361 results about "Geometric transformation" patented technology

The invention relates to a hydraulic engineering concrete crack intelligent identification and quantitative analysis method and system based on machine vision. The method comprises the following steps: firstly, acquiring an original image sequence of the surface of a hydraulic engineering concrete structure, classifying according to illumination intensity, shooting angle and shooting distance, extracting crack edge features through a convolutional neural network, and fusing to obtain a crack feature set; correcting illumination through adaptive histogram equalization, correcting angles and distances through geometric transformation, and combining edge detection and scale invariant feature transformation to obtain standardized geometric parameters including crack length, maximum width and the like; if the parameter exceeds the engineering safety standard threshold value, tracking a crack track through an optical flow method to calculate increment, and inputting a neural network to output a damage trend; and finally, generating a three-color risk distribution diagram by using a finite element based on the trend, extracting high-risk data to calculate a real-time evaluation value, and dynamically adjusting the monitoring frequency to generate an optimization strategy. By adopting the method, the reliability and economy of engineering safety monitoring can be remarkably improved.

The invention discloses a pavement crack detection method based on a Yolov8 model, and belongs to the technical field of crack detection. Comprising the following steps: constructing a pavement crack segmentation data set: acquiring a road crack image through an unmanned aerial vehicle, performing data enhancement processing of geometric transformation and color transformation on the image in combination with a public data set, performing Gaussian filtering denoising on a noise image, and performing crack labeling by using Label; based on a YOLOv8-Seg model, improvement is carried out by introducing a PKIblock multi-scale convolution kernel, generalizing an efficient layer aggregation network GELAN module, an EMA attention mechanism and replacing a spatial pyramid pooling layer SPPF into a SimSPPF, and a road crack recognition model YOLOv8-RCI is constructed; and performing crack detection and instance segmentation on the unmanned aerial vehicle image by using the trained YOLOv8-RCI model, and outputting a crack position and mask information. Through the improvement in the four aspects, the detection segmentation performance of the model is improved, and the model meets the requirement of real-time detection.

The invention discloses an automobile part quality detection method and system based on artificial intelligence visual inspection, and belongs to the field of artificial intelligence machine visual inspection, and the method comprises the steps: firstly, carrying out the registration of a collected RGB image and a depth image, and extracting a part region through a saliency detection network; two-dimensional key points are extracted based on an RGB region image and are matched with key points of a three-dimensional model, an initial three-dimensional attitude is obtained by adopting a PnP algorithm, iterative registration is performed with the three-dimensional model in combination with a point cloud generated by a depth image, and a fine three-dimensional attitude is obtained. And calculating a geometric transformation matrix from the part to a standard front view attitude according to the attitude, and performing attitude correction on the RGB and depth region image. And then matching the corrected image with a standard template image by using a feature detection and matching network so as to correct the position of the detection window. And finally, the three-dimensional size of the part is calculated in the corrected detection window in combination with the depth value, and tolerance judgment is carried out. And the precision, the robustness and the automation level of online detection of the automobile parts can be obviously improved.

The invention provides a face image recognition method and system, and relates to the technical field of data processing, and the method comprises the steps: carrying out the fusion of a topological relation feature vector and a biological feature contour, calculating the curvature manifold of the concave region of five sense organs, and generating a local curvature feature vector through differential geometric transformation; inputting the topological relation feature vector, the local curvature feature vector and the micro feature coding sequence into an evolvable biological feature library, and fusing a space-time weight to update a comparison template; fusing the micro-feature coding sequence, the topological relation feature vector and the local curvature feature vector to obtain a biological feature set, and calculating the multi-scale similarity between the biological feature set and the updated comparison template; and if the multi-scale similarity exceeds an adaptive threshold, generating a face recognition result. According to the invention, the stability and accuracy of face recognition in a complex illumination environment are improved.

The invention belongs to the technical field of defect identification, and particularly relates to a gas turbine blade defect identification method based on an improved YOLOV8 network. Comprising the following steps: capturing a defect image; effective information is extracted from the image or subsequent target detection, classification and segmentation task requirements are met; performing data enhancement through geometric transformation and color perturbation; adding structured labels or annotations to the images, and endowing the images with semantic information; the scheme is improved on the basis of YOLOv8 so as to be realized in gas turbine blade defect detection, and training of the model is completed in a distributed heterogeneous computing framework by using an acquired training data set and an acquired verification data set and is used for a prediction task. The method can achieve the pixel-level detection and positioning of the defects of the blade, can effectively meet the daily detection demands, and gives consideration to the detection precision and real-time performance.

The invention discloses a real-time data stream processing method for multi-device collaborative management of a dome cinema, which relates to the technical field of cinema device data management, and comprises the following steps of: establishing communication connection between a central management node and a plurality of playing devices, collecting device identifiers, projection area numbers and geometric position parameters of the playing devices, and establishing a central management node; generating an equipment mapping relation table; realizing local clock correction of the playing equipment based on a unified time synchronization protocol; the central management node generates a control instruction stream according to the equipment mapping relation and the playing content, the instruction stream comprises space frame distribution information and fusion control parameters, and the space frame distribution information and the fusion control parameters are distributed to the playing equipment through communication channel numbers; and the playing device completes image geometric transformation and edge fusion processing according to the received control instruction and then executes synchronous playing. According to the invention, multi-device automatic mapping is realized, the time synchronization precision is improved, the image playing is coordinated and consistent, and the collaborative display capability and the playing stability of the dome cinema are improved.

The invention discloses a multi-scale linear array camera splicing method and system based on point cloud. The method comprises the following steps: completing acquisition and preprocessing of point cloud data and image data of a target area; determining an overlapping region range between adjacent images; extracting spatial structure characteristics in the point cloud data, and performing multi-scale hierarchical decomposition on the point cloud through a multi-scale segmentation method; meanwhile, multi-scale image feature extraction is carried out on the images of the linear array camera; solving gradients in X and Y directions by adopting an optical flow method aiming at any pixel in the overlapping region, and calculating a motion vector between the pixels; fusing the optical flow information obtained under each scale, and constructing a globally consistent optical flow vector field; according to the fused optical flow vector, calculating to obtain a geometric transformation matrix of the whole overlapping region; and after image transformation and alignment are completed through the transformation matrix, fusion processing is carried out on overlapped areas. And the unification of the visual effect and the spatial integrity of the spliced image is ensured.

The invention relates to the technical field of image processing, in particular to a rubber ring contour burr detection method and system based on geometric transformation, and the method comprises the steps: obtaining a rubber ring image, and extracting the outer contour of a rubber ring and the contour of each burr in the rubber ring image; fitting the outer contour of the rubber ring by adopting a least square method to obtain an ellipse, and mapping the ellipse into a standard circle: carrying out bilinear interpolation resampling on the rubber ring image, and carrying out polar coordinate transformation on the rubber ring image by taking the circle center of the standard circle as an original point to generate an expanded image; generating a reconstructed image with enhanced burr features from the expanded image through wavelet transform, determining burr parameters of the reconstructed image, mapping the burr parameters back to a Cartesian coordinate system of the rubber ring image, and generating a detection result containing burr number, position and size information; according to the invention, the accuracy and robustness of tiny burr detection can be improved.

The invention relates to a GIS (Gas Insulated Switchgear) equipment mechanical defect diagnosis method based on a Gramb angle field and a dual-channel PCNN-Attention neural network, and belongs to the technical field of gas insulated switchgear mechanical vibration defect diagnosis. The method solves the problems that traditional diagnosis depends on artificial feature extraction, so that subjectivity is high, information mining is insufficient, and defect severity evaluation is missing. According to the technical scheme, the method comprises the steps that a one-dimensional vibration signal is converted into a GASF two-dimensional image and a GADF two-dimensional image through a GASF field so as to completely reserve time sequence topological features; carrying out data enhancement by adopting an image geometric transformation technology so as to improve the generalization ability of the model; and a dual-channel PCNN-Attention model is constructed, and synchronous intelligent identification of defect types and severity is realized through parallel feature extraction and dynamic weight optimization of an attention mechanism. According to the method, the diagnosis accuracy, reliability and adaptive capacity are improved, and support is provided for equipment state operation and maintenance.

The invention discloses an airborne visible light image automatic splicing method, and relates to the field of unmanned aerial vehicle image processing, and the method comprises the steps: obtaining a plurality of images collected in the flight process of an unmanned aerial vehicle, and the corresponding spatial position information and attitude information; on the basis of the spatial position relation between the images and the image overlapping information, determining image pairs capable of being registered, and constructing a connection map with the images as nodes and the image pairs capable of being registered as edges; detecting whether spatial connection fracture caused by image missing exists in the atlas or not, if so, inserting a virtual node and constructing a virtual image comprising an edge region and a transition region; and further calculating geometric transformation parameters between the images, resampling all image contents to a unified coordinate system, and executing pixel-level fusion processing in an image overlapping region. According to the invention, the problem of discontinuous image splicing of the unmanned aerial vehicle due to the existence of a no-fly zone, a privacy protection zone and the like of the unmanned aerial vehicle is solved.

The invention relates to the field of projection image calibration, in particular to a self-adaptive calibration method and device of a projection touch system. The method comprises the following steps: identifying pre-projection image information, carrying out projection area layout design, and outputting a pattern layout projection effect; performing pattern positioning calculation on the pattern layout projection effect one by one, and generating a sub-pixel positioning coordinate of each pattern; calculating a central point of a projection area according to the sub-pixel positioning coordinates, and constructing an actual projection positioning coordinate system; theoretical position deviation calculation is carried out according to the actual projection positioning coordinate system, and projection pixel point position deviation of each pattern is extracted; and performing multi-stage progressive pattern calibration compensation based on the projection pixel point position deviation to obtain a geometric transformation compensation result. The pattern position precision of the projection touch control image is improved, and the visual effect and precision of the projection pattern are optimized.

The invention relates to the technical field of medical instruments, and discloses a tracheotomy anatomical structure image recognition method and system, and the method comprises the steps: obtaining real-time visual image data and spatial positioning data of an endoscope, carrying out the time synchronization, carrying out the mode recognition of the spatial positioning data, and locking the remarkable anatomical features in the real-time visual image data, and according to the characteristics and the spatial positioning data, calculating a geometric transformation relationship between an endoscope visual coordinate system and a spatial positioning coordinate system, and finally fusing the data and displaying anatomical structure information and a surgical tool position in real time. The method can effectively solve the problem that in the prior art, an endoscope is irregularly deformed in the narrow trachea with physiological bending of a patient.

The invention relates to the technical field of image updating, in particular to a multi-modal image real-time updating method and system for temporal bone surgery. The method comprises the following steps: acquiring a real-time multi-modal image, performing pixel-by-pixel frequency domain reconstruction optimization, and constructing a frequency spectrum enhanced fusion image; performing reverse geometric transformation compensation on the spectrum enhancement fusion image to obtain a space alignment optimization image; performing real-time visual contrast enhancement on the space alignment optimization image, and constructing a visual enhancement image; carrying out inter-frame difference calculation on the vision enhanced image, carrying out real-time increment updating optimization, and constructing an increment updating image sequence; and performing multi-level cache rendering management and parallel execution based on the incremental updating image sequence. The temporal bone surgery safety and efficiency are improved through real-time and efficient image updating.

The invention relates to an unmanned aerial vehicle indoor three-dimensional reconstruction autonomous acquisition method and system based on reinforcement learning, and the method comprises the following steps: constructing a virtual indoor simulation space and a virtual unmanned aerial vehicle model, and laying target acquisition points; a reinforcement learning model is constructed and trained, in the training process, in each time step, the unmanned aerial vehicle obtains target collection point information and inputs the target collection point information into the reinforcement learning model, the reinforcement learning model generates an action instruction and transmits the action instruction to the virtual unmanned aerial vehicle model, and the virtual unmanned aerial vehicle executes geometric transformation, updates the position and orientation and reconstructs view information; and deploying the trained reinforcement learning model in an untrained virtual simulation environment, carrying out an unmanned aerial vehicle autonomous flight experiment and data acquisition, and carrying out quantitative evaluation on the acquisition performance through a plurality of performance indexes. Compared with the prior art, efficient and stable data acquisition can be realized in a complex environment.

The invention relates to the technical field of Internet television services, in particular to an OTT visual feature extraction system and method based on multi-modal agent driving, and the method comprises the steps: capturing a screen real-time video stream of equipment; processing the target advertisement image and the real-time video stream, and extracting double-flow heterogeneous visual features, including global content perception features and local geometric structure features, through a multi-modal visual perception model; executing a hierarchical matching algorithm, calculating and screening out candidate frames by using global content perception features, matching in the candidate frames by using local geometric structure features to establish a corresponding relation set containing all matched initial key points, performing spatial clustering on the set to separate out advertisement instances, and obtaining bounding boxes of the instances through geometric transformation calculation; and according to the bounding box, performing highlight display on the area where the target advertisement is located on the original video frame to generate a visual broadcast monitoring result. According to the invention, through multi-mode intelligent body driving, OTT advertisement visual feature extraction and broadcast monitoring are realized.

The invention relates to the technical field of optical measurement, and discloses a biaxial galvanometer error calibration method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining a camera internal reference containing an internal reference matrix and a distortion coefficient through a calibration board; establishing an ideal geometric transformation model of the biaxial galvanometer, and defining a first galvanometer transformation matrix determined by the rotation angle of the first galvanometer and the first distance and a second galvanometer transformation matrix determined by the rotation angle of the second galvanometer and the second distance; installing error parameters are introduced, and a complete coordinate transformation model used for describing a complete light path transformation relation from an actual camera coordinate system to a final imaging coordinate system is formed; calibration data are collected, galvanometer system parameters and installation error parameters are optimized based on re-projection errors, re-projection residual errors are constructed, the re-projection residual errors are minimized through a nonlinear optimization algorithm, and optimal parameters are obtained. The installation error can be accurately calibrated and effectively compensated.

The invention discloses a simulation data analysis method for long-distance tunneling of a shield in a complex stratum, and relates to the technical field of underground engineering and intelligent construction, and the method comprises the following steps: multi-source data fusion and dynamic modeling: integrating geological exploration data, real-time shield construction sensor data and historical engineering data, constructing and dynamically updating a digital twinborn body in the shield tunneling process, wherein the digital twinborn body is used for simulating mechanical response in the tunneling process in real time; mechanical characteristic enhanced extraction: extracting a stress-strain tensor field of a key part based on a simulation result of the digital twin, and performing differential geometric transformation on the tensor field; by constructing a system of multi-source data fusion, digital twin dynamic modeling, differential geometric mechanics feature extraction, three-dimensional risk intelligent identification, multi-target parameter optimization and visual closed-loop execution, the pain points of traditional shield analysis data fragmentation, feature surface and difficulty in risk positioning are broken through.

The invention relates to the technical field of computer vision and pattern recognition, and discloses a multi-component object matching method and system based on graph matching. The method comprises the following steps: identifying each discrete component and component type label in a to-be-detected object image; forming a hierarchical set structure corresponding to the hierarchical relationship of the preset template; establishing undirected full-connection edges between adjacent levels, and filtering invalid edges to construct a dynamic graph; searching candidate sub-graphs matched with the topological structure of the preset template in the dynamic graph; calculating an affine transformation matrix between each candidate subgraph and a preset template; calculating a matching degree between the remaining candidate sub-images and a preset template, taking a matching result of which the matching degree exceeds a preset threshold value as effective matching, and outputting an object position coordinate and a geometric transformation parameter corresponding to each effective matching; according to the method, deep learning detection and graph matching technologies are fused, and hierarchical dynamic graph construction and backtracking affine two-stage verification are combined, so that the detection rate and the processing efficiency are greatly improved, and the mismatching rate is effectively reduced.

The invention relates to a navigation mark type and health state detection method based on improved YOLOv9, and belongs to the technical field of computer vision and navigation mark inspection. According to the method, in order to solve the problem of insufficient navigation mark data sets, a shape-damaged navigation mark data set is generated by using a CGAN technology, a navigation mark data set containing stains is generated through expansion by using a Canny algorithm, and a skew navigation mark data set is increased by using an affine transformation algorithm; in addition, traditional data enhancement methods such as geometric transformation and image splicing are introduced to simulate the imaging effects of the navigation mark at various angles and positions, so that the diversity of the image is improved. And the improved YOLOv9 model is utilized to improve the recognition accuracy of the small-size navigation mark. According to the method, the high-efficiency performance is maintained, meanwhile, the detection capability on fine and complex features is greatly improved, and accurate recognition on the navigation mark under different conditions is ensured.

The invention relates to an unmanned aerial vehicle image automatic splicing method for fault photovoltaic module positioning, and the method comprises the steps: extracting a feature point of a photovoltaic slate point edge in an image through employing a SuperPoint deep learning model, and precisely positioning a significant position which can be used for matching in the image; constructing a key point association graph through a SuperGlue algorithm, calculating a feature similarity matrix in combination with an attention mechanism, and establishing a corresponding relation of feature points among different images; solving an optimal feature matching pair by applying a Sinkhorn algorithm to obtain an optimal feature point corresponding relation between adjacent images, and providing accurate matching data for geometric transformation; according to the method, a SuperPoint self-supervision framework is utilized, through a VGG encoder and a double-branch decoder, the photovoltaic panel regular texture region feature point extraction capability is improved, uniform distribution is realized through non-maximum suppression, and the problems of sparse feature points and non-uniform distribution in a traditional algorithm are solved. And a SuperGlue graph neural network is introduced, and a similarity matrix with a direction weight is constructed in combination with a self-crossover attention mechanism, so that the accuracy of the edge feature points of the photovoltaic panel is improved.

The invention provides an outdoor power transmission channel hidden danger identification method and system based on semantic segmentation, and the method comprises the steps: marking a small target region in long-tail distribution from an outdoor power transmission channel image, carrying out the geometric transformation after cutting out the small target region, and carrying out the cross-image migration to generate a migration image; performing illumination correction on the migrated image to generate an enhanced image; constructing a semantic segmentation model for hidden danger recognition, and inputting the enhanced image into the semantic segmentation model to extract each semantic region and corresponding mask in the enhanced image; screening out abnormal masks from the masks and deleting the abnormal masks, and then removing existing cavity areas to obtain residual masks; and clustering and merging the remaining masks to generate a minimum enclosing rectangle with a rotation angle as a final detection frame. Based on the method, the invention also provides an outdoor power transmission channel hidden danger identification system based on semantic segmentation. According to the method, the outdoor power transmission channel hidden danger identification precision and efficiency are remarkably improved, and safe and stable operation of a power transmission line is ensured.

The invention belongs to the technical field of intelligent automobiles, and particularly relates to a driver driving state image data generation method based on a conditional diffusion model. Comprising the following steps: step 1, acquiring original driver driving state data and constructing camera parameters; 2, potential representation extraction and geometric control vector construction of an original image; step 3, potential representation generation of the target image; step 4, image reconstruction and post-processing; according to the method, an image diffusion generation network is taken as a core, a geometric transformation relation between an original camera and a target camera is combined, and generalization generation from a single image to a multi-view image is completed through a potential spatial modeling and condition guidance mechanism; the objective of the invention is to synthesize lifelike images under configuration of other visual angles and camera parameters by using a small amount of original images.

The invention belongs to the technical field of target detection, and particularly relates to a fast target detection algorithm based on multi-modal fusion, which comprises the steps of data acquisition and calibration, synchronous acquisition of images and point clouds and completion of internal and external parameter calibration; performing data enhancement operation such as zooming on the image, performing enhancement such as Pill-processing on the point cloud, and sharing geometric transformation to keep bimodal alignment; obtaining a two-dimensional bounding box and a confidence coefficient by using a YOLOv5 network; obtaining a three-dimensional bounding box by using a PointPillars network, and outputting a projection frame and a confidence coefficient; projecting the three-dimensional bounding box to a pixel plane according to the calibration matrix to form a bounding box; when the intersection-to-union ratio is greater than 0.5, generating a matching pair; each matching pair is coded into a 14-dimensional vector; vectors are input into a multi-layer perceptron, single-mode false detection is effectively inhibited through geometric and semantic consistency screening, 14-dimensional input features learn a nonlinear relation of a multi-mode bounding box through a three-hidden-layer MLP, and fusion precision is remarkably improved.

The invention discloses a digital pathological section paired image block data set construction method and system, and belongs to the technical field of medical image processing. The method comprises the following steps: carrying out local feature matching on HE and IHC images, and respectively solving a homographic transformation model and an affine transformation model based on a robust estimation algorithm; constructing an objective function preferential geometric model; mapping the HE image block to an original resolution coordinate system of the IHC image, determining a local rotation angle according to a local Jacobian, and executing one-time geometric transformation in a target domain; calculating similarity indexes based on tissue masks to perform quality control screening; hierarchically outputting index files by taking slices or patients as granularity; a verification example mask is obtained through an H-channel enhancement and segmentation process, an organization mask is generated based on a hybrid backspacing mechanism, and an example level label is exported. According to the method, the problems of difficulty in accurate registration and uncontrollable quality of the cross-dyeing image are solved, and a high-quality data basis is provided for subsequent model training.

The invention relates to the technical field of flexible circuit board detection, and provides an intelligent detection method and device for a flexible circuit board, and electronic equipment. The method comprises the following steps: S1, controlling an image acquisition device to obtain at least one target image of a to-be-detected flexible circuit board; s2, the control processing unit identifies a plurality of preset optical positioning reference points from the target image, and obtains actual coordinates of each optical positioning reference point in the current target image; s3, the control processing unit compares the actual coordinates of the optical positioning reference points with the theoretical coordinates of the optical positioning reference points in the standard template image to generate a deformation mapping relation of the to-be-detected flexible circuit board; s4, the control processing unit performs nonlinear geometric transformation on the target image or the to-be-detected region of interest in the target image according to the deformation mapping relation to obtain a corrected image which is spatially aligned with the standard template image; and S5, controlling the processing unit to execute a defect detection algorithm on the corrected image so as to identify defects on the flexible circuit board.

The invention provides an OCTA image retinal vessel segmentation method based on a full-resolution network, and the method comprises the steps: carrying out the preprocessing of an input OCTA image through a data preprocessing module, wherein the preprocessing comprises the optical transformation and geometric transformation, and obtaining the preprocessed data; the feature coding module uses five feature extraction units which are arranged in sequence to extract image features of five levels and outputs the image features to the cascade feature enhancement module; a cascade feature enhancement module CFEM combines low-level detail information and high-level semantic information through cross-layer guidance for feature fusion, and then outputs a segmentation result; calculating loss and optimizing the OCTA image retinal vessel segmentation model based on the full-resolution network to obtain an optimized segmentation model; using the optimized segmentation model to obtain a retinal blood vessel image; according to the method, high-precision and high-efficiency segmentation of the retinal blood vessel can be realized, the network complexity is relatively low, and the computing resource consumption is relatively low.

A system and method for implementing experiential manifold cognition that extends persistent cognitive machines beyond discrete thought caching to continuous geometric representation of experience. The system maintains an experiential manifold comprising a differentiable manifold with Riemannian metric tensor encoding semantic relationships, compression pressure field governing memory consolidation, and potential field encoding goals and attention. Input data is projected onto the manifold through adaptive geometric diffusion preserving semantic structure. The system executes geometric transformations including metric evolution, geodesic computation, and curvature estimation. During non-interactive periods, autonomous evolution occurs through trajectory recombination and selective pruning. A user interface enables visualization and direct manipulation of manifold geometry, translating navigation into geodesic traversal and edits into metric modifications. The system maintains persistence across sessions and enables controlled federation between multiple manifolds through consent-bounded synchronization. Applications include persistent narrative worlds, collaborative cognitive spaces, and experiential intelligence systems that learn through geometric evolution.

The invention relates to the technical field of target detection, in particular to a blueberry fruit focusing detection method and system based on a lightweight YOLO model, and the method comprises the steps: obtaining blueberry images to form a multi-dimensional data set, and carrying out the marking of the obtained multi-dimensional data set; performing data enhancement based on the acquired multi-dimensional data set, including performing geometric transformation on the acquired original data set, and performing optical simulation and occlusion simulation by adding Gaussian noise and gradient operators; and constructing a detection model by taking the preprocessed multi-dimensional data set as input, introducing a dynamic attention mechanism on the basis of the detection model, carrying out model training and optimization on the basis of the constructed detection model, and inputting a test set for detection by utilizing an optimal weight obtained by training to generate a final detection result. According to the method, dual optimization of semantic understanding and accurate positioning is realized through feature fusion, and the detection precision of overlapped fruits and small targets is effectively improved.

The embodiment of the specification provides a copper foil defect detection system, comprising: a data acquisition module configured to acquire a copper foil surface image through a multi-spectral line-scan digital camera and output original image data; the preprocessing module is configured to perform dynamic denoising and partition contrast enhancement on the original image data to generate a preprocessed image; the frequency domain processing module is configured to perform Fourier transform on the preprocessed image to obtain a frequency domain feature map, separate a low-frequency contour and a high-frequency texture component through a band-pass filter bank, and generate a composite feature map through inverse transformation after frequency domain mask enhancement; the data enhancement module is configured to implement geometric transformation enhancement and frequency domain noise injection on the composite feature map and output an enhanced sample set; and the multi-scale detection network is configured to extract hierarchical features from the enhanced sample set, fuse the hierarchical features through a gating mechanism and output defect positioning and classification results. According to the scheme, the frequency domain image processing technology is combined with the CNN, and the detection effect of the weak contrast defect is improved.

The invention discloses a multi-layer semantic perception, distillation and semi-supervised cooperative training target detection method, and the method specifically comprises the following steps: constructing a data set: extracting a first part of images from image data, marking the first part of images to construct a supervised target detection data set, and taking the remaining images as an unmarked image data set, the data volume of the unlabeled image data set is greater than that of the supervised target detection data set; teacher model optimization: performing supervision training on the teacher model on the supervised target detection data set; constructing a teacher model, and executing self-distillation training on the teacher model; pseudo labels are generated for the unlabeled images through a teacher model, and adaptive screening is carried out based on confidence distribution; mapping a pseudo label to a strong enhanced sample through enhanced geometric transformation, carrying out semi-supervised training by using the strong enhanced sample and the pseudo label, and introducing a feature layer distillation constraint at the same time; optimizing a student model; and outputting the target detection model obtained through training.