969 results about "Template matching" patented technology

The present invention relates to an adaptive sensing-based lightweight monitoring method for a fine crack in a complex background region. The method comprises the following steps: step S1, on the basis of region division, performing automatic acquisition of crack information, wherein PTZ camera sensors are used to automatically perform block-wise acquisition on crack regions; step S2, performing an adaptive complex scale calibration process, using a multi-scale template matching algorithm to adaptively correct distortion information of all regions, and performing real-scale conversion from pixel precision; step S3, constructing a lightweight crack segmentation network to process data processed in step S2; and step S4, by means of a quantitative crack-tracking algorithm based on Euclidean distance similarity classification, performing real-time monitoring on each piece of crack dynamic information. Compared with the prior art, the present invention has advantages such as achieving efficient, accurate, and online monitoring and analysis of cracks.

The invention relates to an AI-based automatic adaptation method for a work approval process, and the method comprises the steps: collecting multi-source approval data, and carrying out the preprocessing of the multi-source approval data, and forming standardized approval data; analyzing the system rule text by using the pre-trained AI large model, and extracting a structured approval rule comprising a trigger condition, an approval role and a process node sequence; matching a basic examination and approval template according to the form field and the user permission information, and dynamically generating an adaptive process comprising multiple stages of examination and approval nodes, aging parameters and an additional examination and approval link; and through resource scheduling optimization and time domain correlation analysis, establishing an optimized mapping relation based on flow execution characteristics such as approval timeliness deviation and node skipping frequency, and outputting a visual flow chart and execution parameters. A business logic writing mode is replaced by automatic analysis of an AI large model on an unstructured rule; the dynamic template matching and continuous iterative optimization mechanism can quickly respond to business changes, and resource scheduling optimization and automatic process generation shorten the implementation period and reduce the operation and maintenance cost.

The invention discloses a bridge structure health monitoring data anomaly detection method based on deep learning, particularly relates to the technical field of structure health monitoring, and is used for solving the problems of high environmental interference sensitivity and insufficient cross-modal data fusion capability caused by image enhancement and feature extraction process splitting in the existing method. A cross-domain feature mapping relation is generated through combined training of dynamic image enhancement and a deep learning model, and collaborative optimization of enhancement parameters and feature space is achieved; time-frequency resonance parameters of visual images and acoustic emission signals are fused based on cross-modal convolution, and damage feature space distribution is corrected in combination with an attention mechanism; analyzing and quantifying the structural difference of the cross-domain features by using topology persistence coherence, and iteratively optimizing the feature mapping network through an optimal transmission theory; and finally, a multi-level feature template matching and self-adaptive threshold judgment mechanism is adopted to output an abnormal detection result, so that the robustness and generalization ability of bridge structure health detection in a complex environment are remarkably improved.

The invention belongs to the technical field of image processing and robots, and particularly relates to an image processing method and device and a robot, and the method comprises the steps: obtaining a structured light coding image, a polarization imaging image and a binocular depth image; constructing a multi-dimensional feature vector by fusing decoded contour phase, polarization analysis material characteristics and depth image data; dynamically calibrating the multi-dimensional feature vector according to the ambient light and the pose of the robot, correcting color deviation through an adaptive white balance algorithm, compensating a camera installation error, and obtaining a calibrated descriptor; based on the calibrated descriptor, utilizing an improved Hough transformation and template matching algorithm to identify the object category attitude, and calculating six-degree-of-freedom pose parameters through a triangulation method; and according to the six-degree-of-freedom pose parameters, a collision-free grabbing path is generated in combination with a robot kinematics model and obstacle information. Therefore, the problems of insufficient image analysis capability, single compensation mechanism, lack of path optimization capability and the like in the prior art are solved.

The invention relates to the technical field of machine vision, in particular to an intelligent machining positioning method and system based on machine vision, and the method comprises the steps: collecting and preprocessing original visual image data, and extracting geometric structure features and surface morphological features from the preprocessed visual image data, thereby comprehensively mining the characteristics of a workpiece; when the machining positioning offset vector is determined, the geometric structure features are matched with the preset machining reference template, and the workpiece machining position deviation can be preliminarily determined; on the basis of surface morphological characteristics, compensation adjustment is carried out on a machining positioning offset vector to generate an optimized positioning vector, the complex condition of the workpiece surface is considered, positioning deviation possibly generated according to a geometric structure purely is corrected, and positioning is more accurate; a machining path adjusting instruction is generated according to the optimized positioning vector and transmitted to a machining control unit, machining equipment can be effectively guided to adjust the machining path according to the actual state of the workpiece, it is ensured that the machining process is accurate and efficient, and the workpiece machining quality and production efficiency are improved.

The invention discloses a document element extraction method based on an AI large model technology. The method comprises the following steps that various receipts input by a user are received; carrying out analysis and layout analysis on the document through a visual language model or a text analysis technology to generate a unified intermediate representation; a dynamic grading module is adopted to evaluate document complexity from three dimensions of format, structure and variation, and the document complexity is divided into a simple regular type, a structure complex type or a height variation type; adaptively selecting a processing flow according to a rating result, wherein the processing flow comprises high-speed template matching, multi-modal cognitive fusion or intelligent agent driving processing; the processing result is converted into an EasyEX format which is easy to process; the method comprises the following steps of: converting a natural language demand into an execution rule by extracting an Agent and utilizing a Prompt dynamic compiling technology, and positioning and extracting a target field by LLM (Logical Language Model); and finally, outputting structured data after type verification, knowledge graph verification and compliance review. According to the invention, the universality, the accuracy and the automation degree of receipt processing are improved, and the labor cost is greatly reduced.

The invention discloses a textile production defect detection method and system based on image processing, and belongs to the field of textile defect detection. According to the method, the point cloud data is generated by collecting the surface deformation stripes, and the penetrability reflection signal containing the sound wave propagation time is obtained through the multi-frequency ultrasonic probe. Denoising the point cloud data and then positioning a surface deformation candidate area based on a curvature gradient and a normal vector deflection angle; and an internal abnormal region is determined by separating and extracting a reflection wave amplitude value and a time delay parameter through an ultrasonic frequency band. And matching the surface point cloud coordinates with the ultrasonic propagation time by adopting an iterative nearest point algorithm to establish feature association between the normal vector deflection angle and the reflected wave amplitude. And dynamically distributing fusion weights for the surface features and the internal features according to the curvature gradient change rate and the amplitude attenuation rate to generate a multi-modal defect distribution diagram. And geometric shape and reflection intensity parameters are extracted and matched with a preset defect template to realize type and position judgment, and the method can be suitable for production defect detection of the thick textile with a multi-layer composite material structure.

Disclosed herein are a method, an apparatus and a storage medium for image encoding / decoding. In typical image encoding / decoding methods, a decoder-side motion information derivation method may be limitedly used. Therefore, the improvement of encoding efficiency attributable to the decoder-side motion information derivation method may also be limited. In embodiments, a motion information search method used in an inter-prediction mode, an intra block copy mode and an intra template matching prediction mode is disclosed. With the use of various motion search methods, encoding efficiency may be improved.

The invention discloses a fixed dual-runway FOD detection method based on multi-source circumferential scanning, and belongs to the technical field of data fusion detection, and the method specifically comprises the steps: synchronously obtaining dual-runway multi-polarization echo data through a fixed circumferential scanning rada; performing time-varying clutter suppression by combining an adaptive clutter cancellation method matched with a dynamic background template and a runway material feature library, and generating a time-frequency domain three-dimensional feature matrix through time-frequency analysis; synchronously collecting and preprocessing image data, and marking a target area through a constant false alarm rate algorithm and image detection; a multi-source data geographic coordinate mapping model is established, data registration and cross validation are completed under a runway physical coordinate system, and real-time tracking of a real FOD target is realized based on radar and image fusion data through multi-dimensional confirmation of radar polarization / micro-motion features and image visual features; according to the method, the FOD detection accuracy and the position identification precision are improved through multi-source data fusion, the false alarm rate is effectively reduced, and a guarantee is provided for safe operation of an airport.

The invention relates to the technical field of intelligent monitoring, and discloses an intelligent archival repository monitoring system based on Internet of Things, comprising: a plurality of edge sensing units deployed in a repository, monitoring parameter drift in real time and triggering a local perturbation event by dynamically learning a personalized stable state baseline of environmental parameters; cooperative judgment is achieved among the units through perturbation wake-up signals, concerned events are reported to a central platform only when preset conditions are met, and full-spectrum perception of slowly changing and sudden risks is formed through baseline learning and a cooperative wake-up mechanism of the edge units in combination with instantaneous event template matching and targeted enhanced monitoring; based on the dynamic health fingerprint analysis of the perturbation resonance characteristic spectrum, the stability degradation trend of the archive microenvironment can be pre-warned in a prospective manner, and the intelligent level and the preventive protection capability of the monitoring system are remarkably improved.

The invention relates to the technical field of artificial intelligence, can be applied to business scenes such as financial science and technology and medical health, and discloses an image generation method, device and equipment based on theme information and a medium. The method comprises the steps that input information is analyzed to generate theme information and copywriting information, a cue word set is generated, and a figure image set and a background image set are generated; fitting the segmented figure image with the background image to form a head image candidate set, and selecting a head image matching template frame to generate a basic image; decomposing the copywriting to generate a sub-module initial picture set, and adding a gradual change effect to form a sub-module picture set; the adjusted sub-module pictures are obtained based on size adjustment, and a pre-synthesized image is generated through splicing; and identifying the blank area to draw a title text to obtain a final image. Through theme analysis, template matching, image splicing, modular copywriting processing and blank drawing, poster generation efficiency is improved, layout flexibility is enhanced, and visual unification is realized.

The invention provides a nuclear power safety report data extraction method and system based on multi-modal feature fusion, and belongs to the technical field of data processing, and the method comprises the steps: obtaining a nuclear power safety report document; carrying out OCR (Optical Character Recognition) and paragraph segmentation on the report document to obtain first text data; performing text semantic understanding on the first text data by adopting a dynamic template matching and semantic driving extraction mechanism to obtain a text feature vector; segmenting pages of the report document and identifying fonts and fonts to generate first format data; adopting a cross-page table reconstruction algorithm to detect table cells of the first format data and splicing a cross-page table to obtain format feature vectors; performing feature alignment on the text feature vector and the format feature vector, and inputting the text feature vector and the format feature vector into a large pre-training language model to obtain a semantic vector; and fusing the text feature vector, the format feature vector and the semantic vector to form a composite expression unit, and intelligently extracting structured data based on the composite expression unit. The semantic entity and structural relationship in the nuclear power report can be effectively identified.

The invention discloses an unstructured file identification method based on primitive identification, and the method comprises the following steps: (1) preprocessing: cutting and zooming an electrical wiring drawing to a required size, and carrying out the gray processing; (2) primitive recognition: constructing a feature extraction network of an electrical primitive detection model by adopting a YOLO electrical primitive detection algorithm fused with an attention mechanism; (3) character recognition, which comprises the following steps: removing primitives, detecting an annotated text area, constructing a text candidate box, segmenting and adjusting an annotated text boundary, recognizing annotated text content, and filtering an electrical text recognition result; and (4) electrical element association relationship analysis, which comprises the following steps: frame element extraction, primitive group region division, template matching and electrical primitive-annotation text association. According to the method, the primitives and characters of the electrical wiring diagram can be obtained, the problem that different fonts and styles appear in marked characters can be solved, and semantic information can be understood.

The invention discloses a concave-convex point detection and defect evaluation system based on multistage detection and feature change, and the system comprises an image collection unit which is used for capturing a product image in real time; the image preprocessing unit is used for carrying out noise reduction processing on the image and improving the brightness and the contrast ratio of the image; the primary product defect positioning module is used for selecting Gaussian difference extraction, dynamic threshold extraction, fixed threshold extraction, template matching and a deep learning algorithm module to extract and position suspicious defects according to the size of the product and the characteristic form of the defects; the secondary defect classification and judgment module is used for carrying out preliminary classification and type judgment on the defects in the region; the third-stage defect contour extraction unit is used for performing fine contour extraction on each detected defect area after the defect classification is clear; and the four-stage defect feature analysis and evaluation unit is used for deeply extracting key features of each defect and carrying out scientific and quantitative evaluation on the key features.

Disclosed herein are a method, an apparatus and a storage medium for image encoding / decoding. Methods for performing more accurate prediction are disclosed when performing prediction using a block vector. A block vector for a target block is derived by template matching, and a prediction block for the target block is determined based on the derived block vector. Various and detailed embodiments for a plurality of pieces of information used to derive a block vector, such as a search area, a block vector candidate list, a reference region, a buffer, and a tree type, are provided.

The invention relates to a packaging defect detection method and system, and the method comprises the steps: S1, obtaining the size information of packaging bags and the interval information of adjacent packaging bags through an area-array camera and a photoelectric sensor in combination with a multi-scale template matching method; s2, delay time of a conveying belt is calculated, and it is ensured that the distance between adjacent packaging bags is larger than a set value; s3, triggering a line scanning camera to shoot an image of the packaging bag, and collecting the vibration frequency; s31, performing motion compensation correction on the image by adopting an improved optical flow method; s32, inputting the corrected image into a forward propagation network based on an improved OfficientViT model, wherein the model comprises a multi-scale linear attention module and an SPP-ELan module; s33, the images are classified, if the classification result is that the bags are not packaged, the unpackaged bags are grabbed through a grabbing robot, and if it is determined that the bags are not packaged based on a classification confidence coefficient threshold value, the unpackaged bags are removed; and S34, inputting the defect image into an improved ACGAN network for data enhancement, wherein the network comprises a CMFNet denoising module based on a U-Net architecture. The method and the system can monitor and identify the packaging defects on the production line in real time, improve the accuracy and the efficiency of inspection, and reduce the labor cost.

The invention relates to the field of image recognition, and provides a positive and negative sample algorithm combined power transmission line high-altitude anomaly detection method. The objective of the invention is to improve the accuracy and robustness of detection. By comprehensively utilizing the advantages of the positive sample algorithm and the negative sample algorithm, the method effectively solves the problems of background complexity and abnormal data collection. Firstly, a detection picture and a base picture are cut and masked and focused on a power transmission line area; next, a positive sample algorithm is used to identify an abnormal region, while a negative sample algorithm is focused on detecting a specific type of anomaly. Through a multi-stage filtering mechanism including target detection filtering and template matching filtering, false alarms are effectively reduced, and the detection effect is improved. According to the method, a memseg algorithm and a YOLO model are adopted in model training, and an anomaly filtering model is introduced to detect common misinformation targets such as birds and vegetation. Finally, by combining positive and negative sample results, high-precision and high-recall-rate detection performance is realized.

The invention discloses an AI-based archive digitization full-process management and control method, which comprises the following steps: acquiring archive image data and text data, constructing an original archive data set, and generating an initial directory set; constructing an archive directory template set, executing template matching and structure adaptation, and generating an initial structure recognition result; setting a classification rule of directory deviation and structural contradiction, constructing a deviation rule base, executing rule scanning and node comparison, and generating a deviation contradiction label sequence; constructing an evolution state vector, inputting the evolution state vector into an improved GMN model, and generating a structure repair state vector through a structure guiding unit, a deviation sensing unit, a path fusion unit and a node evolution unit in sequence; and executing result updating and conflict rearrangement, completing standard template alignment and semantic coverage verification, and generating a final structure recognition result. According to the method, the structural integrity, the semantic consistency and the automatic repair capability of directory recognition are improved, and the method has relatively high adaptability and intelligent level.

Embodiments of the present disclosure provide a solution for video processing. A method for video processing is proposed. In the method, for a conversion between a current video block of a video and a bitstream of the video, a process is applied to the current video block based on template matching. At least one reference sample of a current template of the current video block is determined based on a block vector (BV) of the current video block during the process. The conversion is performed based on the applying.

The invention provides an automobile part visual inspection correction method and equipment. The method comprises the following steps: extracting a part contour through edge detection, calculating an inclination angle by adopting Hough transform, executing affine transform on an original image, enabling a rotation angle to be negative to the inclination angle, and ensuring that the parts are horizontally aligned; in the corrected horizontal image, calculating a transverse offset through a feature matching algorithm or template matching, and translating an image area to enable the center of the detection window to be aligned with the actual position of the part; if inclination and offset are detected at the same time, inclination correction is executed preferentially, the offset is calculated based on the image after inclination correction, synchronous correction is achieved through a composite transformation algorithm, and accumulative errors are reduced through iterative optimization; and extracting a size parameter from the corrected image, comparing the size parameter with a predetermined tolerance range, and if the measured size is outside the predetermined tolerance range, marking the image as a defective product, and removing the defective product. The method is used for upgrading the visual inspection system for automobile manufacturing.

The invention provides a real-time operation remote video consultation system and method based on virtual reality, and the system comprises a driving data collection module which collects multi-modal data; the data processing module is used for carrying out fusion processing on data from different acquisition sources; the virtual image driving module is used for reconstructing a virtual scene of an operation site and performing three-dimensional modeling and rendering on an image acquired by the high-definition camera and an internal structure image acquired by the medical imaging equipment; the template matching module is used for collecting and sorting standard action templates of common surgical operations and matching the standard action templates; the data collaboration module is fused with the virtual scene to display the medical data of the patient; and the 5G robot assistance module remotely controls the robot to execute an action corresponding to the virtual reality scene through 5G. According to the invention, audio and scene fusion is realized, scene fidelity is improved, and consultation immersion is enhanced. Interaction between experts and virtual scenes in consultation can be achieved, consultation efficiency and sense of participation are improved, doctors are assisted in standard operation, and surgical risks are reduced.

The invention discloses a microscopic image scanning and parallel efficient splicing method based on a spanning tree, which considers the characteristic that the scanning visual angle of a microscope is distortionless and the characteristic that the corresponding relation between the physical scale and the pixel scale of a microscope platform is unchanged, and has the advantages of an image matching algorithm based on a template. Automatically dividing an overlapping region range for a template matching algorithm, fully considering the matching result credibility of the pictures on the surrounding adjacent sides, and calculating a global splicing point location by using a spanning tree algorithm; picture splicing types are divided through rows and columns, the problem of memory access conflicts is avoided, the computing power improvement brought by multiple threads is fully utilized, and the picture splicing efficiency is improved. According to the method, adjacent image matching, high-precision matching of global splicing point locations based on the spanning tree and a multi-thread high-speed image splicing technology are realized.

The invention discloses an instruction-level code optimization method and device based on template matching, a medium and equipment. According to the method, instruction-level codes serve as input, the input instruction-level codes are firstly constructed into corresponding control flow diagrams, then value ranges of active variables and register quantities are initialized based on the control flow diagrams, then variable assignment instructions and related instructions are found out from instruction sequences of basic blocks to form instruction segments to be optimized, and the instruction segments to be optimized are optimized. The method comprises the following steps of: optimizing an instruction fragment, obtaining an optimized instruction fragment through template matching, performing SMT equivalence verification on the instruction fragment before and after optimization based on an active variable and a value range of a register quantity, and finally optimizing an input instruction-level code in a manner of replacing the instruction fragment before optimization with the optimized instruction fragment which is verified to be equivalent. And finally, the optimized instruction-level code is obtained.

The invention discloses a target tracking method and device based on a rolling and pitching type holder, and the method comprises the steps: building a polynomial response model through adaptive color calibration, and achieving the dynamic color correction; an improved YOLOv8 model is combined with a bidirectional feature pyramid network to carry out multi-scale target detection, and template matching is assisted to improve a low-confidence-coefficient scene recognition rate; predicting the position and the speed of the target by using Kalman filtering to cope with a shielding condition; and high-precision smooth rotation of the holder is realized through quaternion attitude solution in combination with PID (Proportion Integration Differentiation) and FOC (Fiber Operating Control) double-loop cooperative control. The device adopts a rolling and pitching type holder structure and comprises a main controller, an angle sensor, a visual sensor and a magnetic encoder, the mechanical structure is simplified, the weight is reduced, and higher response speed and higher control precision are achieved. The target tracking accuracy, stability and anti-interference capability of the intelligent mobile platform in a complex environment are remarkably improved, and the method is suitable for various dynamic scenes such as aerial photography and industrial inspection.

The invention discloses a confidential evaluation compliance document automatic generation system and method based on strategic data management and control and full-link verification, and aims at solving the problems that traditional confidential evaluation document generation is low in efficiency, insufficient in safety protection, poor in expansibility and the like. According to the method, hard coding logic is replaced by a strategy engine center, a scene-template-strategy association mechanism is constructed, and full-process automatic generation of secret evaluation documents is achieved through scene configuration, intelligent template matching, parameter analysis, multi-dimensional data cleaning and desensitization, national secret standard automatic signing and full-process log management and control. The generation efficiency is remarkably improved, a sensitive data'identification-processing-protection 'safety closed loop is formed, the tamper-proof and auditing requirements of secret evaluation are met, a newly-added scene can be quickly accessed through strategy configuration, and the expansibility and reusability are excellent.

The invention provides a defect detection method of a printed circuit board and electronic equipment, which are used for improving the detection accuracy of the printed circuit board. Comprising the steps that a detection image of a target printed circuit board is acquired, and the detection image comprises local detection images; determining a target defect detection strategy corresponding to the element type of the element in the first local detection image by using a preset corresponding relationship between the element type and a defect detection strategy, the defect detection strategy including a template matching strategy, a connected region analysis strategy, a color comparison strategy and a deep learning strategy; the target defect detection strategy is at least one of the defect detection strategies, and the first local detection image is any one of the local detection images; performing defect detection on the first local detection image by using a target defect detection strategy to obtain a defect detection result of the first local detection image; and obtaining a defect detection result of the target printed circuit board according to the defect detection result of each local detection image.

The invention relates to a template-based self-adaptive stack type generation method, system and device and a storage medium, a preset stack type layout template and to-be-stacked material information are obtained, and relative position information and stack type information of each material in the stack type layout template are calculated; according to the information of the to-be-stacked materials, the relative position information and the stack type information, the to-be-stacked materials and the stack type layout template are matched, so that the to-be-stacked materials are arranged according to the stack type layout template, and a single-layer stack type is obtained; the multiple single-layer stack types are stacked according to a preset merging rule, and an overall stacking stack type is generated; and a preset material gap is obtained, arrangement of the materials in the overall stacking type is adjusted according to the preset material gap, and an actual stacking type is generated. Through template matching and automatic calculation, new stack types can be generated for materials of different sizes in a short time, repeated programming and debugging are not needed, and the time cost is reduced.

The invention provides a two-stage SAR (Synthetic Aperture Radar) image registration method based on SAR-SIFT (Scale Invariant Feature Transform) and template matching, which comprises a pre-registration stage and a fine registration stage, and is characterized in that in the pre-registration stage, a down-sampling image is calculated through an SAR-SIFT algorithm to obtain an initial registration image; updating the initial registration image to obtain a final matching pair; after the final matching pair is mapped back to the original image size, a pre-registration image is obtained according to the final affine transformation matrix; in the fine registration stage, stable feature points are extracted from the pre-registration image blocks, and mismatching pairs are eliminated to obtain a fine registration image; and judging whether a region with relatively large topographic relief exists through the clustering mismatching pair, sampling a candidate topographic relief region, and then carrying out local correction to obtain a fine registration image again. The SAR-SIFT algorithm and iterative optimization are utilized to realize initial registration, a similarity measurement method is utilized to realize accurate registration of a flat region, adaptive stretching transformation is utilized to transform a topographic relief region to realize accurate registration of the topographic relief region, and finally accurate registration of the SAR image is realized.

The invention provides an aluminum strip shearing fine adjustment structure based on visual data, and relates to the technical field of aluminum strip shearing, the aluminum strip shearing fine adjustment structure comprises a shearing table, a material receiving roller set is installed in the middle of the shearing table, and a first material rolling roller set and a second material rolling roller set are installed on the two sides, close to the material receiving roller set, in the shearing table correspondingly; a smoothing roller set is installed at the bottom, close to one side of the material collecting roller set, in the shearing table, the smoothing roller set is attached to the material collecting roller set, and a material passing roller set is installed at the bottom, close to one side of the smoothing roller set, in the shearing table. According to the scheme, the cutter abrasion monitoring probe is used for shooting the surface of the cutter in real time in five frames per second, the main control machine is combined with template matching and a deep learning model to accurately recognize abrasion, when the depth of a notch is larger than 0.2 mm or the width of a cutting edge is increased by more than 5%, the cutter sleeve is automatically triggered to rotate by 180 degrees, and cutter replacement is completed through cooperation of the micro air cylinder, the micro motor and the servo motor.

The invention relates to the technical field of industrial vision multispectral imaging detection, in particular to a defect detection method and system based on multispectral fusion imaging, and the method comprises the steps: synchronously collecting a multi-channel spectral image of the surface of an object to be detected, and generating an enhanced feature map fusing multispectral information; identifying and preliminarily marking a suspected defect region by applying a region segmentation algorithm based on anomaly detection; extracting a multi-dimensional spectral response curve of each region in each original spectral channel, and constructing a spectral feature vector; matching the vector with a standard template of a pre-established defect spectral feature database, and classifying and confirming defect types; and carrying out contour refined analysis, calculating the geometric dimension and position of the defect, and generating a structured detection report. According to the method, the sensitivity of defect detection and the classification accuracy are improved through accurate matching of multispectral fusion and spectral features, and automatic and high-precision defect identification is realized.