Digital auxiliary repair method for ancient building

Abstract

A digital auxiliary repair method for an ancient building. The method comprises: point cloud data acquisition, point cloud model establishment, point cloud model optimization and model printing, which involve: scanning and photographing a damaged ancient building by using a three-dimensional scanner and an industrial camera, so as to obtain three-dimensional point cloud data of the ancient building; performing preprocessing, such as data registration, data denoising, data simplification and data segmentation, on the point cloud data to obtain data for establishing a three-dimensional model; optimizing the three-dimensional model; and by means of a 3D printer, obtaining a mold for a damage position of the ancient building, and on the basis of requirements, using the mold to perform molding to obtain a wooden or brick-mixed workpiece.

Description

A digital assisted restoration method for ancient buildings Technical Field

[0001] The present invention relates to the technical field of ancient building restoration, and in particular to a digital-assisted restoration method for ancient buildings. Background Art

[0002] Ancient buildings carry the long history of Chinese culture, and their protection is of great practical significance. Only by preserving cultural relics can their historical, artistic, and scientific value be reflected. The restoration of ancient buildings is extremely important. Strengthening severely damaged buildings and extending their lifespan is a key step in the restoration process.

[0003] Existing restoration methods for ancient buildings typically adhere to traditional methods, employing methods such as one hemp and five ash, one cloth and four ash, or a single ash layer, depending on quality requirements. For example, Chinese invention patent 201610054835.4 discloses a "system and method for simulating and restoring ancient building components." While its fundamental purpose is to restore the original appearance of ancient buildings, these restoration projects are prone to numerous flaws. For example, the filling of damaged areas on ancient buildings can be influenced by subjective considerations, leading to historical distortion, or the overall disharmony of the building after restoration. Therefore, based on the foregoing, those skilled in the art have proposed a digitally assisted restoration method for ancient buildings.

[0004] Summary of the Invention

[0005] In response to the aforementioned shortcomings, the present invention aims to overcome the shortcomings of the prior art by providing a digitally assisted restoration method for ancient buildings. This method utilizes modern methods to digitally process the damaged area to obtain a three-dimensional image and construct a three-dimensional digital model, thereby achieving high-precision repair of damaged areas of ancient buildings.

[0006] In order to achieve the above-mentioned purpose of the invention, the technical solution provided by the patent of this invention is as follows;

[0007] A digital assisted restoration method for ancient buildings, which specifically includes the following assisted restoration process:

[0008] S1: point cloud data acquisition;

[0009] Conduct on-site surveys of ancient buildings, and use 3D scanners and industrial cameras to scan and photograph damaged ancient buildings. After obtaining the 3D scanning point cloud data, pre-process the point cloud data. This step allows the use of 3D scanners and industrial cameras to conduct on-site surveys and data acquisition of ancient buildings, thereby obtaining relatively accurate point cloud data of ancient buildings.

[0010] S2: Build point cloud model;

[0011] Integrate the parameters of the reference ancient buildings, import point cloud data and photos into the 3D model software, and use the 3D reconstruction algorithm in the 3D model software to generate a 3D model and the image texture of each grid face. This step can generate a relatively realistic 3D model with a certain image texture by integrating the reference parameters and point cloud data and using the 3D reconstruction algorithm in the 3D model software.

[0012] S3: point cloud model optimization;

[0013] The obtained three-dimensional model of the ancient building is subjected to fine texture processing and model monomer modification processing; since some damage and details of the ancient building require fine texture processing, this step is very necessary to ensure the details and accuracy of the ancient building.

[0014] S4: Model printing;

[0015] After obtaining the optimized three-dimensional model of the ancient building in step S3, the model data is output to a 3D printer to obtain a mold of the damaged position of the ancient building. The mold is used to perform reverse molding to obtain a workpiece made of wood or brick-concrete material as needed. The output result of this step has a high degree of accuracy, and molds of different materials can be obtained as needed for reverse molding of the workpiece.

[0016] S5: placing the workpiece obtained in step S4 to the damaged part of the ancient building. This step can make the restored ancient building more complete, and thus has a certain reference value for preserving and repairing ancient buildings.

[0017] In the above technical solution of the digital-assisted restoration method for ancient buildings, preferably, the point cloud data preprocessing includes:

[0018] ① Data registration: Point cloud data registration is a key step in point cloud data processing. Using targets for inter-station calibration can improve point cloud data registration accuracy. Utilizing overlapping areas and points of the same name can further improve point cloud data registration accuracy while ensuring the correctness of the stitching results.

[0019] ② Data denoising: Point cloud data denoising can effectively reduce noise in point cloud data, improve point cloud data quality, and thus improve the accuracy of subsequent processing. However, it should be noted that excessive denoising may lead to problems such as data loss and unrealistic models.

[0020] ③ Data reduction: Data reduction can remove redundant and useless information from point cloud data, thereby improving point cloud data processing efficiency. However, the reduction process must be cautious to avoid excessively damaging important information in the point cloud data and affecting the accuracy of subsequent processing.

[0021] ④ Data segmentation: Data segmentation can divide point cloud data into different parts for subsequent accurate processing. For example, the roof, doors, windows, columns, etc. can be segmented and processed separately. This can improve processing efficiency while ensuring the authenticity of the 3D model.

[0022] Data registration specifically includes:

[0023] Stitching and coordinate correction of point cloud data;

[0024] Among them, splicing and correction include:

[0025] Place three or more targets in the common area between the two scanning stations, scan the data and targets of each station in turn, and finally use the same target data from different stations to perform point cloud registration. Each target corresponds to an ID number, and the ID number of the same target at different stations must be consistent;

[0026] During the stitching process, there must be a certain degree of regional overlap when scanning the target object, and the target object feature points must be obvious, and stitching must be performed by finding the same-name points in the overlapping area.

[0027] In the above technical solution of the digital assisted restoration method for ancient buildings, preferably, the data denoising specifically includes:

[0028] Obtain the size and data area information of the data on the terminal computer, and copy the data area to the buffer; overlap the center value of the data with the set pixel position of the filter template; read the grayscale values ​​of each corresponding pixel under the template, sort these grayscale values ​​from large to small, find the one in the middle, and assign the middle value to the center of the corresponding template to obtain the N value; and loop to obtain the pixel value of each point; sort the N*N shielding window centered on the pixel value of the point, including the pixel values ​​of each point, to obtain the middle value; use the middle value as the data acquisition range, and shield the data that is too large or too small to obtain the denoised data.

[0029] The data denoising method adopts the idea based on median filtering and implements denoising through the following steps:

[0030] Get data area information and copy it to the buffer: Get the data size and data area information on the terminal computer, and copy the data in the area to the buffer for subsequent processing.

[0031] Set the filter template: align the center value of the data with the set pixel position of the filter template. The filter template is generally an N*N window.

[0032] Calculate the median N: read the grayscale values ​​of each corresponding position under the filter template, and sort these grayscale values ​​from large to small, then find the value of the middle position after sorting, and assign this value to the center of the template to obtain the median N.

[0033] Loop through the pixel values ​​of each point: Based on the range of the median N, with the pixel value of each point as the center, construct a masking window of size N*N, including each pixel value, and sort these pixel values.

[0034] Get the middle value: Sort the pixel values ​​in the mask window and get the value at the middle position after sorting.

[0035] Perform data masking: Use the middle value as the data acquisition range and mask data that is too large or too small.

[0036] Through the above steps, the denoised data can be obtained.

[0037] Median filtering is a commonly used denoising method that effectively removes noise from data while preserving edge information and detailed features. The specific effect depends on the size of the filter template and the characteristics of the noise. Properly adjusting parameters and using appropriate filtering methods can achieve good denoising results. However, it should be noted that different scenarios and data characteristics may require other more suitable denoising methods, so it is necessary to choose the appropriate method based on the actual situation.

[0038] In the above technical solution of the digital-assisted restoration method for ancient buildings, preferably, the data reduction specifically includes:

[0039] By removing redundancy and simplifying the point cloud data, effective information can be extracted;

[0040] Redundancy removal refers to the data in the repeated area after data registration. This part of the data is mostly useless data and has a great impact on the speed and quality of modeling. This part of the data must be removed;

[0041] The point cloud data is layered and thinned to form a pyramid-like point cloud pyramid model, forming a pyramid structure with the point cloud sparsity ranging from sparse to dense and the data volume ranging from small to large. The point cloud data needs to be thinned and the data organized into blocks. First, the original file is thinned to obtain a sparser point cloud, which is saved as a file. Then, thinning is performed on this basis until the layering is completed. The thinned point cloud data is then divided and stored in blocks. When the point cloud is displayed, the corresponding block data is loaded according to the display area.

[0042] In the above technical solution of the digital-assisted restoration method for ancient buildings, preferably, the data segmentation specifically includes:

[0043] Data segmentation criteria:

[0044] The features of the block area are single and there is no sudden change in normal vector and curvature in the same area;

[0045] The common edges of the split should be as convenient as possible for subsequent splicing;

[0046] The number of blocks should be as small as possible to reduce the complexity of subsequent splicing;

[0047] Each piece after segmentation should be easy to reconstruct the geometric model;

[0048] Edge-based segmentation methods require finding feature lines first. Curvature and normal vector extraction methods usually consider points where the curvature or normal vector suddenly changes as feature points, such as inflection points or corners. After extracting the feature lines, the area enclosed by the feature lines is segmented.

[0049] The surface-based method is an iterative process that finds points with the same surface properties, divides the set of points with the same basic geometric features into the same area, and then determines the surface to which these points belong. Finally, the boundaries between surfaces are determined by the adjacent surfaces.

[0050] The clustering-based method is to classify data points with similar geometric feature parameters. The geometric features can be calculated based on Gaussian curvature and mean curvature, and then clustered, and finally divided according to the class to which they belong.

[0051] In the above technical solution of the digital-assisted restoration method for ancient buildings, preferably, the refined texture processing includes:

[0052] By using oblique images, vertical and four-directional cameras are used to simultaneously capture the same surface, or multiple images from multiple angles are captured at the same time. For unreasonable textures such as occlusion and large color differences, the manual mapping function is needed to perform fine-tuning of the texture.

[0053] In the technical solution of the above-mentioned digital assisted restoration method for ancient buildings, preferably, vector cutting and singulation are performed: using continuous oblique photography data as the basic data source, manual intervention is performed, and the modeling results are imported into the software for refined editing. The separation effect is achieved on the original scene through model reconstruction, and the model is singulated. Combined with ground photos, the missing information of the aerial image on the ground floor, the ground, and urban components is compensated, and the entire scene is modified while the singulation is completed;

[0054] Vector overlay individualization: In the 3D rendering process, GIS application software dynamically overlays the corresponding vector surface onto the oblique photography model. The vector surface generates a bounding box perpendicular to the surface within a certain threshold range. It determines which triangulated networks are within the range and fits the triangulated networks with semi-transparent colors, visually achieving the effect of complete model fitting and individual management operations.

[0055] As can be seen from the above technical solutions, the present invention provides a digital assisted restoration method for ancient buildings. Compared with the prior art, the present invention has the following beneficial effects:

[0056] In the present invention, a 3D scanner and an industrial camera are used to scan and photograph damaged ancient buildings to obtain 3D point cloud data of the ancient buildings. The point cloud data is preprocessed by data alignment, data denoising, data simplification and data segmentation to obtain data for building a 3D model. The 3D model is then optimized, and a mold of the damaged location of the ancient building is obtained by a 3D printer, or a workpiece made of wood or brick-concrete is obtained by casting the mold as needed, thereby achieving high-precision repair of damaged areas of the ancient building. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces and describes the drawings required for use in the embodiments of the present invention or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0058] FIG1 is a schematic diagram of a framework of a digital-assisted restoration method for ancient buildings according to the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In order to make a clearer explanation and description of the technical solutions and implementation methods of the present invention, several preferred specific embodiments for implementing the technical solutions of the present invention are introduced below.

[0060] Example 1

[0061] Referring to FIG. 1 , this embodiment is a digital-assisted restoration method for ancient buildings. The restoration process of digital ancient buildings involved in this method is as follows:

[0062] Point cloud data acquisition;

[0063] Conduct on-site surveys of ancient buildings, and use 3D scanners and industrial cameras to scan and photograph damaged ancient buildings. After obtaining the 3D scanning point cloud data, the point cloud data is pre-processed.

[0064] Point cloud data preprocessing includes data registration, data denoising, data reduction and data segmentation;

[0065] Data registration specifically includes:

[0066] Stitching and coordinate correction of point cloud data;

[0067] Among them, splicing and correction include:

[0068] Place three or more targets in the common area between the two scanning stations, scan the data and targets of each station in turn, and finally use the same target data from different stations to perform point cloud registration. Each target corresponds to an ID number, and the ID number of the same target at different stations must be consistent;

[0069] The point cloud-based stitching method requires a certain degree of regional overlap when scanning the target object, and the target object's feature points must be obvious. Otherwise, the data cannot be stitched together, and stitching must rely on finding the same-name points in the overlapping area. Therefore, the determination of the feature points in the overlapping area is directly related to the quality of the registration result.

[0070] The geodetic coordinates of the common control points are determined by a total station or GPS, and then all the common control points are accurately scanned using a 3D laser scanner. The point cloud data of the scanned multiple stations are then directly spliced ​​with the control points as base stations, and all the scanned point cloud data can be converted into the coordinate system actually required for the project.

[0071] Data denoising specifically includes:

[0072] Based on the ordered point cloud data, the smoothing filter denoising method is used. Currently, the data smoothing filters mainly adopt Gaussian filtering, mean filtering and median filtering;

[0073] Gaussian filtering is a linear smoothing filter that performs a weighted average of data within a specified area and can remove high-frequency information.

[0074] Mean filtering, also known as average filtering, is a typical linear filter. Its principle is to select points within a certain range and calculate their average value to replace the original data points.

[0075] Median filtering is a nonlinear smoothing filter. Its principle is to find the median of three or more adjacent data points of a certain point, and the result replaces the original value.

[0076] Commonly used methods for denoising scattered point cloud data include Laplace denoising, mean curvature flow method, and bilateral filtering algorithm;

[0077] Although the Laplace algorithm can well guarantee the detailed features of the model; the mean curvature depends on the curvature estimation, and has a better denoising effect on data with simple models and fewer noise points; the bilateral filtering algorithm can effectively remove noise points.

[0078] Data reduction specifically includes:

[0079] By removing redundancy and simplifying the point cloud data, effective information can be extracted;

[0080] Redundancy removal refers to the data in the repeated area after data registration. This part of the data is large and mostly useless, which has a great impact on the speed and quality of modeling. This part of the data must be removed;

[0081] Thinning and simplification means that the scanned data density is too high and the quantity is too large, and part of the data is not very useful for later modeling. Therefore, under the premise of meeting a certain accuracy and maintaining the geometric characteristics of the measured object, the data is simplified to improve the data operation speed, modeling efficiency and model accuracy.

[0082] Data segmentation specifically includes:

[0083] Data segmentation criteria:

[0084] The features of the block area are single and there is no sudden change in normal vector and curvature in the same area;

[0085] The common edges of the split should be as convenient as possible for subsequent splicing;

[0086] The number of blocks should be as small as possible to reduce the complexity of subsequent splicing;

[0087] Each piece after segmentation should be easy to reconstruct the geometric model;

[0088] Edge-based segmentation methods require finding feature lines first. Curvature and normal vector extraction methods usually consider points where the curvature or normal vector suddenly changes as feature points, such as inflection points or corners. After extracting the feature lines, the area enclosed by the feature lines is segmented.

[0089] The surface-based method is an iterative process that finds points with the same surface properties, divides the set of points with the same basic geometric features into the same area, and then determines the surface to which these points belong. Finally, the boundaries between surfaces are determined by the adjacent surfaces.

[0090] The clustering-based method is to classify data points with similar geometric feature parameters. The geometric features can be calculated based on Gaussian curvature and mean curvature, and then clustered, and finally divided according to the class to which they belong.

[0091] Build a point cloud model;

[0092] Integrate the parameters of the reference ancient buildings, import point cloud data and photos into the 3D model software, and use the 3D reconstruction algorithm in the 3D software to generate a 3D model and the image texture of each grid surface.

[0093] Point cloud model optimization;

[0094] The obtained three-dimensional model of the ancient building is subjected to fine texture processing and individual model modification.

[0095] Refined texture processing includes:

[0096] By using oblique images and using vertical and four-directional cameras to simultaneously capture the same surface, multiple images from multiple angles can be collected simultaneously. For unreasonable textures such as occlusion and large color differences, the manual mapping function is needed to perform fine-tuning of the texture.

[0097] The texture of the building's top surface is usually acquired using images captured by a vertical camera. The best effect is achieved when the acquisition direction of the top surface is consistent with the normal direction of the facet. Textures projected onto the top surface by temporarily stacked objects or adjacent buildings must be modified. In a building facade facet, one face corresponds to multiple oblique images. You can switch between the oblique images one by one for comparison, and select the best texture to directly replace the defective texture.

[0098] For example, check each oblique image one by one or check by shooting angle, and manually select the best texture. If there are multiple surfaces with the same texture, and one of the surfaces cannot meet the mapping requirements from all viewing angles for various reasons, the texture of another surface with the same texture can be selected as a replacement.

[0099] Use small-scale texture replacement. For solid-color walls, due to occlusion and other reasons, there is no image that can completely cover the entire wall surface. You can use a small-scale texture to map it to the entire large surface of the model, and use part of the texture to replace the texture of the entire surface.

[0100] Use Photoshop software to edit textures. When images from multiple perspectives cannot meet the texture requirements, there is no identical texture to replace them, and small-scale texture mapping is not possible, you can use Photoshop software to edit the texture. Select the surface that needs to be modified, switch to manual mapping, and link Photoshop software to modify the texture. For surfaces with the same texture as the exposed part, the existing texture can be used to derive the texture of the occluded part.

[0101] Model monomer modification processing includes:

[0102] Vector cutting and singulation: Using continuous oblique photography data as the basic data source, manual intervention is performed, and the modeling results are imported into the software for refined editing. Through model reconstruction, a separation effect is achieved on the original scene, and the model is singulated. Combined with ground photos, this method compensates for the lack of information about ground floor shops, ground, and urban components in aerial images. While singulation is being completed, the entire scene is modified.

[0103] Vector overlay individualization: In the 3D rendering process, GIS application software dynamically overlays the corresponding vector surface onto the oblique photography model. The vector surface generates a bounding box perpendicular to the surface within a certain threshold range, determines which triangulations are within its range (including aerial triangulations), and fits the triangulations with semi-transparent colors, visually achieving the effect of a complete model fit and individual management operations.

[0104] Model printing;

[0105] After obtaining the optimized three-dimensional model of the ancient building, the model data is output to the 3D printer to obtain a mold of the damaged position of the ancient building. The mold is then used to cast a wooden or brick-concrete workpiece as needed.

[0106] Finally, it should be noted that the structures, proportions, sizes, etc. depicted in the drawings of this specification are intended solely to complement the contents disclosed herein and to facilitate understanding and reading by those familiar with the art. They are not intended to limit the conditions under which this application can be implemented and therefore have no substantive technical significance. Any structural modifications, changes in proportions, or adjustments in size, provided they do not affect the efficacy and objectives of this application, should still fall within the scope of the technical content disclosed in this application. The present invention is not limited to the above-described preferred embodiment. Anyone should be aware that any structural changes made under the guidance of this invention, as long as they have the same or similar technical solutions as the present invention, fall within the scope of protection of this invention.

Claims

1. A method for digital-assisted restoration of ancient buildings, characterized in that, The following auxiliary repair processes are included: S1: point cloud data acquisition; Conduct on-site surveys of ancient buildings, and use 3D scanners and industrial cameras to scan and photograph damaged ancient buildings. After obtaining 3D scanning point cloud data, pre-process the point cloud data; S2: build point cloud model; Integrate the parameters of the reference ancient buildings, import point cloud data and photos into the 3D model software, and use the 3D reconstruction algorithm in the 3D software to generate the 3D model and the image texture of each grid surface; S3: point cloud model optimization; The obtained three-dimensional model of the ancient building is subjected to fine texture processing and the model is subjected to monomer modification processing; S4: Model printing; After the optimized three-dimensional model of the ancient building is obtained in step S3, the model data is output to a 3D printer to obtain a mold of the damaged position of the ancient building, and the mold is used to perform reverse molding to obtain a workpiece made of wood or brick-concrete material as needed; S5: placing the workpiece obtained in step S4 to the damaged part of the ancient building.

2. The method for digitally assisted restoration of ancient buildings according to claim 1, characterized in that, The point cloud data preprocessing in step S3 includes: ①Data registration; ②Data denoising; ③Data reduction; ④Data segmentation; Among them, data registration specifically includes: Stitching and coordinate correction of point cloud data; Among them, splicing and correction include: Place three or more targets in the common area between the two scanning stations, scan the data and targets of each station in turn, and finally use the same target data from different stations to perform point cloud registration. Each target corresponds to an ID number, and the ID numbers of the same target at different stations must be consistent. During the stitching process, there must be a certain degree of area overlap when scanning the target object, and the feature points of the target object must be obvious, and it is necessary to stitch by finding the same-name points in the overlapping area.

3. The method for digitally assisted restoration of ancient buildings according to claim 2, characterized in that, The specific aspects of the ② data denoising are as follows: Obtain the size and data area information of the data on the terminal computer, and copy the data area to the buffer; overlap the center value of the data with the set pixel position of the filter template; read the grayscale values ​​of each corresponding pixel under the template, sort these grayscale values ​​from large to small, find the one in the middle, and assign the middle value to the center of the corresponding template to obtain the N value; and loop to obtain the pixel value of each point; sort the N*N shielding window centered on the pixel value of the point including the pixel values ​​of each point to obtain the middle value; use the middle value as the data acquisition range, and shield the data that is too large or too small to obtain the denoised data.

4. The method for digitally assisted restoration of ancient buildings according to claim 2, wherein The data reduction specifically includes: By removing redundancy and simplifying the point cloud data, effective information can be extracted; Redundancy removal refers to the data in the repeated area after data registration. This part of data is mostly useless data and has a great impact on the speed and quality of modeling. This part of data should be removed; The point cloud data is thinned in layers to form a pyramid-like point cloud pyramid model, forming a pyramid structure with point cloud sparsity from sparse to dense and data volume from small to large. The point cloud data needs to be thinned and organized into blocks. First, the original file is thinned to obtain a sparse point cloud. The point cloud is saved as a file, and then thinning is performed on this basis until the stratification is completed, and then the thinned point cloud data is divided into blocks for storage, and the corresponding block data is loaded according to the display area when the point cloud is displayed.

5. A method for digital assisted restoration of ancient buildings according to claim 2, characterized in that, The data segmentation (4) specifically includes: Data segmentation criteria: The features of the block area are single and there is no sudden change in normal vector and curvature in the same area; The common edges of the segments should be as convenient as possible for subsequent splicing; The number of blocks should be as small as possible to reduce the complexity of subsequent splicing; Each segmented piece should be easy to reconstruct the geometric model; The edge-based segmentation method needs to find the feature lines first. The curvature and normal vector extraction method usually considers the points where the curvature or normal vector suddenly changes as feature points, such as inflection points or corner points. After the feature lines are extracted, the area enclosed by the feature lines is segmented. The surface-based method is an iterative process that finds points with the same surface properties, divides the point set with the same basic geometric features into the same area, and then determines the surface to which these points belong. Finally, the boundary between surfaces is determined by the adjacent surfaces. The clustering-based method is to classify data points with similar geometric feature parameters. The geometric features can be calculated based on Gaussian curvature and mean curvature, and then clustered, and finally segmented according to the class to which they belong.

6. A method for digital assisted restoration of ancient buildings according to claim 1, characterized in that The refined texture processing in step S3 includes: By using oblique images, vertical and 4-directional cameras are used to simultaneously capture the same surface, or multiple images from multiple angles are captured at the same time. For unreasonable textures such as occlusion and large color differences, the manual mapping function is needed to perform fine-tuning of the texture.

7. A method for digitally assisted restoration of ancient buildings according to claim 1, characterized in that, The model monomer modification process includes: Vector cutting and individualization: With continuous oblique photography data as the basic data source, manual intervention is carried out, and the modeling results are imported into the software for refined editing. The separation effect is achieved on the original scene through model reconstruction, and the model is individualized. Combined with ground photos, it makes up for the lack of information on ground floor shops, ground, urban components, etc. in aerial images. While individualizing, the modification of the entire scene is completed; Vector overlay individualization: In the 3D rendering process of GIS application software, the corresponding vector surface is dynamically superimposed on the oblique photography model. The vector surface generates a bounding box perpendicular to the surface within a certain threshold range, determines which triangulated networks are within its range, and fits the triangulated networks with semi-transparent colors, visually achieving the effect of the model being completely fitted and individually managed.

Images