Historical building three-dimensional supervision modeling method based on multi-source geographic data fusion

Abstract

The application discloses a historical building three-dimensional supervision modeling method based on multi-source geographic data fusion, and belongs to the technical field of data processing, and comprises the following steps: S1, obtaining each image region of a historical building image; S2, obtaining each secondary division region; S3, extracting feature vectors of each specified secondary division region of an image group, and performing feature vector similarity analysis to obtain the comprehensive feature vector similarity of the image group; S4, obtaining a 3D adjacent determination result, if the 3D adjacent determination result is adjacent, then the image group is spliced, otherwise, the temporary storage location of the image group is obtained by analysis, and the adjacent judgment of the image group is continued until the complete image splicing relationship is obtained; and S5, obtaining a historical building three-dimensional model. The historical building three-dimensional model is accurately constructed, and the problems of low three-dimensional modeling accuracy and low efficiency caused by inaccurate feature vector acquisition of historical buildings in the prior art are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a historical building three-dimensional supervision modeling method based on multi-source geographic data fusion. BACKGROUND

[0002] The existing historical building three-dimensional supervision modeling system obtains data such as temperature changes and energy consumption curves in the building operation process, establishes a dynamic response model of the building state, or uses multi-robot cooperation to collect point cloud data of the building structure, and completes three-dimensional modeling through error correction, path optimization and point cloud splicing.

[0003] For example, the modeling method and system for building demand response dynamic process disclosed in Chinese patent application CN108122067A include: determining the power consumption curve of the building electrical equipment according to the temperature change at each time; comparing the power consumption curve with the pre-defined reference power curve to generate a demand response dynamic process simulation curve; determining the fitting function of each stage in the demand response dynamic process simulation curve; and establishing a demand response dynamic model based on the fitting function of each stage.

[0004] For example, the three-dimensional modeling design method and system based on multi-machine cooperation of building robots disclosed in Chinese patent application CN120297628A include: task allocation information including the tasks and paths allocated to each robot; an error-corrected task execution scheme based on real-time feedback and task allocation information; global analysis of the execution errors of all robots for the corrected task execution scheme; construction of a task-robot cooperation matrix based on the optimized task execution scheme, and task scheduling of the task-robot cooperation matrix using a cooperation scheduling algorithm; and merging of the adjusted point cloud data and the optimized three-dimensional model into a complete three-dimensional model.

[0005] However, in the process of implementing the technical scheme of the present application, the present application has found that the above-mentioned technology at least has the following technical problems:

[0006] In the prior art, there are a large number of similar positions in historical buildings, and when taking pictures of historical buildings and extracting features from the pictures, these large numbers of repeated areas will affect the picture stitching accuracy and speed, affecting the speed of historical building modeling, and thus there is a problem of low three-dimensional modeling accuracy and low efficiency due to inaccurate historical building feature vector acquisition. SUMMARY

[0007] In order to solve the problem of low precision and low efficiency of three-dimensional modeling caused by inaccurate feature vector acquisition of historical buildings in the prior art, the embodiment of the present application provides a three-dimensional supervision modeling method for historical buildings based on multi-source geographic data fusion.

[0008] The three-dimensional supervision modeling method for historical buildings based on multi-source geographic data fusion is provided, and the method comprises the following steps: S1, acquiring multi-source geographic data, wherein the multi-source geographic data comprises historical building images and 3D point cloud data, and the historical building images are regionally divided to obtain each image region of the historical building images; S2, extracting features of each image region of the historical building images to obtain feature vectors of each image region, and performing secondary regional division by mapping the feature vectors to the historical building images; S3, combining any two historical building images to obtain an image group, extracting feature vectors of each specified secondary division region of the image group, and performing feature vector similarity analysis to obtain a comprehensive feature vector similarity of the image group; S4, performing adjacent determination based on the comprehensive feature vector similarity of the image group and the 3D point cloud data to obtain a 3D adjacent determination result, if the 3D adjacent determination result is adjacent, performing image stitching on the image group, otherwise, analyzing a temporary storage location of the image group, and continuously performing adjacent determination on the image group until a complete image stitching relationship is obtained; S5, inputting the complete image stitching relationship and the historical building images into a preset 3D modeling platform to perform 3D modeling, thereby obtaining a three-dimensional model of the historical building.

[0009] Further comprising adjusting the execution parameters of the image acquisition device based on the influence of the external environment of the historical building, thereby obtaining each historical building image; obtaining the shooting influence parameters of each to-be-shot region of the historical building, wherein the shooting influence parameters comprise light intensity, building surface reflectivity, light incidence angle and atmospheric visibility; obtaining a preset shooting reference set in the database, and comparing and analyzing the shooting influence parameters of each to-be-shot region of the historical building to obtain a comparison analysis result; introducing a corresponding weighting factor based on the comparison analysis result for coupling processing to obtain a historical building shooting influence complexity value of each to-be-shot region of the historical building; matching the historical building shooting influence complexity value of each to-be-shot region of the historical building with the database to obtain a shooting first influence factor of each to-be-shot region of the historical building; adjusting the photosensitivity and exposure time of the image acquisition device based on the shooting first influence factor of each to-be-shot region of the historical building, and performing multi-source shooting on each to-be-shot region of the historical building after adjustment to obtain each historical building image; the shooting reference set comprises a light intensity standard value, a building surface reflectivity standard value, a light incidence angle standard value and an atmospheric visibility standard value.

[0010] The technical scheme provided by the embodiment of the present application has at least the following beneficial effects:

[0011] 1. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion provided by the application realizes reasonable regional division of historical building images, improves the fineness of feature extraction, and realizes accurate construction of a historical building three-dimensional model, thereby effectively solving the problems of poor three-dimensional modeling accuracy and low efficiency caused by inaccurate historical building feature vector acquisition in the prior art.

[0012] 2. The application realizes determination of the secondary regional division specification by mapping the feature vectors of each image region to the historical building image and matching based on the mean vector distance and the database, thereby ensuring the consistency of the number of image regions.

[0013] 3. The application realizes accurate evaluation of the comprehensive feature vector similarity of the image group by sorting the specified secondary division regions of the image group in a preset order and performing feature vector similarity fluctuation analysis, and simultaneously coupling the preset threshold and the weighting factor in the database, thereby improving the accuracy of image stitching.

[0014] 4. The application realizes reasonable management of the image group by performing adjacent determination and classified storage of the image group based on the multi-level determination method of the comprehensive feature vector similarity and the 3D point cloud depth difference, thereby improving the stitching efficiency and model integrity of the three-dimensional model construction process. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0016] Figure 1 The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion provided by the embodiment of the application is shown in the flowchart.

[0017] Figure 2 The step flowchart of the historical building three-dimensional supervision modeling method based on multi-source geographic data fusion provided by the embodiment of the application is shown in the flowchart.

[0018] Figure 3 The 3D adjacent determination result determination flowchart of the historical building three-dimensional supervision modeling method based on multi-source geographic data fusion provided by the embodiment of the application is shown in the flowchart.

[0019] Figure 4The building image splicing schematic diagram provided by the historical building three-dimensional supervision modeling method based on multi-source geographic data fusion for the embodiment of the application;

[0020] Figure 5 The regional mapping group example diagram of the historical building three-dimensional supervision modeling method based on multi-source geographic data fusion for the embodiment of the application is provided.

[0021] Figure 6 The image uploading interface example diagram of the historical building three-dimensional supervision modeling method based on multi-source geographic data fusion for the embodiment of the application is provided.

[0022] Figure 7 The splicing effect visual display example diagram of the historical building three-dimensional supervision modeling method based on multi-source geographic data fusion for the embodiment of the application is provided. DETAILED DESCRIPTION

[0023] The technical solutions in the application will be described below with reference to the drawings.

[0024] In the embodiments of the application, the words such as "example", "for example" are used to represent an example, illustration or description. Any embodiment or design scheme described as "example" in the application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. In fact, the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0025] In the embodiments of the application, "image" and "picture" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding" and "corresponding" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0026] In the embodiments of the application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1. When the distinction is not emphasized, the meanings expressed are consistent.

[0027] In order to make the technical problems, technical solutions and advantages to be solved by the application more clear, the following will be described in detail with reference to the drawings and specific embodiments.

[0028] As Figure 1As shown, a macro flowchart of a historical building three-dimensional supervision modeling method based on multi-source geographic data fusion provided by the embodiment of the application is shown, and the method comprises the following steps: S1, acquiring multi-source geographic data, the multi-source geographic data comprising historical building images and 3D point cloud data, and performing regional division on the historical building images to obtain each image region of the historical building images; S2, performing feature extraction on each image region of the historical building images to obtain a feature vector of each image region, and mapping to the historical building images for secondary regional division to obtain each secondary division region; S3, combining any two historical building images to obtain an image group, extracting a feature vector of each specified secondary division region of the image group, and performing feature vector similarity analysis to obtain a comprehensive feature vector similarity of the image group; S4, performing adjacent determination based on the comprehensive feature vector similarity of the image group and the 3D point cloud data to obtain a 3D adjacent determination result, if the 3D adjacent determination result is adjacent, performing image stitching on the image group, otherwise, analyzing to obtain a temporary storage location of the image group, and continuously performing image group adjacent determination until a complete image stitching relationship is obtained; and S5, inputting the complete image stitching relationship and the historical building images into a preset 3D modeling platform for 3D modeling, thereby obtaining a historical building three-dimensional model.

[0029] In the embodiment, as shown in Figure 2 Figure 2 A step flowchart of a historical building three-dimensional supervision modeling method based on multi-source geographic data fusion provided by the embodiment of the application is shown, multi-source geographic data is acquired, and regional division is performed on historical building images, feature vectors of each image region are extracted, and secondary division is performed to obtain secondary division regions; subsequently, the historical building images are combined two by two to form an image group, and a comprehensive feature vector similarity of the image group is calculated; 3D adjacent determination is performed according to the similarity result, if the result is adjacent, image stitching is performed, if it is not adjacent, it is stored in a temporary storage location; thereafter, the image stitching is continuously improved through image group adjacent determination to establish a complete image stitching relationship, and finally a historical building three-dimensional model is generated.

[0030] The multi-source geographic data is acquired, and specifically, historical building images can be acquired through a multi-source image acquisition device (such as an industrial camera), and 3D point cloud data can be obtained through ground laser radar scanning. Feature extraction is performed on each image region of the historical building images to obtain a feature vector of each image region, which can be obtained by processing with an image preprocessing tool such as MATLAB.

[0031] It should be noted that the absolute difference processing and the difference processing are both difference processing, and the results obtained are both positive numbers.

[0032] ​The scheme realizes the tilt angle correction and depth change analysis of the historical building image by multi-source geographic data fusion and collaborative processing of the historical building image and 3D point cloud data, thereby accurately dividing the image region and improving the accuracy of the region division; the feature vector mean distance and database matching method are used to complete the secondary region division, thereby improving the correlation of feature extraction and region division; the accurate adjacent determination of the image group is realized through the feature vector similarity fluctuation analysis and weighting coupling combined with the 3D point cloud depth determination, thereby effectively avoiding misjudgment; different temporary zone management is set for the non-adjacent image group, thereby improving the organization efficiency of image stitching; the complete image stitching relationship is input into the 3D modeling platform, thereby ensuring the integrity and accuracy of the 3D model stitching.

[0033] Further, the image regions of the historical building image are obtained, and the specific method is as follows: the tilt photography angle of the historical building image is obtained; the 3D point cloud data of the historical building image is corrected in terms of the tilt photography angle of the historical building image to obtain the corrected 3D point cloud data of the historical building image, and the corrected 3D point cloud data of the historical building image is marked as the tilt correction 3D point cloud data of the historical building image; the point cloud depth change amplitude of the historical building image is extracted based on the tilt correction 3D point cloud data of the historical building image, and the first region division specification of the historical building image is obtained through processing based on the point cloud depth change amplitude; the historical building image is divided into regions based on the first region division specification to obtain the image regions of the historical building image.

[0034] In the embodiment, the tilt photography angle of the historical building image can be obtained by querying the shooting log of the historical building image.

[0035] The point cloud depth change amplitude of the historical building image is extracted based on the tilt correction 3D point cloud data of the historical building image, and the specific method is as follows: the point cloud depth in the tilt correction 3D point cloud data of the historical building image is obtained, and the maximum difference value processing (the maximum point cloud depth value minus the minimum point cloud depth value) is performed based on the point cloud depth, so that the point cloud depth change amplitude is obtained.

[0036] The first region division specification of the historical building image is obtained through processing based on the point cloud depth change amplitude, and the specific method is as follows: the preset point cloud depth change amplitude reference value, the reference region division specification, the single gradient division specification adjustment coefficient and the point cloud depth adjustment gradient in the database are obtained; the point cloud depth reference difference value is obtained through difference processing of the point cloud depth change amplitude reference value and the point cloud depth change amplitude, the point cloud depth multiple is obtained based on the point cloud depth reference difference value divided by the point cloud depth adjustment gradient, the division specification adjustment comprehensive coefficient is obtained based on the multiplication of the point cloud depth multiple and the single gradient division specification adjustment coefficient, and the first region division specification of the historical building image is obtained based on the multiplication of the division specification adjustment comprehensive coefficient and the reference region division specification and then adding the reference region division specification.

[0037] By acquiring the tilt photography angle of the historical building image and correcting the 3D point cloud data thereof according to the tilt angle, the spatial deviation caused by the photography angle can be eliminated, and the spatial matching accuracy between the image and the point cloud data can be improved. The depth change amplitude is extracted based on the corrected 3D point cloud data, which can reflect the complexity of the surface structure of the historical building, so that a more adaptive first region division specification can be formulated accordingly. By dividing the image according to the division specification, adaptive adjustment of the region division is realized: for a building region with complex structure and rich details, more sub-regions are divided to facilitate more detailed feature extraction; for a region with simple structure and small changes, fewer regions are divided to improve the feature extraction efficiency. Thus, the region division has both sufficient detail expression capability and overall processing efficiency, thereby improving the accuracy and speed of subsequent feature extraction and modeling processing.

[0038] Further, each secondary division region is obtained by: performing feature extraction on each image region to obtain a feature vector of each image region; mapping the feature vector of each image region to the historical building image respectively to obtain a feature vector mapping image; analyzing the feature vector of each image region to obtain a mean vector distance of the feature vector; matching the mean vector distance of the feature vector with the database to obtain a second region division specification; and performing secondary region division on the feature vector mapping image based on the second region division specification, thereby obtaining each secondary division region of the historical building image and a feature vector of each secondary division region.

[0039] In the present embodiment, it should be noted that the feature vector of each image region is mapped to the historical building image respectively to obtain a feature vector mapping image, and the specific method is: based on the feature vector extracted from each image region, the feature information is visualized and superimposed on the original historical building image by combining the corresponding image coordinate position, to form a feature vector mapping image. The specific implementation can use the OpenCV library to superimpose and display the feature values in the form of a heat map, for example, using OpenCV to perform color rendering on each image region, and displaying the feature vector intensity in the corresponding position of the original image in the form of color depth.

[0040] The feature extraction is performed on each image region to obtain a feature vector of each image region, and the specific method is: using the DoG (Difference of Gaussian) operator in the SIFT (Scale-Invariant Feature Transform) algorithm to perform feature extraction to obtain the feature vector.

[0041] The second region partitioning specification is obtained by matching the mean vector distance of the feature vectors with the database. Specifically, the gradient of the mean vector distances preset in the database is obtained and matched with the mean vector distances of the feature vectors. If the mean vector distance of the feature vectors falls within a certain mean vector distance gradient, then the region partitioning specification corresponding to that gradient is used as the second region partitioning specification. It should be noted that the mean vector distance gradient refers to the absolute difference between the minimum values ​​of any two adjacent extraction intervals.

[0042] By extracting features from the initially segmented image regions and generating corresponding feature vectors, these feature vectors are then mapped onto the image to form a feature vector mapping image. Further analysis based on the mean vector distance of the feature vectors effectively reflects the differences in feature distribution among the regions. This is then combined with database matching to determine the second region segmentation specifications, and the mapped image is further segmented into secondary regions accordingly, resulting in more consistent region segmentation results, facilitating subsequent region matching.

[0043] Furthermore, the comprehensive feature vector similarity of the image group is obtained through the following method: Any two historical building images are combined to form an image group; feature vectors of each specified secondary partition region in the image group are extracted, with the secondary partition regions defined as the edge regions of the historical building images; the specified secondary partition regions are sorted according to a preset order (image arrangement order) to obtain a secondary partition region order; based on this order, the specified secondary partition regions of the image group are mapped and matched to obtain each region mapping group; feature vector similarity analysis is performed on each region mapping group to obtain the feature vector similarity of each region mapping group; a preset feature vector similarity threshold is obtained from the database and compared with the feature vector similarity of each region mapping group; if the feature vector similarity of a certain region mapping group is below the feature vector similarity threshold, then the feature vector similarity of that region mapping group is matched with the database to obtain the comprehensive influence reference coefficient of that region mapping group; if the feature vector similarity of a certain region mapping group is below the threshold, then the feature vector similarity of that region mapping group is matched with the database to obtain the comprehensive influence reference coefficient of that region mapping group. If the eigenvector similarity is greater than the eigenvector similarity threshold, a preset historical comprehensive influence reference coefficient is obtained from the database as the comprehensive influence reference coefficient for that region mapping group, thus obtaining the comprehensive influence reference coefficient for each region mapping group. Based on the eigenvector similarity of each region mapping group, eigenvector similarity fluctuation analysis is performed to obtain eigenvector similarity fluctuation parameters, including the eigenvector similarity range, eigenvector similarity standard deviation, and eigenvector similarity average. A preset eigenvector similarity fluctuation benchmark set is obtained from the database and compared with the eigenvector similarity fluctuation parameters to obtain the comparison analysis results. Based on the comparison analysis results, corresponding weighting factors are introduced for coupling processing, and multiplicative coupling processing is performed with the comprehensive influence reference coefficient to obtain the comprehensive eigenvector similarity of the image group. The eigenvector similarity fluctuation benchmark set includes the eigenvector similarity range benchmark value, the eigenvector similarity standard deviation benchmark value, and the eigenvector similarity average benchmark value.

[0044] In this embodiment, as Figure 5 As shown, Figure 5 This is an example diagram of regional mapping groups for a three-dimensional monitoring modeling method for historical buildings based on multi-source geographic data fusion provided in this application embodiment. F1 and F4 represent regional mapping group 1, F2 and F5 represent regional mapping group 2, and F3 and F6 represent regional mapping group 3. F1, F2, F3, F4, F5, and F6 are all designated secondary subdivision regions.

[0045] The comprehensive influence reference coefficient of the region mapping group is analyzed because there are many repetitive areas in historical buildings. For example, a red wall in a historical building may have a high degree of similarity in texture and other features. When analyzing this red wall area, each feature point has a very important influence on the position analysis of this red wall. If this feature is ignored, other areas may be mistakenly identified as the correct areas, resulting in a significant deviation. Therefore, each area has a significant impact on the overall effect.

[0046] The method for obtaining the comprehensive feature vector similarity of an image group is as follows:

[0047]

[0048] In the formula, ZH represents the comprehensive feature vector similarity of the image group, and i represents the number of the region mapping group, i=1,2... , X represents the total number of region mapping groups. i ε1 represents the comprehensive influence reference coefficient of the mapping group of the i-th region, JC represents the feature vector similarity range, HC represents the feature vector similarity range benchmark value, BC represents the feature vector similarity standard deviation, HB represents the feature vector similarity standard deviation benchmark value, PC represents the feature vector similarity mean, HP represents the feature vector similarity mean benchmark value, ε1 represents the feature vector similarity range weighting factor, ε2 represents the feature vector similarity standard deviation weighting factor, and ε3 represents the feature vector similarity mean weighting factor.

[0049] If the feature vector similarity of a certain region mapping group is below the feature vector similarity threshold, then the feature vector similarity of the region mapping group is matched with the database to obtain the comprehensive influence reference coefficient of the region mapping group. The specific method is as follows: obtain the preset feature similarity benchmark value, comprehensive influence benchmark coefficient, single gradient comprehensive influence coefficient increase, and feature similarity gradient from the database; perform difference processing based on feature vector similarity and feature similarity benchmark value to obtain feature similarity benchmark difference; divide the feature similarity benchmark difference by the feature similarity gradient to obtain the feature similarity gradient multiple; multiply the feature similarity gradient multiple by the single gradient comprehensive influence coefficient increase to obtain the comprehensive gradient increase; and add the comprehensive influence benchmark coefficient to the comprehensive gradient increase to obtain the comprehensive influence reference coefficient of the region mapping group.

[0050] The eigenvector similarity range represents the range of eigenvector similarity among the mapping groups of regions, the eigenvector similarity standard deviation represents the standard deviation of eigenvector similarity among the mapping groups of regions, and the eigenvector similarity mean represents the average eigenvector similarity among the mapping groups of regions.

[0051] The weighting factors for the eigenvector similarity range, standard deviation, and average value can be obtained from a database. For example, the weighting factor for the eigenvector similarity range can be obtained by analyzing the historical eigenvector similarity range set stored in the database. The difference between each historical eigenvector similarity range and the eigenvector similarity range in the historical eigenvector similarity range set is calculated to obtain the difference value for each historical eigenvector similarity range. A preset threshold range for the difference value of the eigenvector similarity range in the database is obtained and compared with the difference values ​​of each historical eigenvector similarity range. If a certain historical eigenvector similarity range difference falls within the threshold range, the corresponding historical eigenvector similarity range is obtained and marked as the historical control eigenvector similarity range, thus obtaining the similarity range of each historical control eigenvector. The standard deviation of each historical control eigenvector similarity range is then calculated after removing the maximum and minimum values. The database is used to obtain the preset historical reference feature vector similarity range weighting factor, the historical control feature vector similarity range standard deviation benchmark, the historical feature vector similarity range-control standard deviation difference gradient, and the single-level adjustment amount of the feature vector similarity range weighting factor. The difference between the historical control feature vector similarity range standard deviation and the historical control feature vector similarity range standard deviation benchmark is processed to obtain the historical control feature vector similarity range standard deviation difference value. The gradient of the historical control feature vector similarity range standard deviation difference value and the historical control feature vector similarity range standard deviation difference is analyzed using a multiple analysis (dividing the historical control feature vector similarity range standard deviation difference by the historical control feature vector similarity range standard deviation difference gradient) to obtain the historical feature vector similarity range-control standard deviation difference gradient multiple. The gradient factor of the difference between the standard deviation and the feature vector similarity range is multiplied by the single-level adjustment of the feature vector similarity range weighting factor to obtain the comprehensive adjustment of the feature vector similarity range weighting factor. The standard deviation of the historical reference feature vector similarity range is compared with the benchmark value. If the standard deviation of the historical reference feature vector similarity range is greater than the benchmark value, the historical reference feature vector similarity range weighting factor is added to the comprehensive adjustment of the feature vector similarity range weighting factor to obtain the feature vector similarity range weighting factor. Otherwise, the historical reference feature vector similarity range weighting factor is subtracted from the comprehensive adjustment of the feature vector similarity range weighting factor to obtain the feature vector similarity range weighting factor.

[0052] The single-level adjustment of the eigenvector similarity range weighting factor refers to the adjustment amount required for each increase in the gradient of the historical eigenvector similarity range versus the standard deviation difference. The gradient of the historical eigenvector similarity range versus the standard deviation difference refers to the difference between the minimum values ​​of two adjacent gradient intervals. Other weighting factors, such as the eigenvector similarity standard deviation weighting factor and the eigenvector similarity average weighting factor, are obtained in the same way as the eigenvector similarity range weighting factor.

[0053] By selecting edge-divided regions in historical building images for feature vector extraction and performing matching analysis on these regions in a preset order, the structural continuity and information alignment of the image stitching region can be ensured. By quantitatively analyzing the fluctuation of feature vector similarity (such as range, standard deviation, and mean) and introducing a fluctuation benchmark set for comparison and weighting, the interference of outliers in a single region on the overall result can be effectively suppressed, and the stability of the matching result can be improved.

[0054] Furthermore, the 3D adjacency determination result is obtained. The specific method is as follows: obtain the preset feature vector similarity threshold in the database and compare it with the comprehensive feature vector similarity of the image group to obtain the proposed adjacency determination result. If the comprehensive feature vector similarity of the image group is greater than the feature vector similarity threshold, the proposed adjacency determination result of the image group is adjacent; otherwise, the proposed adjacency determination result of the image group is not adjacent. If the proposed adjacency determination result of an image group is adjacent, then 3D adjacency determination is performed. If the proposed adjacency determination result is not adjacent, then the image group is stored in the mutual exclusion temporary storage area for temporary storage processing.

[0055] In this embodiment, as Figure 3 As shown, Figure 3 This document presents a flowchart illustrating the 3D adjacency determination process for a historical building 3D monitoring modeling method based on multi-source geographic data fusion, as provided in this embodiment. First, the comprehensive feature vector similarity of the image group is obtained, and it is determined whether this similarity exceeds a feature vector similarity threshold. If it does not exceed the threshold, the adjacency determination result is considered non-adjacent; if it does exceed the threshold, the adjacency determination result is considered adjacent, and 3D adjacency determination is then performed. Next, 3D point cloud data of the image group is obtained, and the average depth difference value of the 3D point cloud is calculated. Based on this difference value, the depth of the 3D point cloud data is determined. If the depth determination result indicates adjacency, the 3D adjacency determination result is considered 3D adjacent; otherwise, it is considered 3D non-adjacent, and the final 3D adjacency determination result is output.

[0056] By setting a similarity threshold for feature vectors, the initial screening of image groups is achieved. Only image groups with high similarity are subjected to 3D adjacency determination, which reduces unnecessary 3D computation and improves processing efficiency. At the same time, non-adjacent image groups are temporarily stored, which helps to optimize subsequent scheduling and stitching relationships and ensures the continuity and stability of the modeling process.

[0057] Furthermore, if the proposed adjacency determination result for a certain image group is adjacent, then 3D adjacency determination is performed. The specific process is as follows: acquire 3D point cloud data of a specified secondary region of the image group, and perform 3D point cloud data depth analysis to obtain the average depth difference value of the 3D point cloud of the image group. Thus, the 3D point cloud data depth determination result of the image group is obtained. If the 3D point cloud data depth determination result is adjacent, then the 3D adjacency determination result of the image group is 3D adjacent. If the 3D point cloud data depth determination result is not adjacent, then the 3D adjacency determination result of the image group is 3D non-adjacent. If the proposed adjacency determination result is not adjacent, then the image group is stored in the mutual exclusion temporary storage area for temporary storage processing.

[0058] In this embodiment, the average depth difference value of the 3D point cloud of the image group is obtained by obtaining the average depth of the 3D point cloud of each historical building image in the image group and performing absolute difference processing to obtain the average depth difference value of the 3D point cloud.

[0059] By performing depth analysis on 3D point cloud data of a specified region in an image group, the average depth difference value of the 3D point clouds can be obtained, which can effectively determine the spatial adjacency relationship between two image regions. If the depth difference is small, it indicates that the two regions are close in 3D space and are determined to be 3D adjacent, which helps with subsequent stitching modeling. If the depth difference is large, it indicates that the spatial positions are discontinuous and are determined to be 3D non-adjacent, avoiding incorrect stitching. Different adjacency determination results correspond to different storage strategies. Non-adjacent image groups are stored in a mutually exclusive temporary storage area, which facilitates re-determination based on new matching conditions, effectively improving the accuracy and efficiency of modeling.

[0060] Furthermore, the 3D point cloud data depth determination result is obtained as follows: The minimum 3D point cloud depth of a specified secondary region in the historical building images within the image group is obtained. Using the horizontal plane containing the minimum 3D point cloud depth as a reference plane, the relative 3D point cloud depth of each random sample point in the specified secondary region is obtained. The mean depth of the 3D point cloud in the specified secondary region is then calculated. The difference in the average 3D point cloud depth between two historical building images in the image group is analyzed to obtain the difference value of the average 3D point cloud depth of the image group. A preset first threshold for 3D point cloud depth difference in the database is obtained and compared with the difference value of the average 3D point cloud depth of the image group to obtain the 3D point cloud data depth determination result of the image group. If the difference value of the average 3D point cloud depth of the image group is above the first threshold, the 3D point cloud data depth determination result of the image group is dissimilar; otherwise, the 3D point cloud data depth determination result of the image group is similar.

[0061] In this embodiment, it should be noted that the relative depth of the 3D point cloud of each random sample point in the specified secondary region is obtained by obtaining the minimum value of the 3D point cloud depth of the specified secondary region of the historical building image in the image group, and then performing difference processing between the depth of each 3D point cloud and the point cloud depth of the reference plane to obtain the relative depth of the 3D point cloud of each random sample point in the specified secondary region.

[0062] The difference in the average depth of the 3D point clouds of two historical building images in the image group is analyzed to obtain the difference value of the average depth of the 3D point clouds of the image group. Specifically, the difference value is obtained by performing difference processing on the average depth of the 3D point clouds of the two historical building images in the image group.

[0063] By setting a reference plane based on the minimum depth and calculating the relative depth of each random sample point, the average depth difference value of the image region can be obtained, which can more accurately assess the similarity of the image group in spatial structure. Then, by combining the first threshold of depth difference for judgment, an objective judgment on the three-dimensional adjacency relationship of the image group can be achieved, thereby improving the accuracy and robustness of spatial matching in three-dimensional modeling.

[0064] Furthermore, the temporary storage area for image groups is specifically divided as follows: based on the 3D adjacency determination results, if the 3D adjacency determination results for image groups are not adjacent, then the preset second threshold for 3D point cloud depth difference in the database is obtained and compared with the average depth difference value of the 3D point cloud in the image group. If the average depth difference value of the 3D point cloud in the image group is above the second threshold for 3D point cloud depth difference, then the image group is stored in the ordinary image temporary storage area. If the average depth difference value of the 3D point cloud is less than the second threshold for 3D point cloud depth difference, then the image group corresponding to the average depth difference value of the 3D point cloud is stored in the proposed related image temporary storage area. The temporary storage area includes the ordinary image temporary storage area, the proposed related image temporary storage area, and the mutually exclusive temporary storage area.

[0065] In this embodiment, by introducing a second depth difference threshold, 3D non-adjacent image groups are further subdivided and stored in a general image temporary storage area or a designated related image temporary storage area, which effectively improves the hierarchy and flexibility of image management.

[0066] Image groups are divided and stored in different areas (ordinary image temporary storage area, proposed related image temporary storage area, and mutually exclusive temporary storage area) according to the average depth difference value of 3D point clouds. This allows for hierarchical classification and management based on the spatial relationship and similarity difference between image groups. This enables the rapid location and retrieval of data with higher matching degree when supplementing, replacing, or re-stitching image data during subsequent modeling processes, thereby improving modeling efficiency and accuracy.

[0067] Furthermore, the adjacency determination of image groups continues. The specific method is as follows: obtain image groups that have not been determined to be stitched together, and analyze them. Obtain any building image in the image group that has been stitched together with other building images, and mark the building image as a stitched building image. Mark the remaining building images in each image group that has not been stitched together with the stitched building image as remaining related building images. Thus, each remaining related building image is obtained. Mark any stitched building image in the image group that has been stitched together with the stitched building image as an adjacent stitched building image. Perform image group adjacency determination between each remaining related building image and the adjacent stitched building image. If the image group adjacency determination result is adjacent, then image stitching is performed. If they are not adjacent, random building image adjacency determination is performed until a complete image stitching relationship is obtained.

[0068] In this embodiment, as Figure 4 As shown, taking region A as an example, if region A is stitched together with regions B, C, D, and E, and the image group formed when region A is stitched together with region E is determined to be adjacent, while the image group formed when region A is stitched together with regions B, C, and D is determined to be non-adjacent, then region A is the stitched building image, regions B, C, and D are the remaining related building images, and region E is the adjacent stitched building image.

[0069] Each remaining relevant building image is compared with adjacent stitched building images using an image group adjacency check. If the image group adjacency check result is adjacent, image stitching is performed. If not adjacent, random building image adjacency checks are performed. In the random building image adjacency check, the remaining relevant building images stored in the designated relevant image temporary storage area are analyzed first, followed by those stored in the ordinary image temporary storage area, and finally those stored in the mutually exclusive temporary storage area. This stitching order improves stitching efficiency and accuracy. Prioritizing the analysis of images in the designated relevant image temporary storage area, as they have higher 3D point cloud similarity to stitched images and a higher matching success rate, leads to the analysis of images in the ordinary image temporary storage area, which, although less similar, still has a certain probability of matching. Finally, images in the mutually exclusive temporary storage area are analyzed, only as alternatives when the first two categories cannot be stitched, avoiding interference from low-relevance images with stitching accuracy, thereby reducing invalid calculations and improving stitching success rate and overall modeling quality.

[0070] By prioritizing the adjacency determination of image groups based on "already stitched building images" and their associated "remaining related building images," an outward expansion stitching method is achieved from the existing stitching results. This avoids full traversal and unordered attempts of all image combinations, effectively reducing the computational load of image stitching. At the same time, by utilizing known relationships between images, the hit rate and processing efficiency of image stitching are improved, ensuring that the stitching path has higher spatial continuity and structural logic, rather than blind random stitching, thereby improving the overall modeling accuracy and speed.

[0071] Furthermore, the process includes adjusting the execution parameters of the image acquisition device based on the external environmental impact of the historical buildings to obtain images of each historical building; acquiring shooting impact parameters, including light intensity, ambient temperature, ambient color temperature, and atmospheric visibility; acquiring a preset shooting standard set from the database and comparing it with the shooting impact parameters to obtain the comparison analysis results; introducing corresponding weighting factors based on the comparison analysis results for coupling processing to obtain the shooting impact complexity value of the historical buildings; matching the shooting impact complexity value of the historical buildings with the database to obtain the first shooting impact factor; adjusting the sensitivity and exposure time of the image acquisition device based on the first shooting impact factor; and performing multi-source shooting of each area to be shot in the historical buildings after adjustment to obtain images of each historical building; the shooting standard set includes standard values ​​for light intensity, ambient temperature, ambient color temperature, and atmospheric visibility.

[0072] In this embodiment, the light intensity can be obtained by adjusting the light sensor built into the image acquisition device. The ambient temperature can be detected by a temperature sensor, and the ambient color temperature can be detected by a color temperature meter. Atmospheric visibility can be measured using a visibility meter.

[0073] It should be noted that the specific method for obtaining the first impact factor of photography is as follows: obtain the preset reference gradients of the impact complexity of photography on each historical building in the database, and the first impact reference factor of photography corresponding to each historical building's impact complexity reference gradient, and compare them with the historical building's impact complexity reference value. If the historical building's impact complexity reference value is within a certain historical building's impact complexity reference gradient, then the first impact reference factor of photography corresponding to that historical building's impact complexity reference gradient is taken as the first impact factor of photography. The historical building's impact complexity reference gradient corresponds to a range of historical building's impact complexity values. For example, if the range of the historical building's impact complexity value corresponding to the first historical building's impact complexity reference gradient is (0, 20], then the historical building's impact complexity values ​​within that range are all the first historical building's impact complexity reference gradients.

[0074] The method for adjusting the sensitivity and exposure time of the image acquisition device based on the first impact factor of the shooting is as follows: the light intensity is compared with the standard value of light intensity. If the light intensity is above the standard value of light intensity, the sensitivity and exposure time of the device are reduced. If the light intensity is below the standard value of light intensity, the sensitivity and exposure time of the device are increased.

[0075] To reduce the sensitivity and exposure time of your device, the specific methods are as follows: In the formula, ISO1 represents the reduced ISO sensitivity, ISO0 represents the original ISO sensitivity, and SY a This indicates the first impact factor of the film. In the formula, EVH1 represents the reduced exposure time, EVH0 represents the original exposure time, and SY... a This indicates the first impact factor of the film.

[0076] To increase the sensitivity and exposure time of the equipment, the specific methods are as follows: In the formula, ISO1 represents the increased ISO sensitivity, ISO0 represents the original ISO sensitivity, and SY a This indicates the first impact factor of the film. In the formula, EVH1 represents the exposure time after the increase adjustment, EVH0 represents the exposure time before the increase adjustment, and SY a This indicates the first impact factor of the film.

[0077] By analyzing the shooting influence parameters of various areas of historical buildings to be photographed, including light intensity, ambient temperature, ambient color temperature, and atmospheric visibility, the complexity value of the shooting influence of historical buildings is obtained. This considers the interrelationships between these parameters. For example, light intensity directly determines the brightness and detail clarity of the building surface during shooting; stronger light can enhance detail, but excessive light may lead to overexposure. Ambient temperature affects the stability of the light source and the working state of the camera sensor; higher temperatures may cause increased equipment noise, affecting image quality. Ambient color temperature determines the tonal characteristics of the light source; different color temperatures can lead to color deviations in the image, affecting the accuracy of color reproduction, and changes in color temperature will affect the perceived effect of light intensity. Atmospheric visibility reflects the concentration of suspended particulate matter and water vapor in the air, which can significantly affect the propagation path and scattering degree of light, thus affecting the actual effect of light intensity and color clarity. In addition, a decrease in atmospheric visibility is usually accompanied by changes in humidity and temperature; these changes work together to alter the overall optical environment of the shooting area.

[0078] The method for obtaining the impact complexity value of photographing historical buildings is as follows:

[0079] ;

[0080] In the formula, FZ represents the complexity value of the impact of photographing historical buildings, GQ represents light intensity, ZQ represents the standard value of light intensity, GW represents ambient temperature, ZW represents the standard value of ambient temperature, GS represents ambient color temperature, ZS represents the standard value of ambient color temperature, GN represents atmospheric visibility, and ZN represents the standard value of atmospheric visibility.

[0081] μ1 represents the light intensity weighting factor, μ2 represents the ambient temperature weighting factor, μ3 represents the ambient color temperature weighting factor, and μ4 represents the atmospheric visibility weighting factor.

[0082] Light intensity weighting factors, ambient temperature weighting factors, ambient color temperature weighting factors, and atmospheric visibility weighting factors can be obtained from a database. For example, the light intensity weighting factor can be obtained by analyzing the historical light intensity set stored in the database. The difference between each historical light intensity in the historical light intensity set and the actual light intensity is processed to obtain the historical light intensity difference values. A preset light intensity difference threshold range in the database is obtained and compared with each historical light intensity difference value. If a historical light intensity difference value falls within the threshold range, the corresponding historical light intensity value is obtained and marked as the historical control light intensity, thus obtaining each historical control light intensity. The extreme values ​​(maximum and minimum values) of each historical control light intensity value are then removed, and the standard deviation is taken to obtain the standard deviation of the historical control light intensity. The database is used to obtain preset historical reference light intensity weighting factors, historical control light intensity standard deviation benchmarks, historical light intensity control standard deviation difference gradients, and single-level adjustment amounts of the light intensity weighting factors. The difference between the historical control light intensity standard deviation and the historical control light intensity benchmark is calculated to obtain the historical control light intensity standard deviation difference value. A fold analysis is then performed on the historical control light intensity standard deviation difference value and the historical light intensity control standard deviation difference gradient (dividing the historical control light intensity standard deviation difference value by the historical light intensity control standard deviation difference gradient value) to obtain the historical light intensity control standard deviation difference gradient fold. The gradient multiple of the difference in the standard deviation of historical light intensity is multiplied by the single-level adjustment of the light intensity weighting factor to obtain the comprehensive adjustment of the light intensity weighting factor. Based on the comparison between the standard deviation of historical light intensity and the benchmark value of the standard deviation of historical light intensity, if the standard deviation of historical light intensity is greater than the benchmark value, the historical reference light intensity weighting factor is added to the comprehensive adjustment of the light intensity weighting factor to obtain the light intensity weighting factor; otherwise, the historical reference light intensity weighting factor is subtracted from the comprehensive adjustment of the light intensity weighting factor to obtain the light intensity weighting factor.

[0083] The single-level adjustment of the light intensity weighting factor refers to the adjustment amount required for each increase in the historical light intensity comparison standard deviation gradient. The historical light intensity comparison standard deviation gradient refers to the difference between the minimum values ​​of two adjacent gradient intervals. Other weighting factors, such as ambient temperature weighting factor, ambient color temperature weighting factor, and atmospheric visibility weighting factor, are obtained in the same way as the light intensity weighting factor.

[0084] like Figure 6 , Figure 7 As shown, Figure 6 This is an example image of the image upload interface for a three-dimensional monitoring and modeling method for historical buildings based on multi-source geographic data fusion, provided in an embodiment of this application. Figure 7 This image is a visualization example of the stitching effect of the 3D monitoring and modeling method for historical buildings based on multi-source geographic data fusion provided in this application embodiment. When 3D modeling of historical buildings is required, simply upload the images of the historical buildings to the surveying and mapping platform. The platform can then automatically stitch the images together based on the relationships between them to obtain the overall view of the historical building, and then automatically perform the 3D modeling.

[0085] In summary, this embodiment corrects 3D point cloud data and extracts the depth change amplitude of point cloud by using the oblique photography angle of historical building images, thereby achieving reasonable regional division of historical building images, improving the precision of feature extraction, and thus realizing accurate construction of 3D models of historical buildings. This effectively solves the problems of poor accuracy and low efficiency in 3D modeling caused by inaccurate acquisition of historical building feature vectors in existing technologies.

[0086] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0087] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0088] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0089] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0090] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0091] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for historical building three-dimensional supervision modeling based on multi-source geographic data fusion, characterized in that, The method comprises the following steps: S1, acquiring multi-source geographic data, the multi-source geographic data comprising historical building images and 3D point cloud data, and performing regional division on the historical building images to obtain each image region of the historical building images; S2, performing feature extraction on each image region of the historical building images to obtain a feature vector of each image region, and performing secondary regional division on the historical building images based on the feature vector to obtain each secondary division region; S3, combining any two historical building images to obtain an image group, extracting a feature vector of each specified secondary division region of the image group, and performing feature vector similarity analysis to obtain a comprehensive feature vector similarity of the image group; S4, performing adjacent determination based on the comprehensive feature vector similarity of the image group and the 3D point cloud data to obtain a 3D adjacent determination result, if the 3D adjacent determination result is adjacent, performing image stitching on the image group, otherwise, analyzing to obtain a temporary storage location of the image group, and continuously performing adjacent determination on the image group until a complete image stitching relationship is obtained; S5, inputting the complete image stitching relationship and the historical building images into a preset 3D modeling platform to perform 3D modeling, thereby obtaining a historical building three-dimensional model; The comprehensive feature vector similarity of the image group is obtained by the following method: The specified secondary division region is an edge region of the historical building image; sequentially arranging each specified secondary division region according to a preset order to obtain a secondary division region sequence, mapping and matching each specified secondary division region of the image group based on the secondary division region sequence to obtain each region mapping group; performing feature vector similarity analysis on each region mapping group to obtain a feature vector similarity of each region mapping group; comparing the feature vector similarity of each region mapping group with a preset first feature vector similarity threshold in the database, if the feature vector similarity of a certain region mapping group is below the first feature vector similarity threshold, matching the feature vector similarity of the region mapping group with the database to obtain a comprehensive influence reference coefficient of the region mapping group, if the feature vector similarity of a certain region mapping group is greater than the first feature vector similarity threshold, obtaining a preset historical comprehensive influence reference coefficient in the database as the comprehensive influence reference coefficient of the region mapping group, thereby obtaining the comprehensive influence reference coefficient of each region mapping group; performing feature vector similarity fluctuation analysis based on the feature vector similarity of each region mapping group to obtain a feature vector similarity fluctuation parameter, the feature vector similarity fluctuation parameter comprising a feature vector similarity range, a feature vector similarity standard deviation, and a feature vector similarity average value; comparing the feature vector similarity fluctuation parameter with a preset feature vector similarity fluctuation benchmark set in the database to obtain a comparison analysis result, introducing a corresponding first weighting factor based on the comparison analysis result for coupling processing to obtain a coupling processing result, and performing multiplicative coupling processing on the coupling processing result and the comprehensive influence reference coefficient to obtain the comprehensive feature vector similarity of the image group; The feature vector similarity fluctuation benchmark set comprises a feature vector similarity range benchmark value, a feature vector similarity standard deviation benchmark value, and a feature vector similarity average value benchmark value.

2. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion according to claim 1, characterized in that: The method for obtaining each image region of the historical building image comprises the following steps: obtaining the oblique photography angle of the historical building image; performing oblique angle correction on the 3D point cloud data of the historical building image based on the oblique photography angle of the historical building image, to obtain corrected 3D point cloud data of the historical building image, and marking the corrected 3D point cloud data as oblique correction 3D point cloud data of the historical building image; extracting the point cloud depth variation amplitude of the historical building image based on the oblique correction 3D point cloud data of the historical building image, and processing the point cloud depth variation amplitude to obtain a first region division specification of the historical building image; performing region division on the historical building image based on the first region division specification, to obtain each image region of the historical building image.

3. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion according to claim 1, characterized in that: The method for obtaining each secondary division region comprises the following steps: extracting features of each image region to obtain a feature vector of each image region; mapping the feature vector of each image region to the historical building image respectively to obtain a feature vector mapping image; analyzing the feature vector based on the feature vector of each image region to obtain a mean vector distance of the feature vector; matching the mean vector distance of the feature vector with a database to obtain a second region division specification, and performing secondary region division on the feature vector mapping image based on the second region division specification, thereby obtaining each secondary division region of the historical building image and a feature vector of each secondary division region. The method for obtaining the feature vector mapping image comprises the following steps: based on the feature vector extracted from each image region, combining the corresponding image coordinate position, visualizing the feature information, and superimposing the feature information on the original historical building image to form the feature vector mapping image.

4. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion according to claim 1, characterized in that: The method for obtaining the 3D adjacent determination result comprises the following steps: obtaining a preset second feature vector similarity threshold value in the database, and comparing the second feature vector similarity threshold value with the comprehensive feature vector similarity of the image group to obtain a tentative adjacent determination result, wherein if the comprehensive feature vector similarity of the image group is greater than the second feature vector similarity threshold value, the tentative adjacent determination result of the image group is adjacent, otherwise, the tentative adjacent determination result of the image group is not adjacent; if the tentative adjacent determination result of a certain image group is adjacent, performing 3D adjacent determination, and if the tentative adjacent determination result is not adjacent, storing the image group in a repulsion temporary storage area for temporary storage.

5. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion according to claim 4, characterized in that: If the tentative adjacent determination result of a certain image group is adjacent, the method for performing 3D adjacent determination comprises the following steps: obtaining the 3D point cloud data of the specified secondary division region of the image group, and performing 3D point cloud data depth analysis to obtain a 3D point cloud average depth difference degree value of the image group, thereby obtaining a 3D point cloud data depth determination result of the image group, wherein if the 3D point cloud data depth determination result is adjacent, the 3D adjacent determination result of the image group is 3D adjacent, and if the 3D point cloud data depth determination result is not adjacent, the 3D adjacent determination result of the image group is 3D not adjacent. The 3D point cloud average depth difference degree value of the image group is obtained by the following method: obtaining the 3D point cloud depth minimum value of the specified secondary division region of the historical building image in the image group, and taking the horizontal plane where the 3D point cloud depth minimum value is located as the reference surface, thereby obtaining the 3D point cloud relative depth of each random sample point of the specified secondary division region, and obtaining the 3D point cloud average depth of the specified secondary division region through mean processing; The 3D point cloud average depth difference degree of two historical building images in the image group is analyzed to obtain the 3D point cloud average depth difference degree value of the image group; The 3D point cloud data depth determination result is obtained by the following method: The 3D point cloud depth difference first threshold value in the database is obtained and compared with the 3D point cloud average depth difference degree value of the image group to obtain the 3D point cloud data depth determination result of the image group. If the 3D point cloud average depth difference degree value of the image group is above the 3D point cloud depth difference first threshold value, the 3D point cloud data depth determination result of the image group is not adjacent, otherwise, the 3D point cloud data depth determination result of the image group is adjacent.

6. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion according to claim 5, characterized in that: The temporary storage location of the image group is divided by the following method: Based on the 3D adjacent determination result, if the 3D adjacent determination result of the image group is not adjacent, the 3D point cloud depth difference second threshold value in the database is obtained and compared with the 3D point cloud average depth difference degree value of the image group. If the 3D point cloud average depth difference degree value of the image group is above the 3D point cloud depth difference second threshold value, the image group is stored in the ordinary image temporary storage area. If the 3D point cloud average depth difference degree value is less than the 3D point cloud depth difference second threshold value, the image group corresponding to the 3D point cloud average depth difference degree value is stored in the tentative related image temporary storage area. The temporary storage location includes the ordinary image temporary storage area, the tentative related image temporary storage area and the repulsion storage area.

7. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion according to claim 1, characterized in that: The image group adjacent judgment is continued by the following method: An image group that has not been determined to be image spliced is obtained and analyzed to obtain any one historical building image in the image group that has been spliced with other historical building images, and the historical building image is marked as a spliced building image. Each historical building image in the image group that has not been spliced with the spliced building image is marked as a remaining related historical building image, thereby obtaining each remaining related historical building image. Any spliced building image in the image group is marked as an adjacent spliced building image. The image group adjacent judgment is continued by the following method:

8. The historical building three-dimensional supervision modeling method based on multi-source geographic data fusion according to claim 1, characterized in that: The execution parameters of the image acquisition device are adjusted based on the influence of the external environment of the historical building, thereby obtaining each historical building image. The shooting influence parameters are obtained, including light intensity, environmental temperature, environmental color temperature and atmospheric visibility. The preset shooting standard set in the database is acquired and compared and analyzed with the shooting influence parameters to obtain a comparison and analysis result, a corresponding second weighting factor is introduced based on the comparison and analysis result for coupling processing to obtain a historical building shooting influence complexity value; Based on the historical building shooting influence complexity value and the database, a shooting first influence factor is obtained, the photosensitivity and exposure time of the image acquisition device are adjusted based on the shooting first influence factor, and multi-source shooting of each historical building region to be shot is performed after the adjustment to obtain each historical building image; The shooting standard set includes an illumination intensity standard value, an environment temperature standard value, an environment color temperature standard value and an atmospheric visibility standard value; The shooting first influence factor is obtained by the following method: each historical building shooting influence complexity reference gradient and a shooting first influence reference factor corresponding to each historical building shooting influence complexity reference gradient in the database are acquired, and the historical building shooting influence complexity value is compared; if the historical building shooting influence complexity value is within a certain historical building shooting influence complexity reference gradient, the shooting first influence reference factor corresponding to the historical building shooting influence complexity reference gradient is taken as the shooting first influence factor, wherein the historical building shooting influence complexity reference gradient corresponds to a historical building shooting influence complexity value interval.

Images