Lightweight method for oblique photography three-dimensional model

Abstract

The invention discloses a lightweight method for an oblique photography three-dimensional model, and the method comprises the following steps: 1, carrying out the collection of related data of the oblique photography three-dimensional model, and obtaining the oblique photography information; 2, modeling processing is carried out on the collected oblique photography information, and preliminary model information is obtained; 3, performing secondary processing on the preliminary model information to obtain secondary processing model information; 4, processing the secondary processing model information to obtain final model information; and 5, performing model verification evaluation on the final model information, and if the model verification evaluation is normal, exporting the final model information. According to the method, the user experience can be improved, the system requirement can be reduced, the modeling cost can be reduced, the system performance can be optimized, and the portability and compatibility of the model can be improved.

Description

Technical Field

[0001] The invention relates to the field of oblique photography, and in particular to a lightweight method for an oblique photography three-dimensional model. Background Art

[0002] Oblique photography 3D model is a technology that acquires images of an object from multiple angles and constructs its 3D model. Oblique photography technology uses multi-lens photogrammetry technology to take photos of the ground or buildings from multiple angles (including vertical and oblique angles). Through the 3D reconstruction algorithm, the information in these photos can be combined to generate a 3D model of the target. This process is similar to human binocular positioning and can restore the real 3D world.

[0003] In actual use of oblique photography 3D models, lightweight methods are applied to lightweight models, reduce the amount and complexity of model data, and improve the efficiency of data processing, storage, and transmission.

[0004] The existing lightweight methods have poor lightweighting effects, and the model data after lightweighting is easily lost, which has a certain impact on the use of lightweight methods. Therefore, a lightweight method for oblique photography 3D models is proposed. Summary of the invention

[0005] The technical problem to be solved by the present invention is: how to solve the problem that the existing lightweighting method has poor lightweighting effect and the model data after lightweighting is easily lost, which has a certain impact on the use of the lightweighting method, and provides a lightweight method for oblique photography three-dimensional model.

[0006] The present invention solves the above technical problems through the following technical solutions, and the present invention comprises the following steps:

[0007] Step 1: Collect relevant data of the oblique photography 3D model to obtain oblique photography information;

[0008] Step 2: Model the collected oblique photography information to obtain preliminary model information;

[0009] Step 3: Then, the preliminary model information is subjected to secondary processing to obtain secondary processing model information;

[0010] Step 4: Process the secondary processing model information to obtain the final model information;

[0011] Step 5: Perform model validation and evaluation on the final model information. If the model validation and evaluation is normal, the final model information will be exported.

[0012] Furthermore, the specific process of obtaining the oblique photography information is as follows:

[0013] First determine the location where the oblique photography 3D model needs to be established, then collect the take-off position of the drone at the location to obtain the take-off position information of multiple drones;

[0014] Afterwards, the take-off position information of multiple drones is analyzed, and a final drone take-off position is selected, and the drone for oblique photography is transported to the final drone take-off position to collect oblique photography information, and at least two drones are set to the final drone take-off position to collect oblique photography information;

[0015] During the process of collecting oblique photography information by UAV, the status of the UAV is monitored in real time, and the status of the real-time monitored UAV is analyzed. When the status of the UAV is abnormal, control information is generated to control the UAV and re-collect oblique images;

[0016] The specific process of processing the location-related information and obtaining the location evaluation parameters is as follows: extracting the collected location-related information, which includes the number of other buildings, human flow information, visibility information, light intensity information, and the distance between the take-off position and the collection point;

[0017] The number of other buildings is marked as T1, the flow of people is marked as T2, the visibility information is marked as T3, the light intensity information is marked as T4, and the distance from the take-off position to the collection point is marked as T5;

[0018] Score the number of other buildings T1 to obtain the building number score P1. The smaller T1 is, the higher the P1 score is, and the larger T1 is, the lower the P1 score is.

[0019] The pedestrian flow information T2 is scored to obtain a pedestrian flow score P2. The smaller T2 is, the higher the P2 score is, and the larger T2 is, the lower the P2 score is.

[0020] The visibility information T3 is scored to obtain a visibility score P3. The smaller T3 is, the lower the P3 score is, and the larger T3 is, the higher the P3 score is.

[0021] The light intensity information T4 is scored to obtain a light intensity score P4. When T4 is within a preset range, the light intensity score P4 is a preset value a1. When T4 is outside the preset range, P4 is a preset value a2, where a1>a2.

[0022] The distance T5 between the take-off position and the collection point is scored to obtain a distance score P5. The larger T5 is, the lower the P5 score is, and the smaller T5 is, the higher the P5 score is.

[0023] Then the sum of P1, P2, P3, P4 and P5 is calculated to obtain the position evaluation parameters;

[0024] The number of other buildings is the number of buildings with a height greater than a preset value within the preset range of the selected take-off position;

[0025] The specific process of real-time evaluation parameters is as follows: extract the maximum value and the second largest value of the evaluation parameter of the selected position obtained, and mark the positions corresponding to the maximum value and the second largest value of the evaluation parameter of the selected position as Zmax and Zamx-1;

[0026] Two drones were set up at the Zmax position and the Zamx-1 position respectively, and simulated oblique photography data were collected in the same collection order;

[0027] Then, the data collected by the two drones at the Zmax position are marked as f1 and f2, and the mean of f1 and f2 is calculated to obtain the real-time evaluation parameters at the Zmax position;

[0028] Then mark the data volumes collected by the two drones at the Zmax-1 position as e1 and e2, calculate the mean of fe and fe, and obtain the real-time evaluation parameters of the Zmax-1 position.

[0029] Furthermore, the specific process of analyzing the take-off position information of multiple drones and selecting the final drone take-off position is as follows: extract the take-off position information of multiple drones, measure the distance information between each drone take-off position and the location where the oblique photography three-dimensional model needs to be established, and then select the drone take-off position with the closest distance as the final drone take-off position.

[0030] Furthermore, the process of analyzing the real-time monitoring drone status is as follows: extracting the collected real-time monitoring drone status, the real-time monitoring drone status including the real-time speed information and the real-time drone shaking amplitude information when the drone is running;

[0031] Extract the real-time speed information of the drone, calculate the difference between it and the standard speed in the current control command, and obtain the speed difference. When the speed difference exceeds the preset range, it means that the drone is in an abnormal state;

[0032] Then the real-time UAV shaking amplitude information is extracted. When the real-time UAV shaking amplitude information is greater than the preset amplitude, it means that the UAV state is abnormal.

[0033] Furthermore, the acquisition process of the preliminary model information is as follows: extracting the collected oblique photography information, and preliminarily processing the oblique photography information, wherein the preliminary processing includes data preprocessing and data compression;

[0034] The data preprocessing process is as follows: based on the user's actual modeling needs, unnecessary areas or objects are trimmed;

[0035] Then, data normalization is performed: normalizing database design, eliminating duplicate data items, and reducing data redundancy;

[0036] Data deduplication: Deduplication of data is performed to delete duplicate data items to further reduce data redundancy;

[0037] Data merging: Merge multiple data copies into one to reduce data redundancy and improve storage efficiency;

[0038] The specific process of data compression is as follows: Use compression algorithms to reduce redundant data in the model. The compression algorithms used include Quantization and Delta Encoding.

[0039] After data compression is completed, preliminary model information is obtained;

[0040] First, Quantization is used to compress the data to obtain the first compressed data, and then DeltaEncoding is used to compress the data to obtain the second compressed data. Then, Quantization and DeltaEncoding are used to compress the data in sequence to obtain the third compressed data. Then, the smallest data size among the first compressed data, the second compressed data and the third compressed data is selected as the preliminary model information.

[0041] Furthermore, the specific processing process of the secondary processing model information is as follows: extracting the obtained preliminary model information, and performing geometric simplification and texture compression on the preliminary model information;

[0042] The specific process of geometric simplification and texture compression is as follows: the data in the preliminary model information is geometrically simplified by using the Quadric Error Metrics and EdgeCollaps algorithms to obtain two simplified data, and the one with the smallest data volume is selected for texture compression;

[0043] Apply texture compression algorithms to reduce the storage space of texture maps. The texture compression algorithms applied include JPEG, PNG, S3TC / DXTn, ASTC and ETC algorithms;

[0044] After texture compression is completed using JPEG, PNG, S3TC / DXTn, ASTC and ETC algorithms, five texture compression completion data are obtained. Then, the texture compression completion data with the highest quality but the smallest data volume is selected from the five texture compression completion data as the secondary processing model information.

[0045] Furthermore, the specific processing process of the final model information is as follows: extracting the secondary processing model information, first performing model block and stream processing, and then performing root node merging and index optimization, data compression and streaming transmission, and point cloud-based representation;

[0046] The specific process of model segmentation and streaming processing is as follows:

[0047] First, segment the scene and determine the block size based on actual conditions;

[0048] Boundary processing, setting buffers or overlapping areas between blocks;

[0049] Block management priority sorting: prioritize blocks according to the current display position of the model, and load the blocks at the current display position first;

[0050] Compression and optimization: compress and optimize the geometry and texture of each block, and regularly clean up cache blocks that are no longer in use;

[0051] After the compression optimization is completed, the root node is merged and the index is optimized;

[0052] The specific process of root node merging and index optimization is as follows:

[0053] Use a merge tool or algorithm to merge the root nodes into a new root node based on their spatial relationship and similarity, and then use B+ tree index optimization to optimize the index;

[0054] After the root nodes are merged and the index is optimized, data compression and streaming are performed. The specific process of data compression and streaming is as follows: data is compressed using the preset compression method, and after compression is completed, the corresponding streaming method is selected;

[0055] After selecting the streaming mode, the compressed data is analyzed and processed to obtain the complexity of all model data, and the data with a complexity greater than a preset value among all model data is selected for point cloud-based representation;

[0056] After all the model data with a complexity greater than a preset value is represented based on the point cloud, the final model information is obtained.

[0057] Furthermore, the specific process of model verification and evaluation of the final model information is as follows: the final model information is quality evaluated, that is, the lightweight three-dimensional model is quality evaluated, and the first evaluation information is obtained. The quality evaluation content includes geometric accuracy, texture quality, and detail level.

[0058] Then, the original data and the lightweight three-dimensional model are compared to evaluate the effect of data compression and optimization, and obtain the second evaluation information;

[0059] Finally, the rendering efficiency and user experience are evaluated to obtain the third evaluation information. The evaluation of rendering efficiency and user experience includes loading speed and smoothness of roaming operation.

[0060] When there is no abnormality in the first evaluation information, the second evaluation information and the third evaluation information, it means that the model verification evaluation is normal.

[0061] Furthermore, the process of determining that the first evaluation information is free of abnormalities is as follows:

[0062] Extract quality assessment content, and obtain geometric accuracy, texture quality and detail level from the quality assessment content;

[0063] When the geometric accuracy, texture quality and level of detail are within the preset required standards, it means that there is no abnormality in the first evaluation information;

[0064] The process of determining whether the second evaluation information is free of abnormalities is as follows: extracting the effects of compression and optimization of the collected data. When the effects of compression and optimization of the data are within the preset standard requirements, it means that the second evaluation information is free of abnormalities.

[0065] The process of determining whether the third evaluation information is normal is as follows: extract the content for evaluating rendering efficiency and user experience, and obtain the loading speed and the smoothness of roaming operations. When the loading speed and the smoothness of roaming operations meet the preset standard requirements, it means that the third evaluation information is normal.

[0066] Compared with the prior art, the present invention has the following advantages: the lightweight method of the oblique photography three-dimensional model, in the data collection process of the oblique photography three-dimensional model, performs the setting of relevant oblique photography data collection, thereby effectively reducing the invalid data in the collected data, and realizing the lightweight of the collection process, that is, ensuring the lightweight of the subsequently established oblique photography three-dimensional model, and then the oblique photography three-dimensional model undergoes multi-level lightweight processing, which can speed up the loading speed of the oblique photography three-dimensional model and reduce system requirements. The lightweight model is easier to process and display. The lightweight model has a fast response speed and can greatly enhance the user's interactive experience. Under the premise of ensuring the quality and accuracy of the model, lightweighting can reduce the model's demand for computing resources, thereby optimizing the overall performance of the system. Therefore, lightweighting the oblique photography three-dimensional model can not only improve the user experience, reduce system requirements and improve interactivity, but also reduce modeling costs and optimize system performance. These benefits make lightweighting an indispensable part of oblique photography three-dimensional model processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is the overall flow chart of the present invention. DETAILED DESCRIPTION

[0068] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiment.

[0069] like Figure 1 As shown, this embodiment provides a technical solution: a lightweight method for an oblique photography three-dimensional model, comprising the following steps:

[0070] Step 1: Collect relevant data of the oblique photography 3D model to obtain oblique photography information;

[0071] Step 2: Model the collected oblique photography information to obtain preliminary model information;

[0072] Step 3: Then, the preliminary model information is subjected to secondary processing to obtain secondary processing model information;

[0073] Step 4: Process the secondary processing model information to obtain the final model information;

[0074] Step 5: Perform model verification and evaluation on the final model information. If the model verification and evaluation is normal, the final model information will be exported;

[0075] If the model verification and evaluation is abnormal, the data lightweight processing of the oblique photography 3D model establishment will be carried out again;

[0076] The 3D image processing technology of the present invention can realize the light-weighting of the 3D image processing and the light-weighting of the 3D image processing.

[0077] The specific process of obtaining the oblique photography information is as follows:

[0078] First determine the location where the oblique photography 3D model needs to be established, then collect the take-off position of the drone at the location to obtain the take-off position information of multiple drones;

[0079] Afterwards, the take-off position information of multiple drones is analyzed, and a final drone take-off position is selected, and the drone for oblique photography is transported to the final drone take-off position to collect oblique photography information, and at least two drones are set to the final drone take-off position to collect oblique photography information;

[0080] The specific process of processing the location-related information and obtaining the location evaluation parameters is as follows: extracting the collected location-related information, which includes the number of other buildings, human flow information, visibility information, light intensity information, and the distance between the take-off position and the collection point;

[0081] The number of other buildings is marked as T1, the flow of people is marked as T2, the visibility information is marked as T3, the light intensity information is marked as T4, and the distance from the take-off position to the collection point is marked as T5;

[0082] Score the number of other buildings T1 to obtain the building number score P1. The smaller T1 is, the higher the P1 score is, and the larger T1 is, the lower the P1 score is.

[0083] The pedestrian flow information T2 is scored to obtain a pedestrian flow score P2. The smaller T2 is, the higher the P2 score is, and the larger T2 is, the lower the P2 score is.

[0084] The visibility information T3 is scored to obtain a visibility score P3. The smaller T3 is, the lower the P3 score is, and the larger T3 is, the higher the P3 score is.

[0085] The light intensity information T4 is scored to obtain a light intensity score P4. When T4 is within a preset range, the light intensity score P4 is a preset value a1. When T4 is outside the preset range, P4 is a preset value a2, where a1>a2.

[0086] The distance T5 between the take-off position and the collection point is scored to obtain a distance score P5. The larger T5 is, the lower the P5 score is, and the smaller T5 is, the higher the P5 score is.

[0087] Then the sum of P1, P2, P3, P4 and P5 is calculated to obtain the position evaluation parameters;

[0088] The number of other buildings is the number of buildings with a height greater than a preset value within the preset range of the selected take-off position;

[0089] The specific process of real-time evaluation parameters is as follows: extract the maximum value and the second largest value of the evaluation parameter of the selected position obtained, and mark the positions corresponding to the maximum value and the second largest value of the evaluation parameter of the selected position as Zmax and Zamx-1;

[0090] Two drones were set up at the Zmax position and the Zamx-1 position respectively, and simulated oblique photography data were collected in the same collection order;

[0091] Then, the data collected by the two drones at the Zmax position are marked as f1 and f2, and the mean of f1 and f2 is calculated to obtain the real-time evaluation parameters at the Zmax position;

[0092] Then mark the data collected by the two drones at the Zmax-1 position as e1 and e2, calculate the mean of fe and fe, and obtain the real-time evaluation parameters at the Zmax-1 position;

[0093] During the process of collecting oblique photography information by UAV, the status of the UAV is monitored in real time, and the status of the real-time monitored UAV is analyzed. When the status of the UAV is abnormal, control information is generated to control the UAV and re-collect oblique images;

[0094] At least two drones simultaneously go to the final drone take-off position to collect oblique photography information. After obtaining the oblique photography information, at least two oblique photography information are analyzed and compared, and the data file sizes of the at least two oblique photography information are compared, and the oblique photography information with the smallest data file size is selected as the oblique photography information used subsequently;

[0095] During the process of collecting oblique photography information by UAV, the status of the UAV is monitored in real time, and the status of the real-time monitored UAV is analyzed. When the status of the UAV is abnormal, control information is generated to control the UAV and re-collect oblique images;

[0096] The specific process of analyzing the take-off position information of multiple drones and selecting the final take-off position of the drone is as follows: extracting the take-off position information of multiple drones, measuring the distance information between the take-off positions of each drone and the location where the oblique photography three-dimensional model needs to be established, and then selecting the take-off position of the drone with the closest distance as the final take-off position of the drone;

[0097] Through the above process, the take-off position setting of the drone for oblique photography can be better selected, and the impact of the environmental status information of the drone's take-off position on the oblique photography data collection after the drone takes off can be fully considered, further reducing unnecessary data collected by the drone, and then achieving lightweight processing during data collection truncation.

[0098] The process of analyzing the real-time monitoring drone status is as follows: extracting the collected real-time monitoring drone status, the real-time monitoring drone status includes the real-time speed information and the real-time drone shaking amplitude information when the drone is running;

[0099] Extract the real-time speed information of the drone, calculate the difference between it and the standard speed in the current control command, and obtain the speed difference. When the speed difference exceeds the preset range, it means that the drone is in an abnormal state;

[0100] Then extract the real-time UAV shaking amplitude information. When the real-time UAV shaking amplitude information is greater than the preset amplitude, it means that the UAV state is abnormal;

[0101] The abnormality determination process of real-time monitoring of the drone status also includes: real-time monitoring of the data volume of each location collected by the drone, and the estimated data volume of each location. When the data volume of any real-time collection location is greater than the estimated data volume of the location, the drone is determined to be abnormal, and a control quality control drone needs to be generated to return or re-collect the oblique photography data of the location.

[0102] Through the above process, the operating status of the drone for collecting information required for oblique photography can be monitored in real time, and the drone can be controlled when an abnormal status is found, thereby avoiding abnormal data collected due to unstable operating status of the drone, and the need for repeated collection and acquisition resulting in an increase in the amount of data. In addition, it is possible to promptly detect whether the amount of collected data is too much. When it is found that the amount of data is too much, the drone is promptly controlled, thereby ensuring lightweight data.

[0103] The process of obtaining the preliminary model information is as follows: extracting the collected oblique photography information and preliminarily processing the oblique photography information, the preliminary processing includes data preprocessing and data compression;

[0104] The data preprocessing process is as follows: based on the user's actual modeling requirements, unnecessary areas or objects are trimmed;

[0105] Then, data normalization is performed to standardize database design, eliminate duplicate data items, and reduce data redundancy;

[0106] Data deduplication: deduplication of data, deletion of duplicate data items, and further reduction of data redundancy;

[0107] Data consolidation, merging multiple data copies into one to reduce data redundancy and improve storage efficiency;

[0108] The specific process of data compression is as follows: using compression algorithms to reduce redundant data in the model. The compression algorithms used include Quantization and Delta Encoding.

[0109] After data compression is completed, preliminary model information is obtained;

[0110] Data preprocessing, based on application requirements, removes unnecessary areas or objects and focuses on the geographic area of ​​interest;

[0111] The cropping tool can be used to remove unnecessary parts of the scene, such as areas far from the viewpoint or unnecessary details. This processing method can significantly reduce the size and complexity of the model, thereby improving the efficiency of subsequent processing and analysis.

[0112] Data compression: compressing model data through specific algorithms to reduce storage space;

[0113] Data compression is to reduce the redundant information in the data by applying a specific compression algorithm, thereby reducing the size of the data without losing important information;

[0114] Quantization compresses data by reducing the precision or range in the data, while Delta Encoding compresses data by storing only the amount of change in the data. These algorithms can effectively reduce redundant information in the data and improve compression efficiency;

[0115] First, Quantization is used to compress the data to obtain the first compressed data, then DeltaEncoding is used to compress the data to obtain the second compressed data, and then Quantization and DeltaEncoding are used to compress the data to obtain the third compressed data, and then the smallest data size among the first compressed data, the second compressed data and the third compressed data is selected as the preliminary model information;

[0116] Through the above process, preliminary lightweight processing of the oblique photography model was carried out, which effectively reduced the data volume of the oblique photography model.

[0117] The specific processing process of the secondary processing model information is as follows: extract the obtained preliminary model information, perform geometric simplification and texture compression on the preliminary model information;

[0118] The specific process of geometric simplification and texture compression is as follows: the data in the preliminary model information is geometrically simplified through the Quadric Error Metrics and EdgeCollaps algorithms to obtain two simplified data, and the one with the smallest data volume is selected for texture compression. Through the above process, lighter model data can be further selected, thereby ensuring the stability of the final data quantity;

[0119] When performing geometric simplification, it is necessary to comprehensively consider the geometric accuracy and texture quality of the model in order to maintain acceptable visual quality at a smaller file size. This usually requires adjusting the parameters and strategies of the simplification algorithm according to the specific model and application scenario. At the same time, it is also necessary to pay attention to problems such as model distortion and feature loss that may occur during the simplification process, and take appropriate measures to repair and maintain them;

[0120] When compressing textures, you need to choose the appropriate compression algorithm and parameter settings based on the specific model, application scenario, and hardware platform. At the same time, you also need to pay attention to maintaining texture quality to ensure the realism and details of the model rendering. Through a reasonable texture compression strategy, you can maintain acceptable visual quality at a smaller file size, thereby optimizing the performance and loading speed of the model;

[0121] Apply texture compression algorithms to reduce the storage space of texture maps. The texture compression algorithms applied include JPEG, PNG, S3TC / DXTn, ASTC and ETC algorithms;

[0122] After the texture is compressed using JPEG, PNG, S3TC / DXTn, ASTC and ETC algorithms, five texture compression completion data are obtained, and then the texture compression completion data with the highest quality but the smallest data volume is selected from the five texture compression completion data as the secondary processing model information;

[0123] Through the above process, the oblique photography model is further lightweighted to reduce the amount of data and improve processing efficiency, while maintaining sufficient accuracy and reliability to meet the needs of various application scenarios.

[0124] The specific processing process of the final model information is as follows: extract the secondary processing model information to first perform model block and streaming processing, then perform root node merging and index optimization, data compression and streaming transmission, and point cloud-based representation;

[0125] The specific process of model segmentation and streaming processing is as follows:

[0126] First, segment the scene and determine the block size based on actual conditions;

[0127] Boundary processing, setting buffers or overlapping areas between blocks;

[0128] Block management priority sorting: prioritize blocks according to the current display position of the model, and load the blocks at the current display position first;

[0129] Compression and optimization: compress and optimize the geometry and texture of each block, and regularly clean up cache blocks that are no longer in use;

[0130] Benefits of Model Chunking and Streaming

[0131] Reduce initial loading time. By splitting large scenes into multiple smaller tiles and loading them on demand, the initial loading time can be greatly reduced. Users do not need to wait for the entire scene to be fully loaded before they can start interacting with the loaded parts.

[0132] Reduce memory usage. The on-demand loading method allows the system to only load the scene blocks that are currently visible or about to be visible to the user, thereby reducing the demand for memory usage. This not only improves the operating efficiency of the system, but also reduces the risk of performance bottlenecks caused by insufficient memory.

[0133] Improve user experience. By reducing loading time and memory usage, model chunking and streaming technology provide users with a smoother and more efficient experience. Users can enter the scene in a shorter time and start exploring and interacting faster;

[0134] After the compression optimization is completed, the root node is merged and the index is optimized;

[0135] The specific process of root node merging and index optimization is as follows:

[0136] Use a merge tool or algorithm to merge the root nodes into a new root node based on their spatial relationship and similarity, and then use B+ tree index optimization to optimize the index;

[0137] The benefits of root node merging and index optimization are as follows: Improve rendering efficiency. Merging similar nodes or objects can reduce the amount of calculation required for rendering, thereby improving rendering efficiency. At the same time, optimizing the index structure can make data access faster, further improving rendering speed.

[0138] Reduce memory usage. By merging nodes and optimizing indexes, unnecessary memory usage can be reduced, allowing the system to use limited memory resources more efficiently.

[0139] Enhance scene consistency. Root node merging can improve the scene consistency and continuity of the oblique photography 3D model, eliminate the boundaries and breaks between different parts of the model, and make the entire model more coherent and complete.

[0140] After the root nodes are merged and the index is optimized, data compression and streaming are performed. The specific process of data compression and streaming is as follows: data is compressed using the preset compression method, and the corresponding streaming method is selected after the compression is completed;

[0141] After selecting the streaming mode, the compressed data is analyzed and processed to obtain the complexity of all model data, and the data with a complexity greater than a preset value among all model data is selected for point cloud-based representation;

[0142] After all the model data with a complexity greater than a preset value is represented based on the point cloud, the final model information is obtained.

[0143] The benefits of data compression and streaming are as follows: Reduce bandwidth requirements. Using compression algorithms (such as GZIP, LZMA, etc.) can further compress model data, thereby reducing bandwidth requirements during transmission. This is especially important in environments with limited network bandwidth, which can effectively reduce data transmission costs.

[0144] Streaming is implemented. Streaming allows users to start interacting before the data is fully downloaded, greatly improving the user experience. Users can start browsing and using scenes while the data is loading, without having to wait for the entire data set to be fully downloaded.

[0145] Improve security and reliability. Data compression and streaming technologies can also improve the security and reliability of data transmission. Encryption and verification mechanisms can ensure that data is not tampered with or lost during transmission.

[0146] Technologies such as model segmentation and streaming, root node merging and index optimization, data compression and streaming transmission each have their own unique benefits. Together, these technologies improve the processing efficiency, rendering quality and user experience of 3D models. In actual applications, you can choose the appropriate technical solution to optimize performance based on specific needs and scenarios.

[0147] The specific process of model verification and evaluation of the final model information is as follows: perform quality evaluation on the final model information, that is, perform quality evaluation on the lightweight 3D model, and obtain the first evaluation information. The quality evaluation content includes geometric accuracy, texture quality, and detail level.

[0148] Then, the original data and the lightweight three-dimensional model are compared to evaluate the effect of data compression and optimization, and obtain the second evaluation information;

[0149] Finally, the rendering efficiency and user experience are evaluated to obtain the third evaluation information. The evaluation of rendering efficiency and user experience includes loading speed and smoothness of roaming operation.

[0150] When there is no abnormality in the first evaluation information, the second evaluation information and the third evaluation information, it means that the model verification evaluation is normal.

[0151] The process of determining whether the first assessment information is normal is as follows:

[0152] Extract quality assessment content, and obtain geometric accuracy, texture quality and detail level from the quality assessment content;

[0153] When the geometric accuracy, texture quality and level of detail are within the preset required standards, it means that there is no abnormality in the first evaluation information;

[0154] The process of determining whether the second evaluation information is free of abnormalities is as follows: extracting the effects of compression and optimization of the collected data. When the effects of compression and optimization of the data are within the preset standard requirements, it means that the second evaluation information is free of abnormalities.

[0155] The process of judging whether the third evaluation information is normal is as follows: extracting the content for evaluating rendering efficiency and user experience, obtaining the loading speed and the smoothness of roaming operation, and when both the loading speed and the smoothness of roaming operation meet the preset standard requirements, it means that the third evaluation information is normal;

[0156] Through the above process, after the model is lightweighted, the quality of the model is effectively verified again, thereby greatly reducing the situation where the model quality is too low to be used after the oblique photography model is lightweighted.

[0157] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0158] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0159] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A lightweight method for oblique photography three-dimensional model, characterized in that: The following steps are involved: Step 1: Collect relevant data of the oblique photography 3D model to obtain oblique photography information; Step 2: Model the collected oblique photography information to obtain preliminary model information; Step 3: Then, the preliminary model information is subjected to secondary processing to obtain secondary processing model information; Step 4: Process the secondary processing model information to obtain the final model information; Step 5: Perform model validation and evaluation on the final model information. If the model validation and evaluation is normal, the final model information will be exported.

2. The method for lightweighting the oblique photography 3D model according to claim 1, characterized in that: The specific process of obtaining the oblique photography information is as follows: First determine the location where the oblique photography 3D model needs to be established, then collect the take-off position of the drone at the location to obtain the take-off position information of multiple drones; Afterwards, the take-off position information of multiple drones is analyzed, and a final drone take-off position is selected, and the drone for oblique photography is transported to the final drone take-off position to collect oblique photography information, and at least two drones are set to the final drone take-off position to collect oblique photography information; During the process of collecting oblique photography information by the UAV, the status of the UAV is monitored in real time, and the real-time monitored UAV status is analyzed. When the UAV status is abnormal, control information is generated to control the UAV and re-collect oblique images.

3. The method for lightweighting the oblique photography 3D model according to claim 2, characterized in that: The specific process of analyzing the take-off position information of multiple UAVs and selecting the final UAV take-off position is as follows: extract the take-off position information of multiple UAVs, then collect the position-related information of the take-off position information of multiple UAVs, process the position-related information, obtain the position evaluation parameters, analyze the position evaluation parameters, and select the two positions corresponding to the maximum value and the second largest value of the position evaluation parameters as the initial take-off positions, and then evaluate the two initial take-off positions to obtain the real-time evaluation parameters, and select the position corresponding to the minimum value of the two real-time evaluation parameters as the final UAV take-off position.

4. The method for lightweighting the oblique photography 3D model according to claim 2, characterized in that: The process of analyzing the real-time monitoring drone status is as follows: extracting the collected real-time monitoring drone status, where the real-time monitoring drone status includes real-time speed information and real-time drone shaking amplitude information when the drone is running; Extract the real-time speed information of the drone, calculate the difference between it and the standard speed in the current control command, and obtain the speed difference. When the speed difference exceeds the preset range, it means that the drone is in an abnormal state; Then the real-time UAV shaking amplitude information is extracted. When the real-time UAV shaking amplitude information is greater than the preset amplitude, it means that the UAV state is abnormal.

5. The method for lightweighting the oblique photography 3D model according to claim 1, characterized in that: The acquisition process of the preliminary model information is as follows: extracting the collected oblique photography information, and preliminarily processing the oblique photography information, wherein the preliminary processing includes data preprocessing and data compression; The data preprocessing process is as follows: based on the user's actual modeling requirements, unnecessary areas or objects are trimmed; Then, data normalization is performed: normalizing database design, eliminating duplicate data items, and reducing data redundancy; Data deduplication: Deduplication of data is performed to delete duplicate data items to further reduce data redundancy; Data merging: Merge multiple data copies into one to reduce data redundancy and improve storage efficiency; The specific process of data compression is as follows: use compression algorithms to reduce redundant data in the model; After data compression is completed, preliminary model information is obtained.

6. The method for lightweighting the oblique photography 3D model according to claim 1, characterized in that: The specific processing process of the secondary processing model information is as follows: extracting the obtained preliminary model information, and performing geometric simplification and texture compression on the preliminary model information; The specific process of geometric simplification and texture compression is as follows: the data in the preliminary model information is geometrically simplified by using the Quadric Error Metrics and EdgeCollaps algorithms to obtain two simplified data, and the one with the smallest data volume is selected for texture compression; Apply texture compression algorithms to reduce the storage space of texture maps. The texture compression algorithms applied include JPEG, PNG, S3TC / DXTn, ASTC and ETC algorithms; After texture compression is completed using JPEG, PNG, S3TC / DXTn, ASTC and ETC algorithms, five texture compression completion data are obtained. Then, the texture compression completion data with the highest quality but the smallest data volume is selected from the five texture compression completion data as the secondary processing model information.

7. The method for lightweighting the oblique photography 3D model according to claim 6, characterized in that: The specific processing process of the final model information is as follows: extract the secondary processing model information, first perform model block and stream processing, then perform root node merging and index optimization, data compression and streaming transmission and point cloud-based representation; The specific process of model segmentation and streaming processing is as follows: First, segment the scene and determine the block size based on actual conditions; Boundary processing, setting buffers or overlapping areas between blocks; Block management priority sorting: prioritize blocks according to the current display position of the model, and load the blocks at the current display position first; Compression and optimization: compress and optimize the geometry and texture of each block, and regularly clean up cache blocks that are no longer in use; After the compression optimization is completed, the root node is merged and the index is optimized; The specific process of root node merging and index optimization is as follows: Use a merge tool or algorithm to merge the root nodes into a new root node based on their spatial relationship and similarity, and then use B+ tree index optimization to optimize the index; After the root nodes are merged and the index is optimized, data compression and streaming are performed. The specific process of data compression and streaming is as follows: data is compressed using the preset compression method, and after compression is completed, the corresponding streaming method is selected; After selecting the streaming mode, the compressed data is analyzed and processed to obtain the complexity of all model data, and the data with a complexity greater than a preset value among all model data is selected for point cloud-based representation; After all the model data with a complexity greater than a preset value is represented based on the point cloud, the final model information is obtained.

8. The method for lightweighting the oblique photography 3D model according to claim 7, characterized in that: The specific process of model verification and evaluation of the final model information is as follows: perform quality evaluation on the final model information, that is, perform quality evaluation on the lightweight 3D model, and obtain the first evaluation information. The quality evaluation content includes geometric accuracy, texture quality, and detail level. Then, the original data and the lightweight three-dimensional model are compared to evaluate the effect of data compression and optimization, and obtain the second evaluation information; Finally, the rendering efficiency and user experience are evaluated to obtain the third evaluation information. The evaluation of the rendering efficiency and user experience includes the loading speed and the smoothness of the roaming operation. When there is no abnormality in the first evaluation information, the second evaluation information and the third evaluation information, it means that the model verification evaluation is normal.

9. The method for lightweighting the oblique photography 3D model according to claim 8, characterized in that: The process of determining that the first evaluation information is free of abnormalities is as follows: Extract quality assessment content, and obtain geometric accuracy, texture quality and detail level from the quality assessment content; When the geometric accuracy, texture quality and detail level are within the preset requirement standards, it means that there is no abnormality in the first evaluation information; The process of determining whether the second evaluation information is normal is as follows: extracting the effects of compression and optimization of the collected data. When the effects of compression and optimization are within the preset standard requirements, it means that the second evaluation information is normal. The process of determining whether the third evaluation information is normal is as follows: extract the content for evaluating rendering efficiency and user experience, and obtain the loading speed and the smoothness of roaming operations. When the loading speed and the smoothness of roaming operations meet the preset standard requirements, it means that the third evaluation information is normal.

Images