430 results about "Three dimensional surface" patented technology

The invention relates to the technical field of PCB detection, and discloses a PCB defect detection method and system, and the method comprises the steps: collecting multispectral imaging data through an image collection module, and generating an original image data set; the defect analysis server receives the synchronous imaging data to construct a three-dimensional surface topology matrix; in combination with the original image data set and the real-time imaging data, performing multi-scale decomposition on the three-dimensional surface topological matrix, extracting texture features, positioning a defect region, outputting defect type space distribution features through a layered recognition model, and updating the original image data set; and dynamically calibrating the detection parameters according to the feature categories. The system comprises an image acquisition module group, a data transmission module, a three-dimensional modeling module, a defect identification module and a parameter calibration module. According to the scheme, the accuracy, comprehensiveness and efficiency of defect detection are improved, and the detection requirements of modern PCB production are met.

The invention discloses an AI spraying system control method capable of achieving autonomous 3D scanning, and the method comprises the steps: obtaining three-dimensional surface data of a to-be-coated workpiece through high-precision 3D scanning equipment, constructing a surface model of the workpiece according to the data, and extracting surface feature parameters. Coating parameters set by a user are received, the coating parameters comprise the viscosity, the chemical property and the film thickness requirement of paint, the surface characteristic parameters and the coating parameters are input into the spraying optimization model, and an initial spraying track and technological parameters are generated through calculation in combination with historical data. Environment parameters are collected and associated with the initial process parameters, and the adjusted spraying track and the adjusted process parameters are generated through dynamic environment compensation. And through a robot control system, coating operation is executed according to the adjusted parameters, the spraying effect is optimized in real time, and it is ensured that the surface of the workpiece is evenly coated. Efficient automation of the spraying process is achieved, the coating quality and consistency are remarkably improved, and the automatic spraying device is particularly suitable for coating of workpieces with complex curved surfaces or changeable shapes.

The invention provides a burn area measurement method and system, electronic equipment and a storage medium, and relates to the field of area measurement. The method comprises the following steps: photographing a target part to obtain a multi-modal burn image set, and carrying out image registration; and constructing a global three-dimensional surface model based on the registered image set. Carrying out burn area identification segmentation on the multi-mode burn image, mapping a segmentation result to the global three-dimensional surface model, and obtaining a preliminary burn area three-dimensional surface model; and performing multi-scale feature extraction and fusion on the registered image set, and inputting a depth estimation model to obtain burn depth estimation data. And performing deep layering on the preliminary burn area three-dimensional surface model based on the burn depth estimation data to obtain burn area three-dimensional images of different depth levels. And calculating the actual burn area of each depth level based on the burn area three-dimensional images of different depth levels and the camera focal length information. Through the method, the accuracy of burn area measurement is improved.

The invention provides a three-dimensional micro-nano spherical multi-parameter synchronous mapping method based on a longitude and latitude type scanning track, and belongs to the technical field of micro-nano device and cell surface physical property characterization methods. In combination with an eccentric distance real-time compensation control algorithm, latitude scanning path parameters are established, online compensation of eccentric displacement is realized in the process of rotating along the latitude, and a normal bending deformation signal of a probe micro-cantilever is used as feedback through an upper computer, so that displacement control of a scanning positioning platform is realized. And thus, synchronous scanning imaging of the surface morphology and the nano-mechanical property of the sample is realized. A probe integrated with a protuberant tip structure is used as a micro-force sensor, a novel controllable vector angle single-point servo detection method is combined, a normal bending signal of the probe is used as feedback, and synchronous measurement of the three-dimensional surface topography and the three-dimensional surface mechanical property of a micro-nano sphere sample is realized along a longitude and latitude scanning path.

The invention relates to the technical field of photoetching machine calibration, and discloses a photoetching machine calibration method, device and equipment based on multi-view vision, and the method comprises the following steps: carrying out imaging and phase sensitive detection analysis on a mask plane and a wafer plane through a four-path optical beam splitting system to obtain four groups of mask-wafer initial alignment position information; performing scanning white light interference edge enhancement processing to obtain three-dimensional surface contour data; performing wavelet transform processing on the three-dimensional surface contour data, and performing dynamic registration on the four-path optical beam splitting system to obtain multi-view visual feature registration data; a comprehensive error model including mechanical errors, optical errors and environmental errors is established, real-time correction and error compensation are carried out on a six-degree-of-freedom motion platform of the photoetching machine, a multi-view vision calibration result of the photoetching machine is obtained, the influence of environmental vibration and thermal drift on calibration precision is effectively eliminated, and the calibration accuracy of the photoetching machine is improved. The optical system is ensured to be always in the optimal imaging state, and the calibration precision is improved.

The invention discloses an automatic planning method and system for a curved surface path of a grinding robot, and the method comprises the following steps: S1, constructing a to-be-machined three-dimensional curved surface model and a triangular mesh model, and relates to the technical field of automatic control of robots for surface machining in the manufacturing industry. According to the method, a complex three-dimensional surface coverage path planning problem of a workpiece is converted into two low-complexity sub-problems of key point sampling of a two-dimensional path planning domain and three-dimensional shortest path searching on a triangular mesh model, a small number of sampling points are sampled only on the two-dimensional path planning domain, and the sampling process depends on a spatial index framework; according to the method, global traversal of disordered point clouds is avoided, three-dimensional coordinates are positioned through nearest neighbor search of a KD tree algorithm, adjacent three-dimensional ordered key points are sequentially connected through a graph search algorithm, and a continuous machining path is formed, so that the operand is greatly reduced, real-time response of path planning is facilitated, and the problems that traditional online planning is long in time consumption and difficult in real-time response are solved.

The invention relates to a method and system for measuring the three-dimensional morphology of a metal hydrophobic surface based on white light microscopic interference, and belongs to the field of precision measurement, and the method comprises the steps: firstly, measuring a sample, scanning the surface of the sample, and synchronously collecting a white light interference image sequence containing surface contour information; adopting a bicubic interpolation algorithm to carry out interpolation processing on the acquired white light interferogram group, and constructing a high-resolution interference image sequence; analyzing the time-frequency domain interference signal through fast Fourier transform, effectively inhibiting noise interference in combination with a filtering technology, and accurately extracting an envelope curve of the interference signal; and carrying out Gaussian fitting on the envelope peak by adopting a least square method and positioning a peak value coordinate, and finally reconstructing a high-precision three-dimensional surface topography. According to the method, through advantage complementation of the bicubic interpolation algorithm and the fast Fourier transform method, the transverse resolution is remarkably improved while the longitudinal resolution is kept, and an effective technical means is provided for accurate characterization of the three-dimensional morphology of the metal hydrophobic surface structure.

The invention relates to the technical field of lens correction, and discloses an intelligent correction method and system for the optical performance of a functional lens, and the method comprises the steps: obtaining the original data of the lens, and carrying out the preprocessing of the original data of the lens, and obtaining the surface data of the lens; performing surface parameter calculation according to the lens surface data to obtain three-dimensional surface parameters; performing weighted residual calculation according to the three-dimensional surface parameters to obtain a residual matrix; performing spatial correlation analysis according to the three-dimensional surface parameters and the residual matrix to obtain an error region and a normal region; performing compensation analysis according to the residual matrix, the error region and the normal region to obtain a parameter compensation amount; and performing multi-objective optimization according to the parameter compensation amount to obtain an optimal compensation parameter, and correcting the lens according to the optimal compensation parameter. The method has the following effect that the correction accuracy of the optical performance of the functional lens can be improved.

The present invention relates to the field of mine surveying and mapping technology, and in particular to a mine surveying and mapping method and system based on unmanned aerial vehicle remote sensing technology, the method comprising: obtaining terrain data of a mine; generating a flight trajectory of an unmanned aerial vehicle; using an unmanned aerial vehicle to collect image data and hyperspectral data of the mine; constructing a three-dimensional surface model of the mine; obtaining geological structure data and mineral distribution data of the mine; and generating a three-dimensional structural model of the mine based on the geological structure data, mineral distribution data and the three-dimensional surface model. The present invention generates a flight trajectory of an unmanned aerial vehicle to ensure that the unmanned aerial vehicle takes photos at a reasonable flight altitude and speed in different areas, thereby reducing repeated coverage and maximizing regional coverage. By generating a three-dimensional structural model, detailed geological structure and mineral distribution data can be obtained without actually entering the mining area, greatly reducing the safety risks and economic costs that may be caused by on-site surveys.

The invention relates to the technical field of intelligent mechanical arm control, in particular to a mechanical arm grabbing method and system based on image processing, and the method comprises the steps: collecting an image of a target object needing to be grabbed by a mechanical arm through a visual sensor, carrying out the preprocessing and multi-scale decomposition of the collected image of the target object, screening a region of a potential grabbing point, and carrying out the recognition of the target object; and generating a capture point candidate area map. According to the invention, through processing and multi-scale decomposition of the target object image, noise is eliminated, an illumination area is balanced, a clearer target feature image is generated, and environmental adaptability is enhanced. And in combination with texture change detection and gradient analysis, the feature intensity and boundary characteristics of the object are accurately extracted, and more accurate grabbing point screening is realized. Furthermore, the three-dimensional surface form projection is in a two-dimensional form, and the stability and operability of the grabbing area are improved by optimizing the boundary symmetry and curvature fluctuation.

The invention discloses a handheld three-dimensional scanning system based on visible light, which belongs to the technical field of electronic scanning and comprises a calibration module, a self-adaptive adjustment module, a three-dimensional scanning module and a storage module. The calibration module is used for calibrating the three-dimensional scanner by using a known reference object; the self-adaptive adjustment module is used for automatically optimizing scanning parameters according to the material and surface characteristics of a target object so as to adapt to the surface characteristics of different materials; the three-dimensional scanning module obtains a multi-angle reflection image of the surface of a target object in real time, recognizes geometric information of the surface of the target object in the image, constructs a three-dimensional surface model of the target object, reduces an error between the model and the target object through fusion of the multi-angle image, improves the precision of the three-dimensional model, and improves the accuracy of the three-dimensional surface model. The texture information of the target object is captured, and the vivid effect of the model is enhanced; the storage module is used for storing data generated in the scanning process and establishing an efficient data index according to the stored data.

The invention relates to a rubber part finished product forming degree detection method, system and equipment and a medium, and belongs to the technical field of rubber part forming degree detection. The molding degree detection method comprises the following steps: receiving surface point cloud data and surface geometric position information of a to-be-detected rubber part; constructing a three-dimensional surface contour model of the rubber part to be detected; mapping the surface pressure distribution data to the three-dimensional surface contour model, and detecting a pressure abnormal distribution area; extracting light scattering characteristic data according to the optical image data; comparing the light scattering characteristic data with the pressure abnormal distribution area to obtain a defect area analysis result; dynamic deformation characteristics of the rubber part to be tested under the dynamic environment simulation test are obtained; marking the three-dimensional surface contour model, and constructing a comprehensive feature model of the rubber part to be detected; and carrying out difference analysis on the comprehensive feature model to obtain a feature difference result of the to-be-detected rubber part, and generating a forming degree scoring report of the to-be-detected rubber part. According to the invention, the molding degree of the rubber part can be accurately detected.

The invention discloses a highlight reflection scene object three-dimensional surface reconstruction method based on an adaptive plane and jump sampling, which distinguishes reflection interference from a target object through a pixel-by-pixel auxiliary plane, and introduces a Laplace gradient to reduce the influence of external reflection interference. In addition, the influence of reflection interference is reduced by adopting a jump sampling strategy, and the multi-view consistency and the surface fidelity are further improved. Finally, rejection loss is adopted, gradient differences of a target object and a reflection part are compared, the model is helped to more accurately distinguish the target object and the reflection part during initialization, and a more accurate model surface is reconstructed under highlight reflection interference.

The invention discloses a stele stone carving digital rubbing method based on a three-dimensional laser scanning model, and relates to the technical field of cultural relic protection and digitization, and the method comprises the steps: scanning a stele by using a three-dimensional data scanner, obtaining the three-dimensional data of the stele, and carrying out the data splicing to form a three-dimensional data set; preprocessing the three-dimensional data set and constructing a three-dimensional surface model of the stele; extracting a grey-scale map of the surface of the stele from the three-dimensional surface model; and gray information of the concave seam area on the surface of the stele is extracted from the gray-scale map, and a digital rubbing of the stele is generated according to the gray information. Complete three-dimensional data of the stele is obtained by using a three-dimensional laser scanner, the integrity of the data is ensured, a three-dimensional surface model is generated by using an advanced algorithm, the geometric accuracy of the model is ensured, an accurate basis is provided for the extraction of a grey-scale map, the extracted concave seam area information is more accurate, the definition and detail expressive force of the image are enhanced, and the image quality is improved. And the finally generated digital rubbing has high resolution and high contrast.

The embodiment of the invention discloses a high-precision three-dimensional surface reconstruction method and device based on polarization information enhancement. A specific embodiment of the method comprises the following steps: acquiring multi-view polarization images by using a polarization camera to obtain a polarization image set; performing structured light initialization processing on the target image to generate initialized target image information; inversely projecting the accurate depth information set into a three-dimensional space to obtain enhanced Gaussian point cloud data; gaussian point cloud de-reflection processing is carried out on the initialized target image information to generate de-reflection Gaussian point cloud data; generating Gaussian point cloud data based on the enhanced Gaussian point cloud data and the de-reflected Gaussian point cloud data; and performing data optimization and reconstruction processing on the Gaussian point cloud data to generate a three-dimensional triangular patch. According to the embodiment, the precision and efficiency of three-dimensional surface reconstruction can be improved.

The invention discloses a target object pose estimation method, system and device and a medium, and relates to the technical field of pose estimation, and the method comprises the steps of offline data set establishment and online real-time multi-target pose estimation. In the off-line process, continuous RGBD streams surrounding a single object are collected, the relative poses of front and back frames are calculated frame by frame, and meanwhile, the three-dimensional surface of a target object is reconstructed to obtain a key frame off-line data set; in the online process, a visible target set in a data stream is firstly detected, VOS segmentation is performed on each target to obtain real-time effective texture and point cloud, offline key frame images of to-be-detected targets are arranged to realize single matching of all feature points and find out a closest reference frame at the same time, and RANSAC is adopted to calculate relative poses of a current frame target and the reference frame; according to the method, an expensive data labeling process is avoided, rapid integrated deployment of detection of an object without a new target is realized, the matching speed is improved, and meanwhile, online real-time multi-target pose estimation can be realized.

The invention provides an injection molding part surface flaw detection method and device based on double-model collaboration, and the method comprises the steps: carrying out the local abnormal reflection feature analysis of a multi-view optical image collected on the surface of an injection molding part based on a first detection model, and obtaining a candidate flaw region; performing spatial positioning in the multi-view optical image based on the candidate flaw area to obtain a multi-view candidate image segment; performing surface geometric continuity analysis on the multi-view candidate image segments based on a second detection model to obtain a three-dimensional surface geometric consistency result; carrying out authenticity discrimination on the candidate flaw area based on a three-dimensional surface geometric consistency result to obtain a real flaw area; wherein the first detection model is used for capturing an optical response model of local abnormal reflection characteristics; the second detection model is a stereoscopic vision model for analyzing the geometric continuity of the multi-view lower surface. According to the invention, the false alarm rate of surface defect detection of the injection molded part under a complex surface condition is reduced.

The invention discloses a modeling optimization processing method and system based on a digital image, and relates to the technical field of three-dimensional modeling, and the method comprises the steps: using a lightweight Vision Transform to extract global features, using a Harris corner point to detect and recognize feature points, generating a multi-scale feature map through Laplace decomposition, generating a semantic segmentation map in combination with DeepLabv3 +, constructing a hybrid Poisson equation, and carrying out solution optimization. A smooth three-dimensional surface model is generated through a Marching-Cubes algorithm, an initial contour is adjusted, optimization is carried out in combination with a dynamic change vector, and a final optimization model is generated through completion, denoising, simplification, quantification and compression processing. According to the method, the feature extraction quality is enhanced through standardization processing and Vision Transform, the initial model precision is improved through multi-scale fusion and semantic segmentation, efficient and refined modeling is realized through mixing Poisson optimization and multi-scale adjustment, and a high-quality and low-complexity three-dimensional model is generated.

The present application relates to a geographic mosaicking method and apparatus for video images, and a computer device and a storage medium. The method comprises: acquiring real-time live streaming images of a real-world scene that are collected by a plurality of unmanned aerial vehicles, and GNSS information of the plurality of unmanned aerial vehicles; using a visual SLAM algorithm to perform input frame tracking on the real-time live streaming images, and performing pose estimation on successfully tracked input frames by combining visual trajectories and the GNSS information, so as to generate georeferenced camera poses; using the camera poses and a surface reconstruction algorithm to perform densification processing on the input frames, so as to generate a depth map of the real-world scene, and mapping the depth map into dense three-dimensional point clouds, so as to construct a three-dimensional surface model of the real-world scene; and using the camera pose and the three-dimensional surface model to perform orthorectification on the input frames, and integrating the orthorectified input frames into a global mosaicked map, so as to generate a global image of the real-world scene. By means of the embodiments of the present application, a real-time picture of a real-world scene can be quickly acquired, thereby providing robust support for a rapid emergency response in a real-world scene.

The invention discloses an intelligent laser-dry ice composite cleaning method and device, belongs to the field of stain cleaning, and aims to solve the problem that the cleaning effect on large-thickness stains is not ideal in the prior art. The method comprises the following steps: performing visual scanning on a to-be-cleaned workpiece, obtaining surface stain thickness distribution of the to-be-cleaned workpiece through three-dimensional surface topography reconstruction, and dividing the surface of the to-be-cleaned workpiece into a plurality of sub-regions according to the stain thickness; the sub-regions comprise three types: an ultra-thin sub-region, a medium-thickness sub-region and a large-thickness sub-region; performing cleaning path planning to form three cleaning paths; the cleaning process of the ultra-thin sub-region cleaning path comprises the following steps: laser cleaning; the cleaning process of the medium-thickness sub-region cleaning path is laser-dry ice composite cleaning; the cleaning process of the large-thickness subarea cleaning path is laser-dry ice composite cleaning. The cleaning device comprises a ground rail gantry movement system, a first mechanical arm, a second mechanical arm, a dry ice cleaning system, a visual recognition system and a pulse laser cleaning system.

The invention belongs to the technical field of agricultural product automatic detection, and discloses an agricultural product quality identification method and system, and the method comprises the steps: synchronously obtaining a visible light image, a structured light reflection image, a hyperspectral image and other multi-mode images; constructing a three-dimensional surface model by using the structured light phase information, and mapping the multi-source data onto the three-dimensional surface model to form a unified data volume; self-adaptive fusion is carried out through a cross-modal attention network, and depth features are generated; secondly, augmenting features of rare defects in a feature space by adopting a conditional generative adversarial network to solve the problem of data scarcity; finally, a classifier is trained based on the augmented data set, high-precision recognition of the quality of the agricultural products is achieved, and the robustness and the recognition breadth of complex scenes are remarkably improved. According to the method, multi-source information is deeply integrated, the problem of data scarcity is innovatively solved, and the breadth and precision of agricultural product quality identification are remarkably improved.

The invention relates to the technical field of image data processing, in particular to a workshop filter defect detection method and system based on image processing, and the method comprises the steps: calculating the real-time rotating speed of a fan based on image analysis; dynamically adjusting the frame rate of the camera according to a rotating speed comparison result; reconstructing a three-dimensional surface model of the equipment; performing camera vibration compensation through the coding mark points; fusing the visible light image, the thermal imaging and the three-dimensional model data to generate a comprehensive diagnosis result; compared with the prior art that blade images are blurred and illumination and shadow interference cannot be accurately distinguished when a fixed frame rate camera fluctuates in rotating speed, the method has the advantages that the frame rate is dynamically adjusted by tracking the fan rotating speed deviation in real time, and multiple groups of images of each blade at the same rotating angle position are captured; the blade tip gradient vector is combined to analyze illumination shadow distribution characteristics, real defects and shadow artifacts are effectively separated, the microcrack recognition accuracy is remarkably improved, and meanwhile misjudgment caused by sudden change of environment illumination is avoided.

The application relates to the technical field of image processing, in particular to a camera-based progressive feature distribution sampling 6D pose estimation method and system, which comprises the following steps: obtaining dense pixel points by using a multilayer progressive feature sampling network of Unet++; analyzing three-dimensional model surface point features by using graph convolution to form a corresponding distribution relationship between image pixel points and surface points; calculating the probability distribution of the corresponding key of the image pixel coordinate point set features by querying the model to obtain the mask probability of the corresponding three-dimensional surface of the image pixel coordinates; calculating the best probability distribution value; scoring and voting each generated hypothesis; refining the obtained best voting score attitude hypothesis according to the surface distribution to obtain the target final attitude estimation pose. The application solves the problems that the stability and accuracy of pose estimation of a target object are not high when only RGB data is used, and the accuracy is not high when irregular features exist in three-dimensional model surface feature sampling.

The invention belongs to the technical field of bulk material scanning, and relates to a bulk material scanning method based on a laser radar sensor, which abandons a low-efficiency mode that all laser radar sensors continuously work at full power in the prior art, carries out global monitoring by setting a master scanning node, and sets a slave scanning node in a low-power-consumption mode. The method comprises the following steps: registering point cloud data collected by a main scanning node in adjacent scanning periods, generating a change heat map through voxelization differential calculation and spatial clustering, retrieving a high heat change region in the map, and aiming at the high heat change region, considering a geometrical relationship between a sensor and the region and multi-sensor visual angle complementarity. A sensor subset is selected from all slave scanning nodes to execute cooperative enhanced scanning, high-density local point clouds obtained through enhanced scanning and global point clouds of the master scanning nodes are fused, a bulk material pile three-dimensional surface model is generated, and dynamic on-demand scheduling and energy efficiency optimization of scanning resources are achieved.

The invention provides a space intelligent non-planar reflection generation method based on surface normal guidance. Extracting multi-layer visual representation from an original RGB image containing a non-planar reflection medium scene, and identifying a non-planar reflection area by combining brightness distribution of the original RGB image and a spatial symmetry relation; analyzing non-linear deformation features of textures in the non-planar reflection area, and predicting pixel-level three-dimensional surface normal vectors and confidence distribution in combination with multi-layer visual representation to construct a surface normal graph; determining a line-of-sight vector based on a camera imaging model, and executing geometrical optical mapping operation on the line-of-sight vector, the surface normal graph and the confidence distribution to construct a reflection offset field; extracting an intermediate semantic feature representation corresponding to the non-planar reflection region, and generating a pre-distortion reflection feature representation according to the reflection offset field and the intermediate semantic feature representation; and injecting the pre-distortion reflection feature representation, the multi-layer visual representation and the initial reflection area mask into a generative model for image reconstruction, and generating a non-planar reflection image.

The invention discloses a point cloud scanning processing method, and relates to the technical field of computer graphics and three-dimensional reconstruction. According to the method, accurate registration and seamless fusion of multi-layer point cloud data are achieved, noise is effectively removed through a series of preprocessing steps including statistical filtering, Gaussian filtering, average filtering and a random sampling or voxel grid filtering technology, the point cloud data are smoothed, and the data size is remarkably reduced. In the aspect of sub-cluster processing, principal component analysis is used for deep data dimension reduction and feature extraction, sub-clusters are segmented in a recursive mode, and insertion sub-clusters are set and adjusted, so that accurate segmentation and fine adjustment of the sub-clusters are ensured. And finally, based on an advanced Alpha shape algorithm, a three-dimensional surface model which is faithful to a point cloud main body structure and finely presents local details is constructed, so that a high-quality three-dimensional model is accurately extracted, and a solid foundation is laid for subsequent graphic processing and analysis.

The invention belongs to the technical field of forge piece warpage measurement, and particularly relates to a large long beam type die forging warpage measurement method and system based on three-dimensional contour scanning, and the method comprises the steps: S100, scanning a forge piece, and obtaining original point cloud data; step S200, processing the obtained original point cloud data so as to realize de-noising and main body target segmentation; step S300, accurately matching all the point cloud data on the scanning path, and splicing the point cloud data in sequence to obtain final point cloud data of the forge piece; step S400, performing surface reconstruction on the final point cloud data to generate a continuous and complete three-dimensional surface model; step S500, measuring and calculating the warping degree of the forge piece by using the three-dimensional surface model; the warping detection process can be improved from original sampling inspection to sample inspection, so that the defective rate is reduced; the forge piece scrapping risk caused by blind correction is avoided, and equipment and electric furnace energy consumption and labor cost caused by repeated correction are reduced; and the operation is automatic and convenient.

The invention belongs to the technical field of three-dimensional surface reconstruction, and particularly relates to a robust three-dimensional surface reconstruction method based on multi-scale fusion. Acquiring three-dimensional point cloud data, fitting local implicit curved surfaces of neighborhood points by adopting a weighted least square method, and generating a geometric reconstruction grid model; extracting geometric features in different receptive fields through multi-scale convolution, introducing channel attention, space attention and a gating mechanism to perform weighted fusion on the geometric features in different receptive fields, enhancing a global dependency relationship by adopting a feature mixing operation, and generating a deep learning grid model; performing double-branch fusion on the geometric reconstruction grid model and the deep learning grid model, highlighting a geometric sensitive area, and obtaining a fused enhanced feature; inputting the fused enhanced features into an implicit function decoder, and predicting a signed distance function value of each vertex; and based on the predicted signed distance function field, a contour surface extraction algorithm is adopted to generate a triangular mesh model, so that three-dimensional surface reconstruction is completed.

The invention relates to the technical field of image processing, in particular to a chip surface defect detection method based on multi-modal image fusion. The method comprises the following steps: firstly, acquiring three-dimensional surface geometric data and dynamic thermal response data of a chip; performing physical model correction on the dynamic thermal response data based on the three-dimensional surface geometric data to generate a thermal conductivity surface graph; then, guiding a scanning acoustic microscope to carry out sub-surface detection on the chip based on an abnormal region identified by the thermal conductivity surface graph so as to obtain acoustic image data, fusing the three-dimensional surface geometric data, the thermal conductivity surface graph and the acoustic image data in a three-dimensional coordinate system, and constructing a surface-sub-surface integrated three-dimensional model of the chip; then, analyzing the integrated three-dimensional model by adopting a physical information neural network embedded with physical constraints, and outputting a defect type identification result; the physical accuracy and the detection rate of chip surface defect detection can be improved.

The present invention relates to the field of PCB circuit board detection technology, and discloses a PCB circuit board defect detection method and system. The method comprises the following steps: an image acquisition module acquires multispectral imaging data to generate an original image dataset; a defect analysis server receives synchronized imaging data to construct a three-dimensional surface topology matrix; the original image dataset and real-time imaging data are combined to perform multi-scale decomposition of the three-dimensional surface topology matrix, extract texture features, locate defect areas, output spatial distribution features of defect types through a hierarchical recognition model, and update the original image dataset; and detection parameters are dynamically calibrated according to feature categories. The system includes an image acquisition module group, a data transmission module, a three-dimensional modeling module, a defect recognition module, and a parameter calibration module. This solution improves the accuracy, comprehensiveness, and efficiency of defect detection and is suitable for the detection needs of modern PCB production.