network, the initial classification network including a fusion subnetwork and a classification subnetwork, for each set of training data, the fusion subnetwork is used to generate a first fingerprint matching pair based on the feature description matrix of the first native domain data and the second native domain data, and generate a second fingerprint matching pair based on the feature description matrix of the first native domain data and the third native domain data, and the classification subnetwork is used for fingerprint classification and identification based on the first fingerprint matching pair and the second fingerprint matching pair. Through the present application, the problem of low identification accuracy of true and false fingerprints in related technologies is solved, and the effect of improving the identification accuracy of true and false fingerprints is achieved.

The invention discloses a deep learning-based mixed-shooting image correction method and system, and relates to the field of computer vision, and the method comprises the steps: carrying out the illumination normalization processing of a mixed-shooting image group, and constructing a Gaussian pyramid image group; inputting the Gaussian pyramid image group into a ResNet-34 network to extract a multi-scale feature map, and generating spatial position and channel statistical information; and converting the spatial position and the channel statistical information into a DNA sequence fragment through a DNA base mapping rule. Illumination normalization is achieved through Retinex decomposition and Gaussian filtering, the exposure difference of different devices is effectively eliminated, the input quality of follow-up processing is guaranteed, traditional feature description is converted into biological sequence comparison through DNA base coding, geometric parameters are optimized in combination with quantum annealing, and the accuracy of the method is improved. The matching precision of a low-texture area and the correction effect of a large-view-angle-difference scene are remarkably improved, and the sub-pixel-level matching precision is achieved based on weighted RANSAC and multi-scale pyramid fusion.

The invention relates to the technical field of computers, in particular to an AGV task allocation method and system based on multi-objective optimization, and the method comprises the steps: obtaining a task demand set and a state information set of AGV equipment in a current operation scene, and the task demand set comprises a plurality of task units; performing feature extraction processing on the task demand set and the state information set to obtain task feature description of each task unit and capability feature description of each AGV device; performing joint optimization matching processing according to the task feature description and the capability feature description to generate a matching association relationship set of the task and the AGV equipment, wherein the matching association relationship set comprises a corresponding relationship between the task unit and the AGV equipment and a task execution sequence relationship; and generating a task allocation scheme based on the matching association relationship set, wherein the task allocation scheme comprises a to-be-executed task list of each AGV device and execution time sequence arrangement information of each task. Therefore, task requirements and equipment states can be accurately grasped, and efficient and matched innovative task distribution processing is realized.

The invention belongs to the technical field of optical three-dimensional scanning, and relates to a real-time registration method for high-density point clouds, which comprises the following steps: acquiring two groups of sparse point cloud data generated by three-dimensional scanning in an oral cavity, inputting the point cloud data into a constructed efficient registration model, extracting multi-dimensional feature descriptions of a source point cloud and a target point cloud, and constructing a registration model; the point cloud registration method comprises the following steps of: obtaining a point pair set, calculating a feature distance to form an initial point pair set, performing spatial consistency analysis on the initial point pair set, rejecting outlier pairs to obtain a cluster point pair subset, and finally performing point cloud registration calculation on the cluster point pair subset to complete point cloud registration. The invention further relates to a high-density point cloud real-time registration device, scanning equipment and a storage medium. According to the technical scheme provided by the invention, the registration precision and stability can be improved, and meanwhile, the requirement of the intraoral scanner on the real-time performance is ensured.

The invention relates to the technical field of power equipment state monitoring and fault diagnosis, and discloses an infrared diagnosis method and system for power transmission and transformation equipment. The method comprises the following steps: acquiring infrared thermal image data of equipment and removing noise by using a filter to obtain a clear image; carrying out pixel-level segmentation by adopting an image segmentation model, determining the boundary of the equipment part, and extracting an initial feature vector which comprises texture and depth heat distribution information; based on the feature sets, judging a temperature abnormal region, and performing material classification in combination with texture information; according to the method, the structural difference of the components is analyzed through the classification model, structural feature description is obtained, the structural features are further combined with the temperature abnormal area, fine feature matching is carried out, and therefore synchronous recognition is achieved, and the accuracy and real-time performance of power transmission and transformation equipment fault diagnosis are improved.

The invention relates to a food water content calculation method and device based on a visual image, equipment and a medium. First, food image data is collected, and first image data is obtained by using a standardized preprocessing technology. Secondly, extracting surface glossiness and texture feature values from the first image data, segmenting the image by using a feature detection algorithm, marking a key region related to the water content, and forming a visual feature set; thirdly, the relevance of all elements in the visual feature set is analyzed, food types are distinguished through a classification rule, feature weight distribution of different types of food is determined, and optimization of feature description is achieved; and finally, based on the optimized feature description, establishing a mapping relation model between the water content and the visual features, substituting a newly input food image into the model, predicting the water content, and outputting a real-time detection result. An efficient and convenient detection means is provided for the fields of food processing, storage and the like, the detection cost is reduced, and the production efficiency and the product quality control capability are improved.

The invention discloses a robot navigation and positioning method based on machine vision. The robot navigation and positioning method comprises the following steps: acquiring and preprocessing an environment image sequence; performing feature point detection and feature description by using an improved DELF model to generate a visual feature point set; obtaining a visual pose estimation result; obtaining fusion pose information; constructing a local environment map by adopting a sparse point cloud reconstruction method; generating a global environment map; generating an optimal advancing path by using a path planning method; and a robot motion control instruction is generated, path adjustment and navigation correction are performed, and high-precision autonomous positioning and navigation of the robot in a complex environment are realized.

The invention relates to the technical field of radar target recognition, and discloses a radar target recognition method and system. The method comprises the following steps: acquiring an original echo signal sequence acquired by a radar sensor, and performing noise suppression and pulse compression processing on the original echo signal sequence by using a signal preprocessing unit to generate preprocessed radar echo data; based on the preprocessed data, fusing to generate a comprehensive feature description vector of the target; a dynamically updated target feature association network is constructed, network nodes represent feature sets of different targets, network edges represent space-time association or feature similarity among the targets, and an online learning mechanism is adopted to continuously optimize a network topology structure; according to the real-time comprehensive feature description vector, gradually positioning a potential target area in the target feature association network by adopting a hierarchical focusing mechanism; and in the potential target area, a candidate target set is screened through a multi-dimensional similarity measurement algorithm, and candidate targets are sorted and output according to a feature consistency criterion.

The invention relates to the technical field of mine safety monitoring, in particular to a roadway section convergence multi-view three-dimensional reconstruction method and a monitoring system thereof.The method comprises the steps that a rectangular plane coordinate system is established on a roadway section, a plurality of measuring areas are arranged in the circumferential direction, a plurality of equal-height measuring points are evenly distributed in each measuring area in the elevation direction of the section, and every two adjacent measuring areas share at least one measuring point; a laser structured light projector and a CMOS receiving camera are arranged at each measuring point; matching the feature points through an improved stereo matching algorithm to obtain depth feature point information and establish a section point cloud; dividing deformation into local deformation and overall deformation, establishing deformation association between measurement points through deformation feature description, and obtaining overall convergence information of the section by adopting a point cloud registration method after local deformation compensation is formed; a section safety evaluation model is established, section deformation evaluation and reinforcement suggestions are given, all-directional non-blind area monitoring of the section is achieved, and the monitoring precision reaches the millimeter level.

The invention discloses an intelligent system failure analysis method based on fault recognition, and relates to the technical field of intelligent system reliability analysis, and the analysis method comprises the specific steps: employing Modelica to construct a multi-physical domain digital twinborn model, combing historical data, and building a structured fault mode library; developing a fault injection algorithm to acquire data to form a standardized data set; preprocessing data to construct a causal link and a probability flow network; carding task scenes to construct a reliability evaluation model and training the reliability evaluation model; inputting actual parameters to trace faults, generating an optimization plan, and outputting an analysis report and a plan after verification; according to the method, the problem of insufficient fault correlation analysis caused by modeling scale splitting in a traditional method is solved by constructing a multi-physical domain digital twinborn model covering microcosmic, mesoscopic and macroscopic and establishing a structured fault mode library of a standardized feature description system containing three types of faults of hardware, software and environment interaction.

The invention discloses a construction and recognition method for a sound source recognition network of vehicles in a highway tunnel, and the method comprises the steps: setting a distributed microphone array in the tunnel, and collecting a vehicle audio signal; a MobileNetV3 model based on the Mel frequency spectrum is constructed; the Mel spectrum feature extraction module is used for taking the Mel spectrum feature extracted by the Mel spectrum feature extraction module as input, taking a vehicle audio signal type as output, and training a MobileNetV3 model based on the Mel spectrum to obtain a sound recognition model; constructing a sound source localization model; according to the method, the MobileNetV3 model based on the Mel frequency spectrum is adopted for sound recognition, the sound source positioning model is constructed in a combined mode for sound source positioning, the MobileNetV3 model and the sound source positioning model work cooperatively, the feature description capability of abnormal sound of a vehicle is enhanced, the influence of tunnel echo and environment noise on recognition is effectively reduced, accidents in the tunnel are found in time and positioned accurately, and the method is suitable for popularization and application. The recognition and positioning accuracy of the vehicle sound source in the complex tunnel environment is improved, and the technical problem that in the prior art, the recognition precision of the accident sound in the tunnel is not high is solved.

The invention discloses a storage and identification method and system based on multi-dimensional personnel identification data, and the method comprises the steps: obtaining the effective feature data of a to-be-identified personnel, determining the feature value, the initial weight value and the interference quantized value corresponding to each dimension feature vector in the effective feature data, calculating a target weight value corresponding to each dimension feature vector based on the initial weight value of each dimension feature vector and the corresponding interference quantized value, calculating a fusion value corresponding to the effective feature data based on the target weight value and the feature value of each dimension feature vector, and performing feature vector mapping on the fusion value to obtain the effective feature data. And obtaining and storing a target feature description vector corresponding to each dimension feature vector, inputting the target feature description vector into a pre-training recognition model, and outputting a recognition result of the to-be-recognized person. Through the above method, the dimension feature vectors of different dimensions are fused, the accuracy of the target feature description vector is ensured, and the accuracy of the recognition result is improved.

The invention relates to a Web application firewall rule generation method and device, equipment and a medium, and the method comprises the steps: obtaining attack feature description information based on log information, latest vulnerability information and user demand information through a large language model; the log information comprises Web request information sent by the client; the newest vulnerability information represents newest vulnerabilities injected in the Web request information; the user demand information represents Web request information needing to be protected by the Web application firewall; calling a target tool through a model context protocol module; generating a Web application firewall rule based on the attack feature description information through the target tool; the target tool is a tool related to Web application firewall rule generation, and the efficiency and accuracy of Web application firewall rule generation are improved.

The invention relates to the field of mechanical assembly, in particular to a multi-view image fusion and precision real-time monitoring system in a mechanical assembly process, which is characterized in that the core of the system is a main element analysis unit, based on a differential geometry theory, three-dimensional point cloud data is creatively mapped to a Riemannian manifold space, a local geometric structure is constructed, and geometric features based on geodesic distances are extracted; a multi-view manifold structure is fused to form a unified assembly space three-dimensional reconstruction model, an optimal operation path and attitude are generated on a fused manifold through a Riemannian geodesic flow theory, and the system adopts a multi-scale curvature analysis and geodesic distance feature description method to effectively maintain an internal geometric structure of the surface of an assembly part; collaborative fusion of a multi-manifold topological structure is realized through a manifold alignment and stitching technology; path planning is carried out based on the Riemann geodesic flow theory, the optimal assembly path meeting the multi-constraint condition is generated, and the assembly precision is improved.

The invention relates to the technical field of image analysis, in particular to a flexible circuit board production quality detection method and system. The method comprises the following specific implementation steps: firstly, collecting a multi-angle luminosity image and multi-channel spectral data, and constructing a material-perceived flexible circuit board three-dimensional surface model after spatial registration; then, fusing the three-dimensional curvature features and the two-dimensional texture features extracted from the model to generate a multi-dimensional feature description vector; and finally, carrying out deviation calculation on the vector and a standard reference, automatically positioning and identifying surface flaws, mapping a flaw area to a three-dimensional model, and generating a detailed detection report containing quantitative characteristic parameters. According to the method, the spectrum material information and the three-dimensional shape information are fused, and geometric and texture combined feature analysis is adopted, so that automatic and high-precision positioning and quantitative evaluation of the tiny flaws on the surface of the flexible circuit board are realized.

The invention discloses a process behavior portrait system construction method based on resource use, and belongs to the technical field of portrait systems. The method comprises the steps that a target resource range used for constructing the portrait system is obtained, and resource use data in the process running period is screened out in the range; according to the method, a high-quality data input-comprehensive feature description-dynamic early warning output collaborative system is formed, reliable raw materials are provided for feature extraction through accurate data acquisition and preprocessing, and an accurate basis is provided for baseline construction through a comprehensive feature system; dynamic baseline and graded early warning ensure that the portrait system can adapt to process behavior changes in real time and efficiently respond to anomalies; through combination of the three, the accuracy, integrity and timeliness of the process behavior portrait are synchronously improved, and finally the core targets of early recognition of abnormal processes, resource scheduling optimization and strong guarantee of system stability in operation and maintenance of a computer system are achieved.

The invention provides a fault diagnosis method, device and equipment for auxiliary equipment of a thermal power plant, and relates to the technical field of fault diagnosis of thermal power generation equipment, and the method comprises the steps: obtaining a fault diagnosis request, inputted by a user, of the auxiliary equipment of the thermal power plant; wherein the fault diagnosis request comprises equipment parameters, multi-dimensional monitoring data and fault feature description of the auxiliary equipment; generating a retrieval vector based on the equipment parameters, the multi-dimensional monitoring data and the fault feature description; screening a plurality of target question and answer pairs from a pre-constructed thermal power plant field knowledge base, wherein the similarity between the target question and answer pairs and the retrieval vector is greater than a configured similarity threshold; fusing the target question and answer pair and the fault diagnosis request to obtain target input; and inputting the configured thermal power plant domain knowledge rule as a cue word together with the target input into a pre-trained fault diagnosis large model to obtain a fault diagnosis result corresponding to the fault diagnosis request. According to the invention, rapid, accurate and explainable intelligent diagnosis of the auxiliary machine equipment fault is realized.

The invention relates to the technical field of thunder and lightning mode recognition, and discloses a thunder and lightning intelligent target early warning method and system for a local key target, and the method comprises the steps: collecting target feature description, estimating storm motion information based on lightning positioning data and radar echo images at continuous moments, and carrying out the recognition of a thunder and lightning mode. And generating a risk corridor in the upstream direction of each target area, and forming a target area around the target area in combination with the risk corridor. Unified spatio-temporal data is obtained, spatio-temporal feature extraction is carried out to obtain storm features, cross-modal correlation selection and weighting are carried out on the storm features with target feature description as a drive, and target features are constructed. And generating an early warning result in a predetermined time window according to the target feature. And when the thunder risk probability reaches an early warning threshold and the spatial risk distribution intersects with the buffer range of the target area, generating targeted early warning information. According to the invention, thunder and lightning targeted early warning is realized, the hit rate is improved, and false alarms are reduced.

The invention relates to the technical field of steel pipe and copper pipe welding detection, and discloses a method for detecting welding defects of a steel pipe and a copper pipe. The method comprises the steps that original image data of a steel pipe and copper pipe welding area are collected, gray level conversion and noise suppression processing are conducted on the original image data, and preprocessed image data are generated; textural feature distribution and edge gradient distribution of the welding area are extracted based on the preprocessed image data, and a feature description matrix of the welding area is constructed; inputting the feature description matrix into a trained defect classification model, and outputting defect type probability distribution of the welding area; and generating positioning mark data of the welding defects according to the defect type probability distribution, and generating a visual detection result of the welding defects in combination with the original image data. The method can effectively eliminate image interference, comprehensively capture defect characteristics, improve defect identification precision, visually present detection results, adapt to steel pipe and copper pipe welding scene requirements, and meet industrial batch detection requirements.