Mobile rapid modeling system and method based on live-action three-dimensional modeling technology

Abstract

The invention discloses a mobile rapid modeling system based on a live-action three-dimensional modeling technology. The mobile rapid modeling system comprises a multi-view-angle data acquisition module used for acquiring three-dimensional point cloud data and image information of a target area in real time; the data preprocessing module is used for carrying out primary processing on the collected data, detecting an overlapping region in the multi-view data and marking redundant data; the data fusion and feedback module has data fusion, compression and real-time feedback functions; the invention discloses a mobile rapid modeling method based on a live-action three-dimensional modeling technology. The mobile rapid modeling method comprises the following steps: S1, data acquisition and primary processing; and S2, identifying feature points in the overlapped data region by adopting a feature point matching algorithm, and reserving representative point cloud data after performing intelligent analysis on the overlapped data. According to the method, a large amount of point cloud data can be effectively processed, redundant information is removed, the data processing efficiency is improved, the modeling precision can be dynamically allocated according to the importance of the region, and high-precision modeling of the key region is ensured.

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional modeling technology, and in particular to a mobile rapid modeling system and method based on real-scene three-dimensional modeling technology. Background Art

[0002] In recent years, real-scene-based 3D reconstruction technology has been widely used in smart cities, virtual reality, unmanned driving, geographic information systems and other fields. This technology obtains 3D data and image information of objects or scenes through multi-view acquisition equipment, and converts them into highly realistic 3D models. This type of modeling usually relies on high-resolution point cloud data and image information, but the massive data it generates often brings pressure on storage and computing resources, especially when performing real-time modeling in mobile devices, the requirements for computing efficiency and storage capacity are more stringent.

[0003] Traditional 3D modeling methods have repeated coverage in multi-view data collection, and redundant data occupies a large amount of storage space and affects the subsequent data processing speed. In addition, the existing technology usually adopts a unified modeling accuracy when building the model, and does not perform hierarchical control according to different areas in the scene, resulting in waste of resources and reduced modeling efficiency. The 3D modeling system shows high resource consumption and slow response speed in actual applications. Therefore, it is necessary to design a mobile fast modeling system and method based on real-scene 3D modeling technology to solve the above problems. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the problems existing in the above-mentioned existing mobile rapid modeling system and method based on real-scene three-dimensional modeling technology, the present invention is proposed.

[0006] Therefore, the purpose of the present invention is to provide a mobile rapid modeling system and method based on real-scene three-dimensional modeling technology, which is suitable for solving the problems of high resource consumption and slow response speed of three-dimensional modeling systems in practical applications.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a mobile rapid modeling system based on real-scene 3D modeling technology, comprising:

[0008] Multi-view data acquisition module: used to collect 3D point cloud data and image information of the target area in real time;

[0009] Data preprocessing module: performs preliminary processing on the collected data, detects overlapping areas in multi-view data and marks redundant data;

[0010] Data redundancy detection and elimination module: detect redundant data and retain representative data;

[0011] Multi-level modeling control module: Identify and mark key areas and secondary areas, and assign different modeling precisions according to the importance of the areas;

[0012] Data fusion and feedback module: including data fusion, compression and real-time feedback functions;

[0013] Dynamic adjustment module: adjusts the acquisition parameters of the multi-view data acquisition module in real time according to the feedback information from the data fusion and feedback module;

[0014] Data storage and post-processing module: storage and management of optimized 3D data, and further cleaning and compression of data to improve data quality.

[0015] A mobile rapid modeling method based on real-scene 3D modeling technology, the rapid modeling method is applicable to the above rapid modeling system, and the rapid modeling method comprises the following steps:

[0016] S1: data collection and preliminary processing;

[0017] S2: Use feature point matching algorithm to identify feature points in overlapping data areas, perform intelligent analysis on overlapping data, retain representative point cloud data, remove redundant points with high similarity, and construct data redundancy similarity detection formula;

[0018] S3: Introducing machine learning algorithms to automatically identify key areas of the scene and mark them hierarchically as “high-precision areas” and “low-precision areas”;

[0019] S4: Integrate the layered data and perform secondary compression on the modeling results during the data storage stage;

[0020] S5: Real-time feedback on modeling effects. Based on the real-time feedback information, the system dynamically adjusts the modeling parameters and changes the acquisition frequency and resolution in real time;

[0021] S6: After modeling is completed, a final post-processing step is performed to remove any residual redundant data that may exist.

[0022] As a preferred solution of the mobile rapid modeling method based on real-scene 3D modeling technology described in the present invention, the data redundancy similarity detection formula is as follows:

[0023]

[0024] Among them, Ssimi is the similarity score, indicating the degree of overlap and similarity of point cloud data, Ω raw Represents the space of the original collected point cloud data, f(x,y,z) similarity function, which is used to evaluate the similarity between data points from different perspectives, It is an exponential decay function, and the parameter α is used to control the influence of similarity on the score.

[0025] As a preferred solution of the mobile rapid modeling method based on real-scene 3D modeling technology described in the present invention, wherein: in the output result of the data redundancy similarity detection formula, the first threshold is set to T1;

[0026] If S simi <T1, indicating that the redundancy of the current data is low, and the redundancy elimination process is skipped directly;

[0027] If S simi ≥T1, indicating that the current data redundancy is high, and the data cleaning and precision division stage is entered.

[0028] As a preferred solution of the mobile rapid modeling method based on real-scene 3D modeling technology described in the present invention, a multi-level modeling accuracy division formula is constructed according to the output result of the data redundancy similarity detection formula, and the multi-level modeling accuracy division formula is as follows:

[0029]

[0030] Among them, P prec Assign scores to multiple levels of accuracy to determine the modeling accuracy of each point cloud data. multi (x, y, z) is the value of the multi-level modeling control function, which dynamically selects the modeling accuracy according to the importance of the data point. φ(x, y, z) is a smoothing control function based on the Gaussian kernel, which makes the accuracy distribution smooth and coherent in space.

[0031] As a preferred solution of the mobile rapid modeling method based on real-scene 3D modeling technology described in the present invention, the multi-level modeling control function formula is as follows:

[0032]

[0033] Among them, multi (x, y, z) represents the modeling accuracy requirement of the current point cloud data point. The higher the value, the higher the accuracy of the modeling required for the point. The lower the value, the lower the accuracy of the modeling required for the point. (x, y, z) represents the spatial coordinates of the current point cloud data, which is used to identify the position of the point in the overall model. β is the importance weight coefficient, which is used to adjust the influence of regional importance on modeling accuracy. The larger the value, the higher the importance of the system to the importance weight. import(x,y,z) importance function, represents the edge attenuation term, which is used to control the accuracy requirement of each point in the point cloud relative to the edge distance. λ is the edge distance influence coefficient, which is used to control the influence of edge distance on modeling accuracy. edge (x, y, z) is the distance function, which indicates the distance from the current point to the edge or contour of the model.

[0034] As a preferred solution of the mobile rapid modeling method based on real-scene 3D modeling technology described in the present invention, wherein: the second threshold value is set as T2 in the output result of the multi-level modeling accuracy division formula;

[0035] If P prec >T2, it is a critical area and high-precision modeling is adopted;

[0036] If P prec ≤T2, it is a secondary area and is modeled with low precision.

[0037] As a preferred solution of the mobile rapid modeling method based on real-scene 3D modeling technology described in the present invention, a data compression and feedback integration formula is constructed according to the output result of the multi-level modeling accuracy division formula, and the data compression and feedback integration formula is as follows:

[0038]

[0039] Among them, C compre Score compression feedback, indicating the current system's data optimization level and storage load balance. It is a normalization term used to balance the redundant effects of multi-view data acquisition and make the compression feedback score more stable.

[0040] As a preferred solution of the mobile rapid modeling method based on real-scene 3D modeling technology described in the present invention, wherein: in the output result of the data compression and feedback integration formula, a third threshold value T3 is set;

[0041] If C compre >T3, indicating that the current system load is high, and the acquisition accuracy and frequency need to be reduced;

[0042] If C compre ≤T3, the current compression effect is good and the existing acquisition parameters are maintained.

[0043] As a preferred solution of the mobile rapid modeling method based on real-scene three-dimensional modeling technology described in the present invention, in which: in S6, the optimized data is classified and stored, and high-precision data and low-precision data are managed separately for subsequent calls, further reducing the reading and loading time and improving the system response speed.

[0044] Beneficial effects of the present invention: through the combination of the multi-view data acquisition module and the data redundancy detection and elimination module, the system can effectively process a large amount of point cloud data, remove redundant information, improve data processing efficiency, and reduce storage requirements. The multi-level modeling control module can dynamically allocate modeling accuracy according to the importance of the area by identifying key areas and secondary areas, ensuring that key areas are modeled with high accuracy, while unimportant areas can moderately reduce accuracy. This hierarchical processing improves the flexibility of modeling and resource utilization;

[0045] The data fusion and feedback module is combined with the dynamic adjustment module to realize real-time monitoring and adjustment of the modeling process, automatically optimize the acquisition parameters according to the actual feedback, and ensure the efficiency and accuracy of the modeling process. This mobile rapid modeling system improves the efficiency and accuracy of three-dimensional modeling while having good adaptability and scalability. It is suitable for multiple fields, including but not limited to architecture and urban planning, virtual reality and augmented reality, autonomous driving and drone technology, and geographic information systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0047] Figure 1 This is a schematic diagram of the overall framework of a mobile rapid modeling system based on real-scene 3D modeling technology proposed by the present invention;

[0048] Figure 2 The present invention provides a schematic diagram of the steps of a mobile rapid modeling method based on real-scene 3D modeling technology. DETAILED DESCRIPTION

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0050] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0051] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0052] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0053] Embodiment 1

[0054] Reference Figure 1-Figure 2 , is an embodiment of the present invention, and provides a mobile rapid modeling system based on real-scene 3D modeling technology, comprising:

[0055] Multi-view data acquisition module: contains multi-view sensors (such as lidar, depth camera and panoramic camera), which is used to collect 3D point cloud data and image information of the target area in real time. The data collected by this module is transmitted to the data preprocessing module;

[0056] Data preprocessing module: performs preliminary processing on the collected data, marks the overlapping areas of multi-view data, detects the overlapping areas in multi-view data and marks the redundant data, and transmits the processed data to the data redundancy detection and elimination module;

[0057] Data redundancy detection and elimination module: detects redundant data, retains representative data, and transmits processed data to the multi-level modeling control module;

[0058] Multi-level modeling control module: Identify and mark key areas and secondary areas, and assign different modeling precisions according to the importance of the areas;

[0059] Data fusion and feedback module: including data fusion, compression and real-time feedback functions;

[0060] This module optimizes the processed 3D data through downsampling and data compression, and provides real-time modeling effect feedback. If it detects that the key area is under-modeled, it prompts the operator to adjust the acquisition through feedback information, or notifies the dynamic adjustment module. At the same time, the processed data is transmitted to the data storage and post-processing module;

[0061] Dynamic adjustment module: According to the feedback information from the data fusion and feedback module, the acquisition parameters of the multi-view data acquisition module are adjusted in real time, including acquisition frequency, resolution and viewing angle, to ensure modeling quality and efficiency;

[0062] Data storage and post-processing module: storage and management of optimized 3D data, and further cleaning and compression of data to improve data quality.

[0063] A mobile rapid modeling method based on real-scene 3D modeling technology, the rapid modeling method is applicable to the above rapid modeling system, and the rapid modeling method comprises the following steps:

[0064] S1: data collection and preliminary processing;

[0065] S2: A feature point matching algorithm is used to identify feature points in overlapping data areas. After intelligent analysis of overlapping data, representative point cloud data is retained, redundant points with high similarity are removed, and a data redundancy similarity detection formula is constructed. This not only improves the validity and accuracy of the data, but also reduces storage requirements and improves the efficiency of subsequent processing.

[0066] S3: Introducing machine learning algorithms to automatically identify key areas of the scene and mark them hierarchically as “high-precision areas” and “low-precision areas”;

[0067] The machine learning algorithm is used to automatically identify and hierarchically mark the key areas of the scene ("high-precision areas" and "low-precision areas"), realizing intelligent data processing. This function not only improves the flexibility of modeling, but also adjusts the modeling strategy according to actual needs to ensure high-precision modeling of key areas;

[0068] S4: Integrate the layered data and perform secondary compression on the modeling results during the data storage stage;

[0069] S5: Real-time feedback on modeling effects. Based on the real-time feedback information, the system dynamically adjusts the modeling parameters and changes the acquisition frequency and resolution in real time;

[0070] Dynamically adjust modeling parameters (such as acquisition frequency and resolution) through real-time feedback information to effectively adapt to changing scenarios and needs. This process ensures the optimal use of resources in the actual modeling process, improves modeling efficiency and the accuracy of results. In a rapidly changing environment, real-time parameter adjustment can ensure that the model always reflects the actual situation of the current scenario and improves decision support capabilities;

[0071] S6: After modeling is completed, the final post-processing step is performed to remove any residual redundant data that may exist;

[0072] The optimized data is classified and stored, and high-precision data and low-precision data are managed separately to facilitate subsequent calls, further reducing reading and loading time and improving system response speed; in application scenarios that require fast access to historical data, classified storage can greatly improve data availability and system work efficiency.

[0073] The data redundancy similarity detection formula is as follows:

[0074]

[0075] Among them, S simi is the similarity score, indicating the degree of overlap and similarity of point cloud data, Ω raw Represents the space of the original collected point cloud data, f(x,y,z) similarity function, which is used to evaluate the similarity between data points from different perspectives, It is an exponential decay function, and the parameter α is used to control the influence of similarity on the score.

[0076] In the output result of the data redundancy similarity detection formula, the first threshold is set to T1;

[0077] If S simi <T1, indicating that the redundancy of the current data is low, and the redundancy elimination process is skipped directly;

[0078] If S simi ≥T1, indicating that the current data redundancy is high, and the data cleaning and precision division stage is entered.

[0079] A multi-level modeling accuracy division formula is constructed based on the output result of the data redundancy similarity detection formula, and the multi-level modeling accuracy division formula is as follows:

[0080]

[0081] Among them, P prec Assign scores to multiple levels of accuracy to determine the modeling accuracy of each point cloud data. multi (x, y, z) is the value of the multi-level modeling control function, which dynamically selects the modeling accuracy according to the importance of the data point. φ(x, y, z) is a smoothing control function based on the Gaussian kernel, which makes the accuracy distribution smooth and coherent in space.

[0082] The multi-level modeling control function formula is as follows:

[0083]

[0084] Among them, multi (x, y, z) represents the modeling accuracy requirement of the current point cloud data point. The higher the value, the higher the accuracy of the modeling required for the point. The lower the value, the lower the accuracy of the modeling required for the point. (x, y, z) represents the spatial coordinates of the current point cloud data, which is used to identify the position of the point in the overall model. β is the importance weight coefficient, which is used to adjust the influence of regional importance on modeling accuracy. The larger the value, the higher the importance of the system to the importance weight. import (x,y,z) importance function, represents the edge attenuation term, which is used to control the accuracy requirement of each point in the point cloud relative to the edge distance. λ is the edge distance influence coefficient, which is used to control the influence of edge distance on modeling accuracy. edge (x, y, z) is the distance function, which indicates the distance from the current point to the edge or contour of the model.

[0085] The second threshold value is set as T2 in the output result of the multi-level modeling accuracy division formula;

[0086] If P prec >T2, it is a critical area and high-precision modeling is adopted;

[0087] If P prec ≤T2, it is a secondary area and is modeled with low precision.

[0088] The data compression and feedback integration formula is constructed according to the output result of the multi-level modeling accuracy division formula, and the data compression and feedback integration formula is as follows:

[0089]

[0090] Among them, C compre Score compression feedback, indicating the current system's data optimization level and storage load balance. It is a normalization term used to balance the redundant effects of multi-view data acquisition and make the compression feedback score more stable.

[0091] In the output result of the data compression and feedback integration formula, a third threshold value T3 is set;

[0092] If C compre >T3, indicating that the current system load is high, and the acquisition accuracy and frequency need to be reduced;

[0093] If C compre ≤T3, the current compression effect is good and the existing acquisition parameters are maintained.

[0094] During use, through the combination of the multi-view data acquisition module and the data redundancy detection and elimination module, the system can effectively process a large amount of point cloud data, remove redundant information, improve data processing efficiency, and reduce storage requirements. The multi-level modeling control module can dynamically allocate modeling accuracy according to the importance of the area by identifying key areas and secondary areas, ensuring high-precision modeling in key areas, while the accuracy of unimportant areas can be appropriately reduced. The present invention has good adaptability and scalability while improving the efficiency and accuracy of three-dimensional modeling.

[0095] Embodiment 2

[0096] Referring to Table 1, which is a second embodiment of the present invention, this embodiment is different from the first embodiment in that, in order to verify its beneficial effects, experimental comparison data between the present invention and the prior art are provided.

[0097] In this embodiment, we use real-scene 3D modeling technology to quickly model a part of a city square to verify the performance of the present invention. The test site includes high-precision areas (such as the central statue) and low-precision areas (such as the surrounding open areas). The selected equipment includes multi-view sensors (lidar, depth camera and panoramic camera) to capture high-quality 3D point cloud data and image information. During the data collection process, the multi-view acquisition module will transmit all data to the data preprocessing module in real time for preliminary processing;

[0098] The system starts data collection and preliminary processing, identifies overlapping data areas through feature point matching algorithms and removes redundant points. We use similarity detection formulas to calculate the redundancy of point cloud data, remove point cloud data with high similarity, and use machine learning algorithms to automatically identify key areas such as statues in the square and mark them as "high-precision areas". The surrounding ordinary areas are marked as "low-precision areas". After the layered data integration is completed, the system compresses the three-dimensional data to reduce the storage volume, and provides real-time feedback on the modeling effect through the data fusion and feedback module. When the system detects in real time that the modeling effect of the statue is relatively fuzzy, it notifies the dynamic adjustment module through the feedback module to increase the acquisition frequency and resolution of the statue area to ensure high-precision modeling effects in key areas. After the modeling is completed, the system automatically performs data post-processing and stores the optimized data separately by precision to improve the efficiency of subsequent calls and loading.

[0099] Table 1: Performance comparison of rapid prototyping system and existing technology

[0100]

[0101]

[0102] The data in Table 1 above are analyzed. In terms of data volume, the present invention reduces the data volume of high-precision areas by 40% and the data volume of low-precision areas by 33% compared with the prior art through feature point matching algorithm and redundant data detection. In terms of processing time, the intelligent redundant detection and compression processing of the present invention shortens the processing time of high-precision areas by 41.6% and the processing time of low-precision areas by 50%.

[0103] Further analysis of the precision data, using machine learning algorithms to automatically identify and hierarchically label scenes, significantly improved the modeling accuracy of high-precision areas, from 0.85 of the existing technology to 0.95. In comparison, although the accuracy of low-precision areas has been slightly improved (from 0.65 to 0.70), this optimization has allocated resources more reasonably and achieved accurate capture of key areas.

[0104] The redundant data clearing module of the present invention greatly reduces the burden of invalid data storage. Redundant data clearing saves about 5MB of storage space, further reducing the total data volume and system storage requirements. Overall, the present invention is significantly superior to the prior art in terms of data volume, processing time and accuracy.

[0105] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A mobile rapid modeling system based on real-scene 3D modeling technology, characterized in that: include: Multi-view data acquisition module: used to collect 3D point cloud data and image information of the target area in real time; Data preprocessing module: performs preliminary processing on the collected data, detects overlapping areas in multi-view data and marks redundant data; Data redundancy detection and elimination module: detect redundant data and retain representative data; Multi-level modeling control module: Identify and mark key areas and secondary areas, and assign different modeling precisions according to the importance of the areas; Data fusion and feedback module: including data fusion, compression and real-time feedback functions; Dynamic adjustment module: adjusts the acquisition parameters of the multi-view data acquisition module in real time according to the feedback information from the data fusion and feedback module; Data storage and post-processing module: storage and management of optimized 3D data, and further cleaning and compression of data to improve data quality.

2. A mobile rapid modeling method based on real-scene 3D modeling technology, the rapid modeling method is applicable to the rapid modeling system in claim 1 above, characterized in that: The rapid modeling method comprises the following steps: S1: data collection and preliminary processing; S2: Use feature point matching algorithm to identify feature points in overlapping data areas, perform intelligent analysis on overlapping data, retain representative point cloud data, remove redundant points with high similarity, and construct data redundancy similarity detection formula; S3: Introducing machine learning algorithms to automatically identify key areas of the scene and mark them hierarchically as "high-precision areas" and "low-precision areas"; S4: Integrate the layered data and perform secondary compression on the modeling results during the data storage stage; S5: Real-time feedback on modeling effects. Based on the real-time feedback information, the system dynamically adjusts the modeling parameters and changes the acquisition frequency and resolution in real time; S6: After modeling is completed, a final post-processing step is performed to remove any residual redundant data that may exist.

3. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 2 is characterized by: The data redundancy similarity detection formula is as follows: Among them, S simi is the similarity score, indicating the degree of overlap and similarity of point cloud data, Ω raw Represents the space of the original collected point cloud data, f(x,y,z) similarity function, which is used to evaluate the similarity between data points from different perspectives, It is an exponential decay function, and the parameter α is used to control the influence of similarity on the score.

4. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 3 is characterized by: In the output result of the data redundancy similarity detection formula, the first threshold is set to T1; If S simi <T1, indicating that the redundancy of the current data is low, and the redundancy elimination process is skipped directly; If S simi ≥T1, indicating that the current data redundancy is high, and the data cleaning and precision division stage is entered.

5. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 3 is characterized by: A multi-level modeling accuracy division formula is constructed based on the output result of the data redundancy similarity detection formula, and the multi-level modeling accuracy division formula is as follows: Among them, P prec Assign scores to multiple levels of accuracy to determine the modeling accuracy of each point cloud data. multi (x, y, z) is the value of the multi-level modeling control function, which dynamically selects the modeling accuracy according to the importance of the data point. φ(x, y, z) is a smoothing control function based on the Gaussian kernel, which makes the accuracy distribution smooth and coherent in space.

6. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 5 is characterized by: The multi-level modeling control function formula is as follows: Among them, multi (x, y, z) represents the modeling accuracy requirement of the current point cloud data point. (x, y, z) represents the spatial coordinates of the current point cloud data, which is used to identify the position of the point in the overall model. β is the importance weight coefficient, which is used to adjust the influence of regional importance on modeling accuracy. The larger the value, the higher the importance the system attaches to the importance weight. import (x,y,z) importance function, represents the edge attenuation term, which is used to control the accuracy requirement of each point in the point cloud relative to the edge distance. λ is the edge distance influence coefficient, which is used to control the influence of edge distance on modeling accuracy. edge (x, y, z) is the distance function, which indicates the distance from the current point to the edge or contour of the model.

7. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 5 is characterized by: The second threshold value is set as T2 in the output result of the multi-level modeling accuracy division formula; If P prec >T2, it is a critical area and high-precision modeling is adopted; If P prec ≤T2, it is a secondary area and is modeled with low precision.

8. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 5 is characterized by: The data compression and feedback integration formula is constructed according to the output result of the multi-level modeling accuracy division formula, and the data compression and feedback integration formula is as follows: Among them, C compre Score compression feedback, indicating the current system's data optimization level and storage load balance. It is a normalization term used to balance the redundant effects of multi-view data acquisition and make the compression feedback score more stable.

9. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 8, characterized in that: In the output result of the data compression and feedback integration formula, a third threshold value T3 is set; If C compre >T3, indicating that the current system load is high, and the acquisition accuracy and frequency need to be reduced; If C compre ≤T3, the current compression effect is good and the existing acquisition parameters are maintained.

10. The mobile rapid modeling method based on real-scene 3D modeling technology according to claim 2, characterized in that: In S6, the optimized data is classified and stored, and high-precision data and low-precision data are managed separately to facilitate subsequent calls, further reducing the reading and loading time and improving the system response speed.

Images