Precise correction monitoring method based on VR panoramic position and map coordinates

Abstract

The invention discloses an accurate correction monitoring method based on a VR panoramic position and map coordinates, which belongs to the field of monitoring calibration methods, and comprises the following steps: constructing a high-precision three-dimensional geographic information model comprising a building contour, a ground marker and a high-altitude suspension through multi-modal data fusion of a laser radar three-dimensional point cloud and a VR panoramic image; meanwhile, a multi-exposure synthesis technology is adopted, dynamic object interference is effectively eliminated, image data quality is improved, multi-channel ResNet50 coding of visible light texture features and infrared heat source features is combined, and high-precision geometric feature matching is achieved through cosine similarity matching and Delaunay triangulation verification; a grading early warning mechanism is set to be a three-level early warning mechanism, differential safety strategies can be achieved, and the accuracy of early warning mechanism triggering conditions and the distinction of triggering execution mechanisms are improved.

Description

Technical Field

[0001] The present invention relates to the field of monitoring calibration methods, and more specifically, to a precise correction monitoring method based on VR panoramic position and map coordinates. Background Art

[0002] With the advancement of the construction of smart cities, traditional monitoring methods face multi-dimensional technical bottlenecks. For example, existing monitoring methods mostly rely on single-sensor data, resulting in local missing or insufficient accuracy in environmental modeling. Especially in complex scenarios, it is difficult to completely capture the spatial relationships of building outlines, ground markers, and high-altitude hanging objects. In addition, geometric feature matching algorithm technology is also used in related technologies of traditional monitoring methods, but its geometric feature matching algorithm technology is prone to failure in complex lighting or texture-lacking scenarios and is difficult to meet the high-precision positioning requirements. In terms of the early warning mechanism, existing monitoring methods generally adopt a single-threshold trigger strategy, with rough early warning grading and a high false alarm rate, making it difficult to adapt to differentiated security requirements.

[0003] Based on the above problems, there is an urgent need for a monitoring method that can fuse multi-modal data to improve the accuracy of constructing markers, effectively eliminate the interference of dynamic objects, improve the quality of image data, and have a multi-level threshold trigger strategy to enhance the accuracy of the early warning mechanism trigger conditions and the discrimination of the trigger execution mechanism. Therefore, we propose a precise correction monitoring method based on VR panoramic position and map coordinates to solve the above existing technical problems. Summary of the Invention

[0004] 1. Technical Problems to be Solved

[0005] Aiming at the problems existing in the prior art, the purpose of the present invention is to provide a precise correction monitoring method based on VR panoramic position and map coordinates. It fuses multi-modal data of lidar three-dimensional point clouds and VR panoramic images to construct a high-precision three-dimensional geographic information model including building outlines, ground markers, and high-altitude hanging objects, improving the integrity of environmental modeling. At the same time, it adopts a multiple-exposure synthesis technology to effectively eliminate the interference of dynamic objects and improve the quality of image data. Combining multi-channel ResNet50 encoding of visible light texture features and infrared heat source features, through cosine similarity matching and Delaunay triangulation verification, high-precision geometric feature matching is achieved. In addition, the hierarchical early warning mechanism is set as a three-level early warning mechanism, which can implement differentiated security strategies, enhance the accuracy of the early warning mechanism trigger conditions and the discrimination of the trigger execution mechanism. Through the implementation of encrypted storage on the blockchain storage platform in the three-level early warning, the non-tamperable characteristic of the location information is effectively ensured. These technical breakthroughs fill the existing market gaps and provide a new technical paradigm for the safety management of smart cities.

[0006] 2. Technical Solutions

[0007] To solve the above problems, the present invention adopts the following technical solutions.

[0008] A precise correction monitoring method based on VR panoramic position and map coordinates, comprising the following steps:

[0009] S1. Benchmark construction stage:

[0010] Using a lidar to scan the target area to obtain three-dimensional point cloud data, synchronously collecting VR panoramic images and performing multiple exposure synthesis to eliminate the interference of dynamic objects;

[0011] Fusing the three-dimensional point cloud data and VR panoramic images, constructing a three-dimensional geographic information model including building contour lines, ground markers and high-altitude suspended objects, and generating a spatial mapping relationship library, where the spatial mapping relationship library is a set of feature points generated by the three-dimensional geographic information model, and each feature point includes material reflectivity, thermal radiation coefficient and geometric invariant moment;

[0012] S2. Real-time positioning stage:

[0013] Using a monitoring terminal to collect pose data and real-time images, and extracting visible light texture features and infrared heat source distribution features in the real-time images;

[0014] S3. Precise matching stage:

[0015] Encoding the visible light texture features and infrared heat source distribution features into feature vectors, performing similarity matching with the environmental feature vectors in the spatial mapping relationship library, and eliminating invalid matching points through geometric verification, and calculating the spatial position offset;

[0016] S4. Dynamic correction stage:

[0017] Constructing a filtering model based on sensor data, real-time correcting the current offset, and predicting the future offset trend;

[0018] Fusing the filtering result and the prediction result to generate dynamic compensation parameters, and driving the actuator to perform positioning correction;

[0019] S5. Cooperative verification stage:

[0020] Using the communication network between devices to screen benchmark nodes, constructing a spatio-temporal grid and assigning dynamic weight coefficients, sending an encrypted verification request to low-confidence terminals, and generating a cooperative positioning report;

[0021] S6. Security response stage:

[0022] Triggering a hierarchical warning mechanism according to the offset, and performing corresponding security operations, including alarm log recording, permission restriction and encrypted location information storage.

[0023] Further, the exposure time of the multiple exposure synthesis in step S1 is dynamically adjusted according to the illumination intensity of the target area, and by default includes 10 ms, 30 ms, and 50 ms. The feature point density of the spatial mapping relationship library is 120 points / m in the core area 2 and 15 points / m in the edge area 2 , and each feature point is additionally included with an environmental feature vector containing the material reflectivity, thermal radiation coefficient, and geometric invariant moment.

[0024] Further, in step S3, the visible light texture feature and the infrared heat source distribution feature are encoded by using a pre-trained ResNet50 convolutional neural network, then the feature vectors are matched by cosine similarity, and the geometric rationality is verified by Delaunay triangulation, where the side length error tolerance range is ±5 cm. The input of the ResNet50 network is the RGB-IR multi-channel data of the real-time image.

[0025] Further, in step S4, the filtering model is the extended Kalman filter, and its state variables include the terminal motion acceleration, the environmental illumination intensity, and the electromagnetic interference level. The LSTM network is used to predict the future offset trend, and the input data includes the historical offset sequence and the real-time wind speed data. The historical offset sequence is extracted by the sliding window method, and the window length is 10 seconds. The dynamic compensation parameter fuses the filtering result and the prediction result through the fuzzy control algorithm, where the weight of the filtering result α = 0.6 and the weight of the prediction result β = 0.4.

[0026] Further, in step S5, the improved RANSAC algorithm is used to screen the reference nodes, and the inertial navigation weighting factor γ = 0.3 is added. The encrypted verification request uses the AES-256 algorithm and the differential privacy technology of the Laplace noise mechanism (ε ＝ 0 . 1).

[0027] Further, in step S6, the hierarchical early warning mechanism is a three-level early warning mechanism, and the triggering conditions of the three-level early warning mechanism are:

[0028] Level 1 early warning: The offset < 0.5 m, triggering the warning icon and log record;

[0029] Level 2 early warning: 0.5 m ≤ offset < 2 m, starting the buzzer alarm and restricting the operation authority;

[0030] Level 3 early warning: The offset ≥ 2 m, switching to the Beidou / GPS dual-mode positioning module, and encrypting and storing evidence through the blockchain storage platform. The data encrypted and stored by the blockchain storage platform includes the offset, the timestamp, and the terminal device ID, and multi-party verification is realized through the smart contract.

[0031] A monitoring system for precise correction based on VR panoramic position and map coordinates, comprising:

[0032] Data acquisition module: A wearable device integrating a lidar, a multispectral camera, and an IMU sensor;

[0033] Cooperative processing unit: A heterogeneous computing platform composed of an FPGA and a GPU, where the FPGA optimizes convolutional computing, and the GPU runs an LSTM network;

[0034] Correction actuator: Comprising a piezoelectric ceramic micro-adjuster and a servo motor;

[0035] Secure communication module: A Mesh communication device supporting national cryptographic algorithms and a blockchain evidence storage unit.

[0036] Further, the lidar is a 64-line lidar with a scanning frequency ≥ 20 Hz, the multispectral camera supports visible light and near-infrared bands with a frame rate ≥ 30 fps, the resolution of the piezoelectric ceramic micro-adjuster ≤ 0.1 μm, and the response time of the servo motor ≤ 5 ms.

[0037] Further, the FPGA optimizes convolutional computing using the Winograd algorithm with a processing speed ≥ 130 frames per second, the evidence storage interval of the blockchain evidence storage unit is 1 second, and the block hash value is generated through the SHA-256 algorithm.

[0038] 3. Beneficial effects

[0039] Compared with the prior art, the advantages of the present invention are as follows:

[0040] (1) In this solution, through the multi-modal data fusion of lidar three-dimensional point cloud and VR panoramic image, a high-precision three-dimensional geographic information model including building outlines, ground markers, and high-altitude hanging objects is constructed, improving the integrity of environmental modeling. Moreover, the multiple exposure synthesis technology is adopted to effectively eliminate the interference of dynamic objects, improve the quality of image data. At the same time, through the multi-channel ResNet50 encoding combining visible light texture features and infrared heat source features, and through cosine similarity matching and Delaunay triangulation verification, high-precision geometric feature matching is achieved;

[0041] (2) In this solution, the hierarchical early warning mechanism is set as a three-level early warning mechanism, which can implement differential security strategies, improve the accuracy of the early warning mechanism triggering conditions and the distinguishability of the triggering execution mechanism, facilitating the precise execution of corresponding early warning operations. Moreover, through the implementation of encrypted evidence storage on the blockchain evidence storage platform in the three-level early warning, the non-tamperable property of the location information is effectively ensured. Description of the drawings

[0042] Figure 1Schematic diagram of the steps of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention;

[0043] Figure 2 Mind map of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention Figure 1 ;

[0044] Figure 3 Mind map of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention Figure 2 ;

[0045] Figure 4 Mind map of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention Figure 3 ;

[0046] Figure 5 Mind map of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention Figure 4 ;

[0047] Figure 6 Mind map of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention Figure 5 ;

[0048] Figure 7 Mind map of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention Figure 6 ;

[0049] Figure 8 Mind map of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention Figure 7 ;

[0050] Figure 9 Schematic diagram of the system architecture of a precise correction monitoring method based on VR panoramic position and map coordinates according to the present invention. Detailed implementation manners

[0051] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention specification; obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0052] Embodiment 1:

[0053] Implementation of the precise correction monitoring method based on VR panoramic position and map coordinates

[0054] Please refer to the accompanying drawings of the specification Figures 1 - 8, this embodiment details the implementation process of the method described in claims 1-6, and the specific steps are as follows:

[0055] Step S1: Benchmark construction stage

[0056] Use a 64-line lidar (scanning frequency 20Hz) to collect three-dimensional point cloud data of the target area, and simultaneously use a multispectral camera (supporting visible light and near-infrared bands, frame rate 30fps) to capture VR panoramic images. Dynamically adjust the multiple exposure synthesis parameters according to the illumination intensity of the target area, with the default exposure times being 10ms, 30ms, and 50ms to eliminate interference from dynamic objects. Integrate the three-dimensional point cloud data and the VR panoramic images to construct a three-dimensional geographic information model containing building contour lines, ground markers, and high-altitude hanging objects. During the model generation process, set the feature point density to 120 points / m in the core area 2 , and 15 points / m in the edge area 2 , and attach an environmental feature vector of material reflectivity (obtained through spectral analysis), thermal radiation coefficient (measured by an infrared sensor), and geometric invariant moment (calculated based on point cloud curvature) to each feature point, finally forming a spatial mapping relationship library.

[0057] Step S2: Real-time positioning stage

[0058] The monitoring terminal (integrated with an IMU sensor and a multispectral camera) continuously collects pose data (including acceleration and angular velocity) and RGB-IR multi-channel images. Extract visible light texture features (such as SIFT feature points) and infrared heat source distribution features (based on temperature threshold segmentation) from the real-time images through image processing algorithms.

[0059] Step S3: Precise matching stage

[0060] Input the extracted visible light texture features and infrared heat source distribution features into a pre-trained ResNet50 convolutional neural network (the input layer is adapted to RGB-IR four-channel data), and output a 512-dimensional feature vector. Match the real-time feature vector with the environmental feature vectors in the spatial mapping relationship library through the cosine similarity algorithm, and screen out feature point pairs with a similarity greater than 0.85. Further use Delaunay triangulation to verify the geometric rationality, eliminate invalid matching points with side length errors exceeding ±5cm, and calculate the spatial position offset.

[0061] Step S4: Dynamic correction stage

[0062] Build an Extended Kalman Filter model (state variables include terminal acceleration, ambient light intensity, and electromagnetic interference level), fuse IMU data with matching results, and correct the current offset in real time. Meanwhile, use an LSTM network (input historical offset sequences and real-time wind speed data, with a sliding window length of 10 seconds) to predict future offset trends. Generate dynamic compensation parameters through a fuzzy control algorithm (filtering result weight α = 0.6, prediction result weight β = 0.4), and drive a piezoelectric ceramic micro-adjuster (resolution 0.1μm) and a servo motor (response time 5ms) to perform positioning correction.

[0063] Step S5: Collaborative verification stage

[0064] Devices screen reference nodes through a Mesh communication network supporting national cryptographic algorithms, and use an improved RANSAC algorithm (add an inertial navigation weighting factor γ = 0.3) to construct a spatio-temporal grid. Send a verification request encrypted by AES-256 to low-confidence terminals, and inject Laplace noise (ε = 0.1) to achieve differential privacy protection. Finally, generate a collaborative positioning report containing multi-node consistency verification.

[0065] Step S6: Security response stage

[0066] Trigger a three-level early warning mechanism according to the offset:

[0067] First-level early warning (offset < 0.5m): The warning icon is displayed on the system interface, and the log is recorded in the local database;

[0068] Second-level early warning (0.5m ≤ offset < 2m): Start the buzzer alarm and restrict the terminal operation authority

[0069] Third-level early warning (offset ≥ 2m): Switch to the Beidou / GPS dual-mode positioning module, and encrypt and store the offset, timestamp, and device ID through a blockchain evidence storage platform (evidence storage interval 1 second, block hash value generated by SHA-256), and the smart contract triggers a multi-party verification process.

[0070] Embodiment 2:

[0071] Implementation of the monitoring system

[0072] Please refer to the accompanying drawings of the specification Figures 1 - 9 This embodiment corresponds to the system described in claims 7-9, and the specific implementation is as follows:

[0073] Data acquisition module: The wearable device integrates a 64-line lidar (scanning frequency 20Hz), a multi-spectral camera (frame rate 30fps), and an IMU sensor (sampling rate 100Hz) to synchronously collect 3D point clouds, multi-spectral images, and pose data of the target area in real time.

[0074] Collaborative Processing Unit: An heterogeneous computing platform composed of an FPGA (optimizing convolutional calculations with the Winograd algorithm, processing speed of 130 frames per second) and a GPU (running an LSTM network). The FPGA is responsible for real-time feature extraction and filtering calculations, while the GPU processes prediction models and encryption verification tasks.

[0075] Correction Actuator: A piezoelectric ceramic micro-adjuster (resolution of 0.1μm) for high-precision micro-adjustment, and a servo motor (response time of 5ms) responsible for large-range displacement compensation. The two work together through a PID controller.

[0076] Secure Communication Module: Mesh communication devices support the transmission of data using national cryptographic algorithms SM2 / SM4. The blockchain evidence storage unit uploads encrypted data (encrypted with AES-256) to the blockchain in real time, with an evidence storage interval of 1 second. The block hash value is generated through the SHA-256 algorithm to ensure the immutability of the data.

[0077] Example 3:

[0078] Intelligent Warehouse Logistics Monitoring Application

[0079] Scenario Background

[0080] In a large intelligent warehouse center, it is necessary to achieve high-precision positioning and dynamic path planning of automated guided vehicles (AGVs), while monitoring the stacking status of goods to avoid collisions and misalignments.

[0081] Implementation Steps

[0082] Benchmark Construction Phase

[0083] Use a 64-line lidar (scanning frequency of 20Hz) to perform 3D scanning of the warehouse, generating 3D point cloud data containing shelf contours, ground guiding lines, and safety channels.

[0084] Synchronously collect VR panoramic images of a multi-spectral camera (frame rate of 30fps), and eliminate dynamic interference caused by the movement of AGVs through multiple exposure synthesis.

[0085] Construct a 3D geographic information model of the warehouse, with the feature point density in the core area (such as the sorting area) set to 120 points / m 2 , and in the edge area (storage area) to 15 points / m 2 , and attach the material reflectivity (to distinguish metal shelves from plastic goods) and thermal radiation coefficient (to detect overheating of equipment) to each feature point.

[0086] Real-time Positioning and Dynamic Correction

[0087] The AGV is equipped with a monitoring terminal (integrating an IMU and an RGB-IR camera) to collect pose data and environmental images in real time.

[0088] Extract the visible light texture features and infrared heat source features of the shelf QR code through the ResNet50 network, match them with the spatial mapping relationship library, and calculate the AGV position offset.

[0089] The extended Kalman filter fuses the AGV motion acceleration and environmental light data, predicts the path deviation, drives the piezoelectric ceramic micro-adjuster (resolution 0.1μm) to adjust the AGV wheel hub steering, and the servo motor (response time 5ms) corrects the lateral displacement to ensure that the positioning error ≤ ±2cm.

[0090] Collaborative verification and safety response

[0091] Multiple AGVs construct a spatio-temporal grid through the Mesh communication network, use the improved RANSAC algorithm to screen the reference nodes (γ = 0.3), and initiate AES-256 encryption verification for the AGVs with abnormal offsets.

[0092] If the AGV offset ≥ 0.5m (secondary warning), the system restricts its speed and triggers a buzzer alarm; if the offset ≥ 2m (tertiary warning), switch to the Beidou / GPS dual-mode positioning, and record the trajectory data through the blockchain evidence storage platform (evidence storage interval 1 second) for the operation and maintenance personnel to trace and analyze.

[0093] Implementation effect

[0094] The AGV positioning accuracy is improved to ±2cm, the path deviation correction response delay < 100ms, the warehousing efficiency is increased by 30%, and the collision accident rate is reduced by 95%.

[0095] Example 4:

[0096] Application of construction site safety monitoring

[0097] Scene background

[0098] During high-rise building construction, it is necessary to monitor the positions of workers, tower cranes and heavy equipment in real time to prevent risks of out-of-bounds operation and falling objects from height.

[0099] Implementation steps

[0100] Benchmark construction stage

[0101] Use a drone to carry a 64-line lidar to scan the construction site and generate a three-dimensional model including the floor structure, tower crane track and safety warning area.

[0102] The multi-spectral camera takes panoramic images of the construction site, fuses the point cloud data and then marks the positions of high-altitude hanging objects (such as steel cables), and the feature points are attached with geometric invariant moments (to identify the shape of the tower crane arm) and thermal radiation coefficients (to detect abnormal heating of equipment).

[0103] Real-time monitoring and dynamic warning

[0104] Workers wear helmets with integrated IMU and infrared cameras to upload posture data and surrounding heat source distribution in real time.

[0105] The system uses the LSTM network to predict the swing trend of the tower crane and combines wind speed data (input window 10 seconds) to predict potential collision risks.

[0106] If a worker enters a high-risk area (such as the operating radius of a tower crane), the real-time matching of the feature vector will trigger a first-level warning (interface warning). At the same time, the servo motor will be used to adjust the angle of the surveillance camera to track the movements of the person.

[0107] Collaborative encryption and emergency response

[0108] A collaborative verification chain is built between multiple devices through a Mesh network using a national encryption algorithm, and Laplace noise (ε=0.1) is injected into low-confidence terminals (such as devices in signal-blocking areas) to protect privacy data.

[0109] When the tower crane offset is ≥2m, the system automatically cuts off its operating authority, triggers blockchain evidence storage (recording timestamp, device ID and offset), and notifies the supervisor through the smart contract to intervene in the inspection.

[0110] Implementation Effect

[0111] Construction site safety accidents are reduced by 80%, tower crane positioning accuracy reaches ±5cm, emergency response time is <150ms, and it complies with ISO45001 safety management standards.

[0112] The above is only a preferred specific implementation of the present invention; however, the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solution and its improved conception within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A precise correction monitoring method based on VR panoramic position and map coordinates, characterized in that, Including the following steps: S1. Benchmark construction stage: Using a lidar to scan the target area to obtain three-dimensional point cloud data, synchronously collecting VR panoramic images and performing multiple exposure synthesis to eliminate the interference of dynamic objects; Fusing the three-dimensional point cloud data and VR panoramic images, constructing a three-dimensional geographic information model including building contour lines, ground markers, and high-altitude hanging objects, and generating a spatial mapping relationship library. The spatial mapping relationship library is a set of feature points generated from the three-dimensional geographic information model. Each feature point includes material reflectivity, thermal radiation coefficient, and geometric invariant moments; S2. Real-time positioning stage: Using a monitoring terminal to collect pose data and real-time images, and extracting visible light texture features and infrared heat source distribution features in the real-time images; S3. Precise matching stage: Encoding the visible light texture features and infrared heat source distribution features into feature vectors, performing similarity matching with the environmental feature vectors in the spatial mapping relationship library, and eliminating invalid matching points through geometric verification, and calculating the spatial position offset; S4. Dynamic correction stage: Constructing a filtering model based on sensor data, real-time correcting the current offset, and predicting the future offset trend; Fusing the filtering result and the prediction result to generate dynamic compensation parameters, and driving the actuator to perform positioning correction; S5. Cooperative verification stage: Using the communication network between devices to screen benchmark nodes, constructing a spatio-temporal grid and assigning dynamic weight coefficients, sending an encrypted verification request to low-confidence terminals, and generating a cooperative positioning report; S6. Safety response stage: Triggering a hierarchical warning mechanism according to the offset, and performing corresponding safety operations, including warning log recording, permission restriction, and encrypted location information storage.

2. The precise correction monitoring method based on VR panoramic position and map coordinates according to claim 1, wherein, The exposure time of the multiple exposure synthesis described in step S1 is dynamically adjusted according to the illumination intensity of the target area, and by default includes 10 ms, 30 ms, and 50 ms. The feature point density of the spatial mapping relationship library is 120 points / m in the core area 2 and 15 points / m in the edge area 2 , and each feature point is additionally included with an environmental feature vector containing the material reflectivity, thermal radiation coefficient, and geometric invariant moment.

3. The precise correction monitoring method based on VR panoramic position and map coordinates according to claim 1, characterized in that, In step S3, the visible light texture features and infrared heat source distribution features are encoded by a pre-trained ResNet50 convolutional neural network, then the feature vectors are matched by cosine similarity, and the geometric rationality is verified by Delaunay triangulation. The tolerance range of the side length error is ±5 cm. The input of the ResNet50 network is the RGB-IR multi-channel data of the real-time image.

4. A precise correction monitoring method based on VR panoramic position and map coordinates according to claim 1, characterized in that In step S4, the filtering model is an extended Kalman filter, and its state variables include terminal motion acceleration, environmental light intensity, and electromagnetic interference level. The future offset trend is predicted by an LSTM network. The input data includes the historical offset sequence and real-time wind speed data. The historical offset sequence is extracted by the sliding window method, and the window length is 10 seconds. The dynamic compensation parameters are fused with the filtering result and the prediction result by a fuzzy control algorithm, where the weight of the filtering result α = 0.6 and the weight of the prediction result β = 0.

4.

5. A precise correction monitoring method based on VR panoramic position and map coordinates according to claim 1, characterized in that, In step S5, the screening reference node adopts an improved RANSAC algorithm, with an added inertial navigation weighting factor γ = 0.

3. The encrypted verification request adopts the AES-256 algorithm and the differential privacy technology of the Laplace noise mechanism (ε ＝ 0 . 1).

6. The precise correction monitoring method based on VR panoramic position and map coordinates according to claim 1, wherein, In step S6, the hierarchical warning mechanism is a three-level warning mechanism, and the triggering conditions of the three-level warning mechanism are: Level 1 warning: offset < 0.5 m, triggering warning icons and log records; Level 2 warning: 0.5 m ≤ offset < 2 m, starting the buzzer alarm and restricting operation permissions; Level 3 Early Warning: When the offset ≥ 2m, switch to the Beidou / GPS dual-mode positioning module, and encrypt and store evidence through the blockchain evidence storage platform. The encrypted evidence storage data of the blockchain evidence storage platform includes the offset, timestamp, and terminal device ID, and multi-party verification is achieved through smart contracts.

7. A precise correction monitoring system based on VR panoramic position and map coordinates, for implementing the method according to any one of claims 1-6, characterized in that, It includes: Data acquisition module: A wearable device integrating a lidar, a multispectral camera, and an IMU sensor; Collaborative processing unit: A heterogeneous computing platform composed of an FPGA and a GPU. The FPGA optimizes convolutional calculations, and the GPU runs the LSTM network; Correction actuator: Includes a piezoelectric ceramic micro-adjuster and a servo motor; Secure communication module: A Mesh communication device supporting national cryptographic algorithms and a blockchain evidence storage unit.

8. A precise correction monitoring system based on VR panoramic position and map coordinates according to claim 7, characterized in that, The lidar is a 64-line lidar with a scanning frequency ≥ 20Hz. The multispectral camera supports the visible light and near-infrared bands with a frame rate ≥ 30fps. The resolution of the piezoelectric ceramic micro-adjuster ≤ 0.1μm, and the response time of the servo motor ≤ 5ms.

9. The precise correction monitoring system based on VR panoramic position and map coordinates according to claim 7, characterized in that, The FPGA uses the Winograd algorithm to optimize convolutional calculations with a processing speed ≥ 130 frames per second. The evidence storage interval of the blockchain evidence storage unit is 1 second, and the block hash value is generated through the SHA-256 algorithm.

Images