1071 results about "Projection image" patented technology

The invention relates to a three-dimensional point cloud registration method and system based on a two-dimensional visual large model. A source point cloud and a target point cloud are decoupled into two-dimensional projection representation; constructing a structure consistency restoration domain, restoring geometric structure fracture and semantic deficiency in the projection image, and generating a dense and continuous restoration image; inputting the repaired image into a pre-trained visual geometry large model, carrying out cross-modal feature matching and pose resolving, and synchronously outputting a pixel-level matching confidence map; and inversely mapping a two-dimensional pose calculation result to a three-dimensional space to complete coarse registration, constructing a high-credibility anchor point set in the three-dimensional space by using a confidence map, and guiding a point cloud fine registration algorithm to converge to global optimum. According to the method, the registration large model for natural image training can be seamlessly migrated to the point cloud data, and retraining is not needed; in a weak texture and partially overlapped point cloud scene, a robust rough registration initial value can still be provided; the convergence speed and the anti-noise performance of fine registration are remarkably improved, and adaptive improvement from coarse to fine is achieved.

In various examples, self-supervised learning may be used to pre-train an encoder network of a masked prediction model to reconstruct masked regions of an input representation of 3D detections such as LiDAR point cloud(s). Spatial and / or temporal masking may be applied to a projected representation of 3D detections (e.g., a two-dimensional (2D) projection image), and the masked prediction model (e.g., a masked auto-encoder or joint-embedding predictive architecture) may be used to reconstruct a representation of the masked regions (e.g., reflection characteristic(s) stored in corresponding pixels or cells of the projected representation, a latent representation of the reflection characteristic(s)) during iterations of self-supervised learning. As such, the pre-trained encoder network of the masked prediction model may be used as a foundation model and fine-tuned with a task-specific output head or its pre-trained weights may be used to initialize a task-specific model.

The invention discloses a building outer wall falling risk detection method and system based on deep learning, and the method comprises the steps: obtaining an original surface image, extracting vertical face geometric features, constructing a homography matrix based on vanishing points, carrying out the geometric correction, and reconstructing an orthographic projection image; performing enhancement processing on the orthographic projection image, extracting an effective detection area and intercepting a detection image; inputting the detection image into a deep learning segmentation model into which edge weight constraint is introduced, and performing semantic segmentation, binaryzation and optimization to obtain a falling region; a mapping relation is established according to the actual physical size of the building outer wall, and pixel information of the falling area is converted into multi-dimensional parameters such as the actual physical area, the elevation, the width and the height; and combining the defect intensity input quantification risk assessment model to calculate a comprehensive risk index, determining a risk level and outputting a result. Automatic identification and quantitative evaluation can be realized, distortion and background interference are eliminated, accurate conversion from pixels to a multi-dimensional physical space is realized, and a scientific basis is provided for safety investigation.

The invention relates to an optical distortion test system and method. According to the method, a light source assembly is used for outputting projection light; the target assembly is fixed in front of the projection light output end of the light source assembly. The to-be-tested assembly is located on the side, away from the light source assembly, of the target assembly and connected with the adjusting assembly. The adjusting assembly comprises a pitching adjusting sub-assembly, a horizontal adjusting sub-assembly, a height adjusting sub-assembly and an inclined sub-assembly, the pitching adjusting sub-assembly is used for ensuring that the to-be-detected assembly is perpendicular to the projection optical axis, and the horizontal adjusting sub-assembly and the height adjusting sub-assembly are used for ensuring that the optical center of the to-be-detected assembly is coaxial with the projection optical axis and the lens center of the detection assembly; the inclined sub-assembly is used for adjusting a space angle between an optical central axis of the to-be-detected assembly and a projection optical axis of the light source assembly so as to detect edge distortion of the to-be-detected curved surface element; the detection assembly is located on the side, away from the to-be-detected assembly, of the adjusting assembly and used for shooting a projection image of the target assembly which generates distortion after being acted by the to-be-detected assembly.

The invention provides a component identification method and system based on a three-dimensional point cloud, and relates to the field of industrial detection. The method comprises the steps of determining a main geometric region of a geometric component based on first point cloud data; processing the first point cloud data of the main geometric region in a layered and partitioned manner to obtain second point cloud data corresponding to different structure hierarchies; projecting the second point cloud data to a two-dimensional plane, and determining a two-dimensional projection image; determining geometric feature information based on the two-dimensional projection image; and determining a three-dimensional geometric model based on the geometric feature information. The system comprises an area module, a layering and partitioning module, a projection module, a sensing network platform, a feature information module and a modeling module. According to the method, the three-dimensional geometric model of the component can be stably and efficiently determined, the recognition accuracy and efficiency are improved, and the actual requirements of industrial production detection are met.

The invention discloses a reflector structure for a projection imaging lens, which comprises a reflector body, the reflector body is placed in an optical projection imaging lens with more than two lenses, and the reflector body comprises a middle light-passing area, a middle light-passing area and a middle light-passing area, so that light of the optical projection imaging lens is emitted through the middle light-passing area to become direct emergent light; the upper end of the first reflecting region is small and the lower end is large; the upper end of the second reflecting region is large and the lower end is small; the optical projection imaging lens is vertically placed, the light source or the image source is arranged below the optical projection imaging lens, and the lens is arranged above the optical projection imaging lens; the reflector body is configured to be matched with a circular projection imaging lens in rotational symmetry or a projection imaging lens in a non-rotational symmetry structure. The surface brightness and uniformity of the light-emitting surface of the projection lens are improved.

The invention provides a building structure deformation measurement method, system and device, and relates to the technical field of building structure deformation measurement. The measuring method comprises the following steps: acquiring a laser point physical coordinate, an included angle between a laser emission direction and a world coordinate system and a building image; after the building image is preprocessed, laser point pixel coordinates are extracted; after the relative coordinates of the object surface laser points are calculated, solving a homography transformation matrix, carrying out perspective correction on the building image to obtain an orthographic projection image, and solving a conversion coefficient between pixels and an actual distance; inputting the preprocessed building image into a pre-trained semantic segmentation model to obtain a beam body segmentation mask and a beam body edge curve; and the actual span of the beam body is obtained through conversion of the conversion coefficient. According to the scheme, three-dimensional movement is achieved through the electric driving platform, the positioning position of the laser point can be adjusted according to the size and position of the building structure, buildings of different heights and angles are adapted, and the limitation that only building structures of specific types or sizes can be measured in the prior art is solved.

The invention relates to the technical field of AR, and discloses an AR war chess projection and identification method and system based on spatial positioning and region segmentation. The method comprises the following steps: acquiring a map local image, identifying angular point pixel coordinates and angular point world coordinates of a reference mark in the map local image, and calculating a working range polygon of a map surface currently covered by a projection device based on the angular point pixel coordinates and the angular point world coordinates; the coordinate information of the working range polygon is sent to a server, and the server screens local situation data located in the working range polygon from the global situation data; and receiving the local situation data pushed by the server, rendering a projection image limited in the working range polygon, driving the projection device to project the projection image to the surface of the map, and displaying situation enhancement information in a region corresponding to the working range polygon in an overlapping manner. According to the invention, the efficient multi-device cooperative work capability is realized, and the technical problem of large-size map AR enhancement is solved.

The invention provides an SPECT pinhole collimator imaging truncation artifact filling method and related equipment, and relates to the technical field of CT imaging. The method comprises the following steps: reconstructing real projection data to obtain an original reconstructed image, and determining a high-confidence region in the original reconstructed image; mapping the real projection data to a projection image of an ideal detector radial distance to obtain mapping projection data; predicting projection data outside an effective area of a real acquisition detector and taking the projection data as simulation projection data; carrying out weighted reconstruction on the simulation projection data to obtain a complemented image containing a truncation region; and according to the high-confidence region, fusing the original reconstructed image and the complemented image to obtain a final image. The method aims at solving the problem of truncation artifacts caused by the fact that the probe cannot completely cover the whole human body projection image, negative effects caused by the truncation artifacts can be made up, and the effect of improving the imaging quality and the diagnosis accuracy is achieved.

The invention provides a vehicle-mounted instrument driving record panoramic image fusion three-dimensional projection method and system, and relates to the technical field of vehicle-mounted display, and the method comprises the steps: obtaining original panoramic image data collected by cameras around a vehicle and real-time driving state information of the vehicle; distortion correction and multi-view splicing processing are carried out on the original image to generate a spliced panoramic image; determining the type of a driving scene according to the real-time driving state information, setting corresponding three-dimensional viewpoint parameters including viewpoint height, viewpoint distance and pitch angle, performing three-dimensional space coordinate transformation on the spliced panoramic image based on the parameters, and generating a three-dimensional aerial view image adaptive to the current driving scene; an obstacle target is extracted, the boundary of an image area of the obstacle target is recognized, height stretching rendering is carried out at the corresponding position of the three-dimensional aerial view image, and a three-dimensional panoramic projection image containing an obstacle three-dimensional identifier is generated; and finally adaptively outputting to a vehicle-mounted instrument panel display screen for real-time display. The visual angle can be dynamically adjusted according to different driving scenes, the stereoscopic perception of the obstacle is enhanced, and the driving safety is improved.

The invention relates to the field of computer vision, and discloses a laser radar and camera joint calibration method, electronic equipment and a storage medium. The method comprises the following steps: acquiring point cloud data of a plurality of laser radars and image data of a camera; performing motion distortion compensation on the point cloud data to generate a global point cloud map; converting the global point cloud map into a plurality of three-dimensional Gaussian voxels with color attributes and covariance attributes; minimizing the first loss function by jointly optimizing the attribute of the three-dimensional Gaussian voxel and the external parameter of the camera so as to complete the calibration of the external parameter of the camera; wherein the first loss function comprises a color consistency constraint term and a rendering error term, and the color consistency constraint term is used for constraining that observed color attributes of a single three-dimensional Gaussian voxel at multiple moments and multiple visual angles are kept consistent; the rendering error term is a pixel color error between a projection image generated by the three-dimensional Gaussian voxel and image data of the camera. Therefore, the precision of external parameter calibration in a dynamic scene is improved.

The invention discloses a radiotherapy equipment image guiding device and method and related equipment, and relates to the technical field of radiotherapy, and the device comprises a ray generation module which is provided with a ray source and is arranged at one side of a treatment room isocenter; the radiation source is used for moving along a first preset path; the ray source is used for emitting rays, and the rays penetrate through a patient; the detection and acquisition module is provided with a detector, is arranged on the other side of the isocenter of the treatment room and is distributed opposite to the ray source along the isocenter; the detector is used for moving along a second preset path, collecting rays penetrating through the patient and converting the rays into corresponding projection images; the registration module is used for performing registration on the projection image and a planned CT image by executing a first registration mode or a second registration mode so as to calculate a positioning deviation; and the first preset path and the second preset path are straight lines, so that the adaptability, precision and stability of radiotherapy are improved.

The invention provides an insulator defect identification method based on cross-modal data fusion. The insulator defect identification method comprises the following steps: acquiring and preprocessing an insulator visible light image, a thermal infrared image and point cloud data which are synchronously acquired; projecting all points in the point cloud to obtain a projection image and a projection mapping matrix; registering the visible light image and the thermal infrared image with the projection image to obtain a corresponding affine transformation matrix, and constructing a shielding matrix for shielding the background according to the projection image; performing feature extraction on the pre-processed visible light image, thermal infrared image and point cloud data to respectively obtain a texture feature map, a temperature feature map and a geometric feature map; fusing the texture feature map, the temperature feature map and the geometric feature map according to the shielding matrix and the affine transformation matrix to obtain cross-modal fusion features; and the cross-modal fusion features are used to discriminate and position insulator defects. According to the method, the problem of leak detection of insulator defects in different scenes by a traditional single-mode method can be solved.

Aspects of the method, apparatus, non-transitory computer readable medium, and system include estimating extrinsic camera parameters of a plurality of cameras based on detection of corners in a camera calibration arrangement by projecting an image of the detected corners onto an image plane of at least one of the plurality of cameras. The aspects further include incrementally tuning a location estimate for the detected corners based on pixels of the projected image, and adjusting the extrinsic camera parameters of the at least one of the plurality of cameras based on the tuned location estimate. The aspects further include generating a virtual image based on the pixels of the projected image mapped to a virtual image plane.

The invention relates to the technical field of projection modules, and particularly discloses an LCOS laser projection module, which comprises a laser light source module, a laser speckle eliminator, a reflector, a diffuser, a collimating lens, a fly-eye lens, a relay lens, a first linear polarizer, a polarizing beam splitter, an LCOS chip, a second linear polarizer and a projection lens which are sequentially arranged along a light transmission path, the laser light source module is provided with a light source optical axis and is used for emitting light, and the light is projected by the projection lens after being modulated along a light transmission path to form a projection image. The light emitted by the laser light source has the advantages of narrow divergence angle, small light-emitting area and good light concentration and collimation, the resolution of a projected image can be improved, the projected image is bright in color and high in definition, and the requirements of simple structure and small size are met. Meanwhile, the laser speckle can be eliminated through the matching of the laser speckle eliminator and the diffusion sheet, and the projection image can meet the high contrast through the existence of the first linear polarizer and the second linear polarizer.

The present invention provides an X-ray imaging apparatus and X-ray tube that can reduce artifacts caused by a limited number of imaging angles while suppressing the increase in imaging time required to acquire projected image data. [Solution] This X-ray imaging apparatus 100 comprises an X-ray tube 1 including a first electron irradiation unit 10a and a second electron irradiation unit 10b, a detector 2, a subject placement unit 3, a rotation mechanism 4, and a control unit 21. The control unit 21 performs the following controls: for each of the multiple imaging angles 6, it controls the simultaneous irradiation of electrons from the first electron irradiation unit 10a and the second electron irradiation unit 10b; and it controls the first relative focal position 7a of the first electron irradiation unit 10a at the first imaging angle 6a, the second relative focal position 7b of the second electron irradiation unit 10b at the first imaging angle 6a, the third relative focal position 7d of the first electron irradiation unit 10a at the second imaging angle 6b, and the fourth relative focal position 7e of the second electron irradiation unit 10b at the second imaging angle 6b, so as not to overlap with each other.

A projection method based on the three-dimensional scanning projection line of the structured light relates to three-dimensional information reconstruction technology field and comprises: switching different projection image groups periodically or randomly in three-dimensional measurements, to collect a plurality of continuous projection image groups (S101); projecting several images of the projection image group onto a surface of an object to be measured (S102); imaging projection images projected onto the surface of the object to be measured to obtain imaged images (S103); and decoding the imaged images and determining a corresponding relation between respective imaging positions and projection positions, to provide three-dimensional information for reconstructing the surface of the object to be measured (S104).

A computer implemented method for determining a 3D position of a surgical object relative to a region of interest in a patient's anatomy. The method includes receiving an X-ray image and a 3D data set. The X-ray image is a 2D projection image generated with an imaging direction and defining a 3D image coordinate system. The 2D projection image may depict a part of a patient's anatomy and a part of the surgical object. The received 3D data set describes the part of the patient's anatomy and a region of interest in a 3D data coordinate system. In order to determine a spatial relation between the 3D image coordinate system and the 3D data coordinate system, the part of the patient's anatomy is localized based on the X-ray image. Subsequently, the 3D position of the surgical object relative to the region of interest is determined in the 3D coordinate system.

The invention discloses a curved surface film covering flatness self-adaptive control method based on machine vision, and relates to the technical field of mechanical arm control. The system comprises a machine vision sensing module, a robot end effector, a composite controller and a distributed clamping jaw array. The method comprises the following steps: acquiring a fringe projection image sequence and dynamic trajectory data of an end effector; performing phase calculation and three-dimensional reconstruction on the image sequence, constructing a digital elevation model and calculating flatness error data; inputting the dynamic trajectory data into a feedforward controller to generate a compensation instruction, inputting the flatness error data into a feedback controller to generate a correction instruction, and superposing to obtain a comprehensive correction instruction; and acquiring a strain sensitivity matrix, distributing the comprehensive correction instruction into displacement execution parameters of each clamping jaw by utilizing pseudo inverse operation, and issuing and executing the displacement execution parameters. According to the method, real-time sensing and active intervention of dynamic deformation are realized, and the yield and batch consistency of curved surface film coating are remarkably improved.

An augmented reality collaboration system includes an assembly platform, a first projector, a camera and a computing circuit. The first projector projects a first projection image to the assembly platform. The camera shoots the assembly platform and the first projection image projected on the assembly platform to obtain a first shooting image. The computing circuit creates a three-dimensional model in a virtual three-dimensional space, arranges a standard operating procedure, sets a mark pattern on the three-dimensional model according to the standard operating procedure, analyzes the first shooting picture, and performs a calibration operation on the first projector to obtain a first calibration parameter. The computing circuit enables the first projector to generate a first corrected pattern according to the first calibration parameter and the mark pattern, and projects the first corrected pattern to the object-to-be-assembled according to the standard operating procedure.

The invention relates to a three-dimensional projection deformation measurement and error compensation method based on a single camera, and the method comprises the steps: collecting a plurality of projection images at multiple moments through a single camera, and recording an image intensity value at each pixel coordinate in each projection image at each moment; inputting the image intensity value into a pre-trained deep learning model, and outputting three-dimensional height prediction of a corresponding pixel coordinate at a corresponding moment; performing correlation search based on the image intensity values of the pixel coordinates in each projection image at the reference moment and the deformation moment to obtain an optimal pixel displacement component; based on the three-dimensional height prediction of each pixel coordinate at the reference moment and the deformation moment, constructing an image surface displacement vector at the deformation moment by using the optimal pixel displacement component, calculating a three-dimensional displacement field at the deformation moment based on the image surface displacement vector at the deformation moment, and then performing three-dimensional strain field calculation; and finally, an efficient and robust closed-loop process is formed through error compensation, and measurement and compensation of three-dimensional deformation in a single-camera scene are realized.

The invention discloses an automatic driving scene rapid high-fidelity reconstruction method based on semantic guidance, and belongs to the technical field of automatic driving and computer vision. The invention aims to solve the problem that the fidelity of a security key object is insufficient due to the fact that resources are allocated to all elements in a scene without distinguishing in the prior art. Comprising the following steps: establishing a three-dimensional Gaussian full-scene synthesis model; obtaining pixel-level masks of all scene elements in the two-dimensional image of the target scene; calculating the overlapping degree of the two-dimensional Gaussian projection image and the key category semantic mask to obtain a static semantic score, and combining the gradient contribution score to obtain a mixed importance score; carrying out iterative optimization on the three-dimensional Gaussian full-scene synthesis model; carrying out adaptive pruning and random selective discarding based on semantic guidance; and according to the three-dimensional Gaussian configuration high-fidelity or low-fidelity label, rendering is carried out respectively, and a high-fidelity reconstruction image is obtained. According to the invention, task-oriented efficient and high-fidelity reconstruction is realized.

The invention relates to the field of three-dimensional vision, in particular to a label-free three-dimensional point cloud segmentation method based on a visual large model, which comprises the following steps: acquiring three-dimensional point cloud data, and extracting a boundary point set according to the three-dimensional point cloud data in combination with curvature and normal vector to construct a target function; solving the objective function by using an optimization algorithm, selecting an optimal projection visual angle combination, and extracting two-dimensional features of all two-dimensional projection images and three-dimensional features of three-dimensional point cloud data by using a visual large model; performing consistency matching on the two-dimensional features and the three-dimensional features through a matching algorithm to obtain a plurality of matching results; acquiring a definition weight factor and a coverage rate weight factor to calculate an importance weight; fusing the two-dimensional features according to the weight factors to obtain overall fused two-dimensional features; and inversely mapping the integrally fused two-dimensional features to a three-dimensional space to realize three-dimensional point cloud unmarked segmentation. The three-dimensional point cloud segmentation method has the effect of realizing three-dimensional point cloud automatic segmentation without labels.

The invention provides a panel lamp light spot uniformity detection method and system and electronic equipment, and the method comprises the steps: analyzing the non-uniform characteristics of brightness gradient distribution according to the intensity and range information of a light spot region, carrying out the image processing of a projection image region, carrying out the quantitative evaluation of the brightness uniformity of a screen, and determining the specific region and degree data of the non-uniform brightness; if the dynamic influence range of the light interference exceeds a preset threshold value, meeting room layout adjustment and equipment position change are determined according to a light interference distribution diagram and equipment configuration parameters, the adjusted light distribution mode change is identified, and the detection priority is determined; according to the detection priority and the brightness deviation index, a brightness compensation method is adopted to carry out local correction on the specific area with the uneven brightness, and corrected brightness distribution data are obtained; and identifying the image definition index and the color rendition degree index after the brightness distribution is corrected, obtaining an area coverage proportion for clearly presenting the information, and judging whether the projection display effect reaches a preset definition standard value or not.

The invention discloses an anti-fatigue driving signal lamp and atmosphere lamp control method and a vehicle, and relates to the technical field of vehicle intelligent auxiliary driving and active safety, and the method comprises the steps: collecting a cab video stream, and positioning face key points after image preprocessing; calculating an eye PERCLOS index and a mouth yawning frequency, and dynamically adjusting a weight based on a face angle so as to fuse and calculate a comprehensive fatigue index; carrying out grading judgment on the index by utilizing a hysteresis interval containing a starting threshold value and a reset threshold value, and reducing a judgment threshold value in real time by utilizing a linear model according to the vehicle speed; and driving an indoor atmosphere lamp to execute a breathing or stroboscopic mode according to the judged sober, mild or severe fatigue grade, controlling an outdoor signal lamp and a road surface projection assembly to execute corresponding warning actions, and performing reverse pre-distortion processing on a projected image. According to the invention, through multi-source feature fusion, hysteresis threshold determination and vehicle speed correction mechanisms, the stability of fatigue monitoring and the accuracy of warning are improved.

An X-ray imaging device emits an X-ray to irradiate an object and includes an object moving mechanism that rotates the object; an X-ray image detector to detect a projection image of the object; an object temperature adjusting mechanism configured to change a temperature of the object; and a processing unit configured to acquire a cross-sectional image of the object by a reconstruction calculation from the projection images detected by the X-ray image detector by rotating the object. The processing unit is acquires a plurality of cross-sectional images and of the object captured at a plurality of different temperatures T1 and T2, and divides an inside of the object into regions for each of substances using a distribution of pixels of each of the cross-sectional images based on CT values obtained at the plurality of temperatures T1 and T2 at which the cross-sectional images of the object are acquired.

The invention discloses a multi-label field adaptive method based on prompt driving, which belongs to the technical field of computer vision and comprises the following steps: generating multi-dimensional semantic text description containing visual attributes, semantic levels and category names of common scenes for each type; embedding the semantic text description into an optimizable category vector, and embedding the optimized category vector into a CLIP text prompt; extracting and projecting multi-layer style statistical characteristics of the image, and synchronously realizing image-text cross-modal alignment and source-target domain distribution alignment in a frame through a cross-domain style mapping network and various alignment losses; semantic priori knowledge of CLIP is combined with a label co-occurrence mode of a source domain, so that the model can sense and adapt to a possibly changed label dependency relationship in a target domain; performing semantic propagation on label embedding by using a graph convolutional network; the whole system is jointly optimized through a multi-task loss function, and end-to-end cross-domain multi-label classification is achieved.

An information display device that displays image information as a virtual image on a projection surface, the device including: an opening portion at a part of the housing, image light generating means for generating image light, image light processing means for performing a predetermined optical process on the image light, and image projecting means for projecting the image light such that a viewer can recognize the image information as the virtual image in front of the projection surface, and an optical element in which a reflective polarizing plate is fixed to a substrate having a transmissive property with an adhesive agent or with a pressure-sensitive adhesive agent is provided in a part of an optical path inside the housing, as means for selectively reflecting a P-polarized light component of light in a visible light region, and a moisture-proof film is provided on a surface of the reflective polarizing plate.

The invention provides a volume photocuring three-dimensional printing method and device based on oxygen permeation interface dynamic oxygen inhibition control, and relates to the technical field of photocuring additive manufacturing. The volume photocuring three-dimensional printing method comprises the following steps that a resin cavity is prepared, the closed container is formed by compounding an oxygen permeable membrane and a transparent rigid material; establishing an oxygen inhibition layer: maintaining air or oxygen-containing atmosphere outside the closed container, so that oxygen continuously permeates into a resin boundary area through the oxygen permeation membrane to form the stable oxygen inhibition layer; volume light field projection: projecting a projection image generated by a computed tomography algorithm into the resin volume by adopting a calculation axial photoetching type multi-angle projection mode to realize multi-angle rotary projection exposure; edge curing is delayed through the boundary oxygen inhibition layer, the problem that the boundary light dosage of traditional volume printing is not uniform is solved, the final forming volume is enlarged by 1.5-3 times compared with the prior art, and the size bottleneck of volume light curing printing is effectively solved.

The invention provides a three-dimensional object labeling method and system and electronic equipment. The method comprises the following steps: constructing a visual large model for identifying features and feature coordinates in a two-dimensional projection drawing; obtaining a target two-dimensional projection drawing of the target three-dimensional object under various preset visual angles, and identifying the target two-dimensional projection drawing by using the visual large model to generate an object file containing target features and target feature two-dimensional coordinates; determining a first three-dimensional coordinate of the target feature according to a projection matrix between target feature two-dimensional coordinates under each target two-dimensional projection drawing in the object file; and performing intersection detection in the target three-dimensional object based on the first three-dimensional coordinate to correct the first three-dimensional coordinate to obtain a second three-dimensional coordinate, and generating a label containing semantic feature information and coordinate information. Through multi-view information fusion and intersection check, random errors and view ambiguity of single identification can be eliminated, an optimal solution in probability is calculated, the processing efficiency is improved, and manual intervention is reduced.