Urban style and feature element identification method and system based on VR depth visual perception

Abstract

The invention belongs to the technical field of urban style and feature recognition, and discloses an urban style and feature element recognition method and system based on VR depth visual perception, and the method comprises the steps: collecting the three-dimensional data and image data of an urban environment in real time through a multi-sensor device; performing data preprocessing to generate a high-quality three-dimensional point cloud and a panoramic image; performing feature extraction on the preprocessed data by adopting a deep learning algorithm, identifying features of urban style and feature elements, and generating feature vectors; utilizing a trained deep neural network model to classify and identify the feature vectors, and determining categories and positions of different urban style and feature elements; a three-dimensional model of urban styles and features is displayed through VR equipment, and a user can perform interaction and exploration in a virtual environment; and performing data analysis on the identification result, and generating a statistical report and a change trend of the urban style and feature elements. According to the method, the problems of low efficiency, insufficient accuracy, data updating lagging and the like in a traditional method are solved, and the method has wide application prospect and practical value.

Description

Technical Field

[0001] The present invention belongs to the technical field of urban landscape recognition, and in particular relates to a method and system for recognizing urban landscape elements based on VR depth visual perception. Background Art

[0002] With the acceleration of urbanization, urban landscape protection and planning have become important issues that need to be addressed urgently. However, traditional urban landscape recognition methods often rely on manual surveys and two-dimensional image analysis, which have problems such as low recognition efficiency, insufficient accuracy, and delayed data updates. Especially in complex urban environments, these methods are difficult to fully and accurately obtain and analyze urban landscape elements. Therefore, there is an urgent need for a solution based on advanced technology to improve the efficiency and accuracy of urban landscape element recognition and support urban planning and smart city construction.

[0003] The technical problems existing in the prior art are as follows:

[0004] 1. Inefficient identification and monitoring methods: Traditional urban landscape recognition methods rely on manual surveys and two-dimensional image analysis, which are inefficient and difficult to meet the needs of real-time monitoring.

[0005] 2. Lack of accuracy: Two-dimensional image analysis is prone to errors when identifying elements in complex urban environments, resulting in low recognition accuracy and failure to fully reflect the city's appearance.

[0006] 3. Data update lag: The data update frequency of traditional methods is low and cannot reflect the changes in the urban landscape in real time, resulting in data lag and affecting the timeliness of urban planning and protection.

[0007] 4. Lack of three-dimensional visualization: Existing technologies mainly display two-dimensional images and lack three-dimensional display methods, which makes it difficult for users to intuitively understand and analyze urban landscape elements. Summary of the invention

[0008] In view of the problems existing in the prior art, the present invention provides a method and system for identifying urban features based on VR depth visual perception.

[0009] The present invention is implemented as follows: a method for identifying urban features based on VR depth visual perception, characterized in that the method specifically includes:

[0010] S1: Data collection, using multi-sensor equipment to collect three-dimensional data and image data of the urban environment in real time, and transmit these data to the data processing center;

[0011] S2: Data preprocessing: preprocessing the collected 3D data and image data to generate high-quality 3D point clouds and panoramic images;

[0012] S3: Feature extraction: Use deep learning algorithms to extract features from preprocessed data, identify the features of urban features, and generate feature vectors;

[0013] S4: Feature recognition, using the trained deep neural network model to classify and identify feature vectors and determine the categories and locations of different urban features;

[0014] S5: VR display, embedding the recognition results into the VR environment, and displaying the three-dimensional model of the city through VR devices, so that users can interact and explore in the virtual environment;

[0015] S6: Data analysis and application: perform data analysis on the identification results to generate statistical reports and change trends of urban landscape elements for decision support in urban planning and landscape protection.

[0016] Furthermore, the S1, multi-sensor device includes a depth camera, a panoramic camera, a LiDAR (laser radar), etc.

[0017] Furthermore, S1 installs or carries the above-mentioned device in an urban environment to collect three-dimensional data and image data of the urban environment in real time; and uses GPS and IMU (inertial measurement unit) to synchronize device data to ensure the temporal and spatial consistency of the data.

[0018] Further, the step S2, data preprocessing, includes:

[0019] Denoising: Use filtering algorithms to remove noise points in the data.

[0020] Alignment: Use the ICP (Iterative Closest Point) algorithm to align data collected at different times or locations.

[0021] Stitching: Stitching point cloud data collected multiple times into a complete 3D point cloud.

[0022] Furthermore, S3, feature extraction, uses a deep learning algorithm (such as CNN, RNN, etc.) to extract features from the preprocessed data; identifies the features of urban landscape elements and generates feature vectors.

[0023] Furthermore, the S4, element recognition, uses the trained deep neural network model to classify and recognize the feature vectors to determine the categories and locations of different urban features elements, which include buildings, roads, green belts, etc.

[0024] Furthermore, the data collection step in S1 further includes:

[0025] The use of multi-sensor equipment including depth cameras, panoramic cameras and LiDAR ensures comprehensive and accurate collection of three-dimensional data and image data of the urban environment under different time periods and weather conditions.

[0026] Furthermore, the feature extraction step in S3 further includes:

[0027] The pre-trained deep learning model ResNet is used for feature extraction. The convolutional neural network (CNN) is used to automatically identify and extract the features of urban landscape elements such as buildings, roads, and green belts to generate feature vectors.

[0028] Furthermore, the element identification step in S4 further includes:

[0029] The trained YOLOv5 deep neural network model is used to perform real-time classification and identification of feature vectors, determine the categories and locations of different urban landscape elements, and generate corresponding classification labels and positioning information.

[0030] Another object of the present invention is to provide a system for identifying urban features based on VR depth visual perception, the system comprising:

[0031] Data acquisition module, used to collect three-dimensional data and image data of the urban environment;

[0032] A data preprocessing module is used to preprocess the collected data;

[0033] Feature extraction module, used to identify the features of urban features and generate feature vectors;

[0034] Feature recognition module, used to classify and identify feature vectors;

[0035] VR display module, used to display the three-dimensional model of the city landscape through VR equipment;

[0036] The data analysis and application module is used to perform data analysis on the recognition results.

[0037] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0038] 1. High efficiency: Utilizing multi-sensor equipment and deep visual perception technology, real-time collection and processing of urban features are achieved, greatly improving recognition efficiency.

[0039] 2. High accuracy: The deep learning algorithm is used to extract and classify the features of urban landscape elements, which significantly improves the accuracy and precision of recognition.

[0040] 3. Three-dimensional visualization: VR technology is used to present the recognition results as a three-dimensional model. Users can observe and interact intuitively in the virtual environment, which enhances the effect of data display.

[0041] 4. Decision support: The generated statistical reports on urban landscape elements and change trend analysis provide scientific data support for urban planning and landscape protection, and improve the rationality and operability of decision-making.

[0042] 5. Timely data update: The system can update urban landscape data in real time, keep the latest status of urban landscape element identification, and avoid problems caused by data lag.

[0043] Significant progress in industrial applications

[0044] 1. Improve urban planning efficiency: The efficient identification and real-time update functions of the present invention enable urban planning departments to quickly obtain the latest urban landscape data, formulate and adjust urban planning plans in a timely manner, and improve planning efficiency and scientificity.

[0045] 2. Enhance the ability to protect urban appearance: Through high-accuracy identification and real-time monitoring, the system can promptly detect and report changes and potential problems in urban appearance, provide scientific data support for the appearance protection department, and enhance the ability to protect urban appearance.

[0046] 3. Support the construction of smart cities: The application of 3D visualization technology makes the display and analysis of urban features more intuitive, supports various applications in smart city construction, such as smart transportation, smart security, smart management, etc., and promotes the efficient construction and management of smart cities.

[0047] 4. Improve the level of decision-making support: Through detailed statistical reports and trend analysis, the system provides comprehensive data support for urban planning and landscape protection decisions, improves the rationality and operability of decisions, and promotes the realization of scientific decision-making.

[0048] In summary, the present invention provides an efficient and accurate method and system for identifying urban features by combining VR technology and deep visual perception technology, which solves the problems of low efficiency, insufficient accuracy and delayed data update in traditional methods, and has broad application prospects and practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of a method for identifying urban features based on VR depth visual perception provided by an embodiment of the present invention;

[0050] Figure 2 is a data preprocessing flow chart provided by an embodiment of the present invention;

[0051] Figure 3 It is a module diagram of a city features element recognition system based on VR depth visual perception provided by an embodiment of the present invention;

[0052] Figure 4 It is a comparison effect diagram between the embodiment of the present invention and the prior art. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0054] The following are two specific embodiments of the urban landscape element recognition system based on VR depth visual perception:

[0055] Example 1: Digitalization and VR display of historical block features

[0056] Application scenarios: Digitally archive the appearance of historical blocks and use VR technology for immersive display for cultural heritage protection, tourism promotion, etc.

[0057] System composition:

[0058] Data collection equipment: mobile lidar scanner, panoramic camera.

[0059] Data processing software: point cloud processing software (such as CloudCompare, PolyWorks), image processing software (such as Photoshop), and 3D modeling software (such as SketchUp, 3ds Max).

[0060] VR display equipment: VR helmet, VR interactive equipment.

[0061] Implementation steps:

[0062] Data collection: Use a mobile LiDAR scanner to perform a three-dimensional scan of the historical district to obtain point cloud data of buildings, streets, landscapes, etc. At the same time, use a panoramic camera to capture panoramic images of the district.

[0063] Data processing: perform pre-processing such as filtering, segmentation, and simplification on the point cloud data, and align and fuse it with the panoramic image to generate a realistic three-dimensional model.

[0064] Feature recognition: Extract geometric features (such as height, volume, roof form) and semantic features (such as architectural style and age) of buildings to identify buildings with historical value.

[0065] VR scene construction: Import the identified historical building models into the VR engine, add textures and materials, and build VR scenes of historical blocks.

[0066] VR display: Users wearing VR helmets can roam freely in the VR scene, observe the details of historical buildings up close, and understand their historical background and cultural value.

[0067] Example 2: VR evaluation of urban planning schemes

[0068] Application scenarios: During the urban planning stage, the planning scheme is converted into a VR scene, allowing planners, decision makers, the public and other participants to conduct immersive experience and evaluation.

[0069] System composition:

[0070] Data source: urban planning design drawings, CAD models, and BIM models.

[0071] Data processing software: 3D modeling software (such as SketchUp, 3ds Max), VR engine (such as Unity, UnrealEngine).

[0072] VR display equipment: VR helmets, VR interactive equipment, large-screen VR display system.

[0073] Implementation steps:

[0074] Model conversion: Convert planning and design drawings, CAD models, and BIM models into a format recognizable by the VR engine.

[0075] VR scene construction: Build the VR scene of the planned area in the VR engine, including buildings, roads, green spaces, public facilities and other elements.

[0076] Interaction design: Design user interaction methods to enable users to roam freely in the VR scene, switch perspectives, adjust time and weather, etc.

[0077] VR evaluation: Invite planners, decision makers, the public and other participants to experience VR scenarios, collect their feedback, and evaluate the rationality, feasibility, and aesthetics of the planning scheme.

[0078] Scheme optimization: Based on the evaluation results, the planning scheme is optimized and adjusted to improve the scientific nature of the planning and public participation.

[0079] These two examples demonstrate the application potential of the urban landscape element recognition system based on VR deep visual perception in different fields. With the continuous development of technology, the system will play an increasingly important role in urban planning, cultural heritage protection, smart city construction and other fields.

[0080] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying urban features based on VR depth visual perception, and the method specifically includes:

[0081] S1: Data collection, using multi-sensor equipment to collect three-dimensional data and image data of the urban environment in real time, and transmit these data to the data processing center;

[0082] S2: Data preprocessing: preprocessing the collected 3D data and image data to generate high-quality 3D point clouds and panoramic images;

[0083] S3: Feature extraction: Use deep learning algorithms to extract features from preprocessed data, identify the features of urban features, and generate feature vectors;

[0084] S4: Feature recognition, using the trained deep neural network model to classify and identify feature vectors and determine the categories and locations of different urban features;

[0085] S5: VR display, embedding the recognition results into the VR environment, and displaying the three-dimensional model of the city through VR devices, so that users can interact and explore in the virtual environment;

[0086] S6: Data analysis and application: perform data analysis on the identification results to generate statistical reports and change trends of urban landscape elements for decision support in urban planning and landscape protection.

[0087] In S1, the multi-sensor equipment includes a depth camera, a panoramic camera, a LiDAR (laser radar), etc. The acquisition method is:

[0088] The above-mentioned equipment is installed or carried in the urban environment to collect three-dimensional data and image data of the urban environment in real time.

[0089] Synchronize device data using GPS and IMU (Inertial Measurement Unit) to ensure data consistency in space and time.

[0090]

[0091] like Figure 2 As shown, the step S2, data preprocessing includes:

[0092] Denoising: Use filtering algorithms to remove noise points in the data.

[0093] Alignment: Use the ICP (Iterative Closest Point) algorithm to align data collected at different times or locations.

[0094] Stitching: Stitching point cloud data collected multiple times into a complete 3D point cloud.

[0095]

[0096] The step S3, feature extraction comprises:

[0097] Use deep learning algorithms (such as CNN, RNN, etc.) to extract features from the preprocessed data.

[0098] Identify the characteristics of urban landscape elements and generate feature vectors.

[0099]

[0100]

[0101] Convert point cloud data into model input format

[0102] Here we assume that there is already a function point_cloud_to_image() that implements the conversion from point cloud to image

[0103]

[0104] The S4, element recognition, uses the trained deep neural network model to classify and recognize feature vectors to determine the categories and locations of different urban features elements, which include buildings, roads, green belts, etc.

[0105]

[0106]

[0107] S5, VR display, embeds the recognition result into the VR environment, and displays the three-dimensional model of the city through the VR device. Users can interact and explore in the virtual environment.

[0108]

[0109] S6, data analysis and application, performs data analysis on the recognition results to generate statistical reports and change trends of urban features elements, which are used for decision support of urban planning and landscape protection.

[0110]

[0111]

[0112] like Figure 3 As shown, an embodiment of the present invention provides a system for identifying urban features based on VR depth visual perception, the system comprising:

[0113] Data acquisition module, used to collect three-dimensional data and image data of the urban environment;

[0114] A data preprocessing module is used to preprocess the collected data;

[0115] Feature extraction module, used to identify the features of urban features and generate feature vectors;

[0116] Feature recognition module, used to classify and identify feature vectors;

[0117] VR display module, used to display the three-dimensional model of the city landscape through VR equipment;

[0118] The data analysis and application module is used to perform data analysis on the recognition results.

[0119] The working principle of the urban landscape element recognition system based on VR deep visual perception.

[0120] Depth sensor: The core of the system is the depth sensor (such as lidar, depth camera), which can capture three-dimensional point cloud data of the urban environment and record information such as the spatial coordinates (X, Y, Z) and reflection intensity of each point.

[0121] Image sensor: Works with the depth sensor to obtain high-resolution color images of the urban environment, providing a basis for subsequent texture mapping and visual recognition.

[0122] Data fusion: align and fuse depth data and image data, associate depth information and color information together to form a realistic three-dimensional point cloud model.

[0123] Point cloud filtering: remove noise points and outliers, smooth the point cloud surface, and improve data quality and reliability.

[0124] Point cloud segmentation: Segment the point cloud data into different areas, such as buildings, roads, vegetation, sky, etc., to prepare for subsequent feature extraction.

[0125] Point cloud simplification: Reduce the amount of point cloud data and improve the efficiency of subsequent processing while maintaining the main features of the model.

[0126] Geometric features: Extract geometric features of buildings such as height, volume, orientation, roof shape, facade texture, etc.

[0127] Semantic features: Use deep learning models (such as convolutional neural networks) to perform semantic segmentation on image data and identify the functional type of buildings (such as residential, commercial, office), road types, vegetation types, etc.

[0128] Feature fusion: The geometric features and semantic features are integrated to form a comprehensive feature vector that fully describes the attributes of urban landscape elements.

[0129] Classifier training: Use labeled urban landscape feature datasets to train classifiers (such as support vector machines, random forests) or deep learning models.

[0130] Feature classification: Input the feature vector into the classifier to obtain the category label of each feature (such as building, road, vegetation).

[0131] Feature identification: Based on the classification results and preset rules, identify features with specific features, such as historical buildings, landmark buildings, green parks, etc.

[0132] Model optimization: Optimize the three-dimensional model of the identified urban features, add textures and materials, and enhance the visual effect.

[0133] VR scene construction: Import the optimized model into the VR engine to build an immersive virtual city environment.

[0134] Interaction design: Design user interaction methods to enable users to freely roam, observe, and query information about urban features in a VR environment.

[0135] Statistical analysis: Conduct statistical analysis on the recognition results and calculate indicators such as the number, distribution, and density of different types of elements.

[0136] Spatial analysis: Analyze the spatial relationship between elements, such as the distance between buildings, road connectivity, green space coverage, etc.

[0137] Application scenarios: Apply the analysis results to urban planning, design, management and other fields to provide support for decision-making. For example, a protection plan can be formulated based on the distribution of historical buildings; the ecological environment quality of the city can be evaluated based on the coverage rate of green space.

[0138] The embodiment of the present invention provides a system for identifying urban features based on VR deep visual perception, which acquires three-dimensional data and image data of the urban environment through depth sensors and image sensors, and processes and analyzes the data using technologies such as deep learning, thereby realizing automatic identification and VR display of urban features. The system has the advantages of high precision, high efficiency, and wide application range, and can provide strong technical support for urban planning, design, management and other fields.

[0140] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0141] Example 1: Historical district protection based on VR deep visual perception

[0142] VR deep visual perception technology is used to identify and monitor urban landscape elements such as buildings, roads, green belts in historical blocks, helping relevant departments to formulate protection and restoration plans.

[0143] Implementation steps:

[0144] 1. Data collection:

[0145] Equipment: Multi-sensor equipment installed in the block, including depth cameras, panoramic cameras and LiDAR.

[0146] Methods: Three-dimensional data and image data of the block were collected under different time periods and weather conditions.

[0147] Transmission: Data is transmitted to the data processing center in real time for processing.

[0148] 2. Data preprocessing:

[0149] Denoising: Use statistical filtering to remove noise points in the collected data.

[0150] Alignment and stitching: Use the ICP algorithm to align and stitch data collected multiple times to generate high-quality 3D point clouds and panoramic images.

[0151] 3. Feature extraction:

[0152] Algorithm: Use the pre-trained ResNet model for feature extraction.

[0153] Feature vector: Extract feature vectors of urban features such as buildings, roads, and green belts.

[0154] 4. Element identification:

[0155] Model: Use the trained deep neural network model to classify and identify feature vectors to determine the category and location of different elements.

[0156] Result: Identify historical buildings, modern buildings, roads, green belts and other elements within the block.

[0157] 5. VR display:

[0158] Create a 3D model: embed the recognition results into the VR environment to generate a 3D model of the block.

[0159] Interaction: By displaying the three-dimensional model through VR devices, users can interact and explore in the virtual environment and observe the details of different elements.

[0160] 6. Data analysis and application:

[0161] Report: Generate statistical reports on the style and features of historical districts and analyze the status and changing trends of buildings.

[0162] Application: Help conservation departments develop protection and restoration plans to prevent damage to historic buildings.

[0163] Implementation effect:

[0164] Efficiency: Real-time data collection and processing greatly improves recognition efficiency.

[0165] Accuracy: Deep learning algorithms improve recognition accuracy, ensuring the preservation of historic buildings.

[0166] Three-dimensional visualization: VR technology provides intuitive display and interaction methods to enhance user experience.

[0167] Decision support: Detailed reports and analysis results support scientific decision-making and protect the appearance of historical districts.

[0168] Example 2: Smart city construction based on VR deep visual perception

[0169] VR deep visual perception technology is used to identify and monitor buildings, roads, public facilities, etc. in new urban areas to support the planning and construction of smart cities.

[0170] Implementation steps:

[0171] 1. Data collection:

[0172] Equipment: Multi-sensor equipment deployed in new urban areas, including depth cameras, panoramic cameras and LiDAR.

[0173] Methods: 3D data and image data of the new area were collected regularly, covering all stages of construction.

[0174] Transmission: The collected data is transmitted to the data processing center for processing.

[0175] 2. Data preprocessing:

[0176] Denoising: Use Gaussian filtering to remove noise points in the data.

[0177] Alignment and stitching: Use a multi-view stereo matching algorithm to align and stitch data collected multiple times to generate high-quality 3D point clouds and panoramic images.

[0178] 3. Feature extraction:

[0179] Algorithm: YOLOv5 model is used for feature extraction.

[0180] Feature vector: Extract feature vectors of urban features such as buildings, roads, and public facilities.

[0181] 4. Element identification:

[0182] Model: Use the trained deep neural network model to classify and identify feature vectors to determine the category and location of different elements.

[0183] Result: Identify elements such as buildings under construction, completed buildings, roads, and public facilities in the new area.

[0184] 5. VR display:

[0185] Create a 3D model: embed the recognition results into the VR environment to generate a 3D model of the new area.

[0186] Interaction: By displaying the three-dimensional model through VR devices, users can interact and explore in the virtual environment and observe the details of different elements.

[0187] 6. Data analysis and application:

[0188] Report: Generate a statistical report on the new district’s landscape elements and analyze construction progress and changing trends.

[0189] Application: Support the planning and construction of smart cities and optimize resource allocation and management.

[0190] Implementation effect:

[0191] Efficiency: Real-time data collection and processing to improve the monitoring efficiency of new district construction.

[0192] Accuracy: Deep learning algorithms improve recognition accuracy, ensuring precise monitoring of the construction process.

[0193] Three-dimensional visualization: VR technology provides intuitive display and interaction methods to enhance user experience.

[0194] Decision support: Detailed reports and analysis results support scientific decision-making and promote the efficient construction of smart cities.

[0195] Data, charts and comparisons:

[0196] Data comparison table:

[0197]

[0198] Contrast effect like columnar Figure 4 shown.

[0199] The above bar chart shows the comparison between the traditional method and the method of the present invention in terms of data collection efficiency, recognition accuracy and user experience.

[0200] Data comparison summary:

[0201] Data collection efficiency: The method of the present invention is significantly higher than the traditional method, reflecting the high efficiency of data collection through VR deep visual perception technology.

[0202] Recognition accuracy: The recognition accuracy of the method of the present invention significantly exceeds that of traditional methods, indicating the advantages of deep learning algorithms in the recognition process.

[0203] User experience: The method of the present invention provides a better user experience, thanks to the intuitive display and interaction provided by VR technology.

[0204] Specific advantages:

[0205] 1. High efficiency: The method of the present invention uses multi-sensor equipment and automated processing in the data collection process, thereby improving the overall efficiency.

[0206] 2. Accuracy: Through deep learning algorithms and high-precision sensors, the method of the present invention can more accurately identify and classify different elements.

[0207] 3. User experience: VR technology enhances user interactivity and immersion, enabling users to observe and analyze data more intuitively.

[0208] These advantages provide strong technical support for the protection of historical blocks and the construction of smart cities, and promote the development of related fields. Through the detailed description of the above embodiments and technical effects, the present invention has made significant breakthroughs and advantages in VR depth visual perception technology, which can greatly improve the efficiency and accuracy of historical block protection and smart city construction, and provide strong technical support and application prospects for related fields.

[0209] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for identifying urban features based on VR depth visual perception, characterized in that: The method specifically includes: S1: Data collection, using multi-sensor equipment to collect three-dimensional data and image data of the urban environment in real time, and transmit these data to the data processing center; S2: Data preprocessing: preprocessing the collected 3D data and image data to generate high-quality 3D point clouds and panoramic images; S3: Feature extraction: Use deep learning algorithms to extract features from preprocessed data, identify the features of urban features, and generate feature vectors; S4: Feature recognition, using the trained deep neural network model to classify and identify feature vectors and determine the categories and locations of different urban features; S5: VR display, embedding the recognition results into the VR environment, and displaying the three-dimensional model of the city through VR devices, so that users can interact and explore in the virtual environment; S6: Data analysis and application: perform data analysis on the identification results to generate statistical reports and change trends of urban landscape elements for decision support in urban planning and landscape protection.

2. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: The data collection step in S1 further includes: The use of multi-sensor equipment including depth cameras, panoramic cameras and LiDAR ensures comprehensive and accurate collection of three-dimensional data and image data of the urban environment under different time periods and weather conditions.

3. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: The feature extraction step in S3 further includes: The pre-trained deep learning model ResNet is used for feature extraction. The convolutional neural network (CNN) is used to automatically identify and extract the features of urban landscape elements such as buildings, roads, and green belts to generate feature vectors.

4. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: The element identification step in S4 further includes: The trained YOLOv5 deep neural network model is used to perform real-time classification and identification of feature vectors, determine the categories and locations of different urban landscape elements, and generate corresponding classification labels and positioning information.

5. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: The S1, multi-sensor device includes a depth camera, a panoramic camera, and LiDAR (laser radar).

6. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: Said S1, installs or carries the above-mentioned device in an urban environment, collects three-dimensional data and image data of the urban environment in real time; uses GPS and IMU (inertial measurement unit) to synchronize device data to ensure the temporal and spatial consistency of the data.

7. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: The step S2, data preprocessing, includes: Denoising: Use filtering algorithms to remove noise points in the data; Alignment: Use the ICP (Iterative Closest Point) algorithm to align data collected at different times or locations; Stitching: Stitching point cloud data collected multiple times into a complete 3D point cloud.

8. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: The S3, feature extraction, uses a deep learning algorithm (such as CNN, RNN, etc.) to extract features from the preprocessed data; identifies the features of urban landscape elements and generates feature vectors.

9. The method for identifying urban features based on VR depth visual perception according to claim 1 is characterized in that: The S4, element recognition, uses the trained deep neural network model to classify and recognize feature vectors to determine the categories and locations of different urban features elements, which include buildings, roads, green belts, etc.

10. A system for identifying urban features based on VR depth visual perception as claimed in claims 1-9, characterized in that: The system includes: Data acquisition module, used to collect three-dimensional data and image data of the urban environment; A data preprocessing module is used to preprocess the collected data; Feature extraction module, used to identify the features of urban features and generate feature vectors; Feature recognition module, used to classify and identify feature vectors; VR display module, used to display the three-dimensional model of the city landscape through VR equipment; The data analysis and application module is used to perform data analysis on the recognition results.

Images