Cooperative sensing network construction method based on Beidou and low-altitude remote sensing data

Abstract

The invention aims to provide a cooperative sensing network construction method based on Beidou and low-altitude remote sensing data. The method comprises the following steps: shooting the same place by adopting a plurality of unmanned aerial vehicles to obtain a plurality of low-altitude remote sensing data; taking the low-altitude remote sensing data of which the key frame similarity between every two low-altitude remote sensing data exceeds a threshold value as overlapped low-altitude remote sensing data pairs; fusing each pair of overlapped low-altitude remote sensing data to obtain three-dimensional low-altitude remote sensing data; acquiring Beidou positioning data; and carrying out cooperative processing on the three-dimensional low-altitude remote sensing data and the Beidou positioning data to obtain accurate positioning data. According to the invention, cooperative sensing is carried out through the Beidou positioning system and the low-altitude remote sensing system, so that the three-dimensional model map of the to-be-measured destination can be accurately obtained, and land resource investigation, agricultural resource investigation and urban daily remote monitoring and daily patrol can be assisted.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data. Background Art

[0002] Beidou positioning refers to the technology of positioning using the Beidou satellite navigation system. The Beidou satellite navigation system is a global satellite navigation system independently developed by China. It is one of the four major satellite navigation systems in the world, along with GPS, GLONASS, and GALILEO. The working principle of Beidou positioning is to calculate the longitude, latitude, altitude, time and other information of the receiver by receiving the signals sent by Beidou satellites. At least 4 satellites are required for the first positioning to achieve the accuracy of 3D positioning. Low-altitude remote sensing refers to the science and technology of obtaining information on the earth's surface from the air by using aviation platforms with low flying altitudes (generally referring to airspace below 1,000 meters), such as drones and light aircraft, equipped with various sensors. Low-altitude remote sensing is a non-contact, long-distance detection technology that obtains the reflected, radiated or scattered electromagnetic wave information of the target object through sensors or remote sensors, and processes and analyzes it. Low-altitude remote sensing is mainly used in the following scenarios: Land and resources survey: used for emergency monitoring, key area monitoring, remote sensing sampling and calibration, etc. Agriculture: Conduct agricultural resource surveys, precision fertilization, pesticide spraying, land rights confirmation, etc. Urban governance: Used for river patrols, illegal building monitoring, etc. Forest fire prevention: Provide remote monitoring, daily patrols, emergency flights and other services.

[0003] Collaborative Perception is a technology that enhances perception through multi-sensor collaboration, involving data processing, distributed computing, and artificial intelligence (AI). It integrates the data collected by sensors through the collaboration between multiple sensors to produce more accurate and complete perception results. Collaborative perception solves the problems of long-distance occlusion and sparse data in single perception. The algorithms and technologies usually involved include data processing, distributed computing, and artificial intelligence. In the fusion of multi-UAV low-altitude remote sensing data and Beidou data, traditional technology usually requires time synchronization of multiple sensors and then fusion of the data. Time synchronization is a complex algorithm and requires a lot of calculations, resulting in low efficiency in collaborative perception of Beidou data and low-altitude remote sensing data. Summary of the invention

[0004] The purpose of the present invention is to provide a method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data. The method can accurately obtain a three-dimensional model map of the target area to be measured through collaborative perception of the Beidou positioning system and the low-altitude remote sensing system, and can assist in land and resources surveys, agricultural resources surveys, and daily remote monitoring and daily patrols in cities.

[0005] The collaborative perception network construction method based on Beidou and low-altitude remote sensing data includes: Use multiple drones to shoot the same location to obtain multiple low-altitude remote sensing data; The low-altitude remote sensing data with key frame similarity exceeding a threshold between each other are regarded as overlapping low-altitude remote sensing data pairs; Each pair of overlapping low-altitude remote sensing data is fused to obtain three-dimensional low-altitude remote sensing data; Get Beidou positioning data; The three-dimensional low-altitude remote sensing data and Beidou positioning data are collaboratively processed to obtain precise positioning data.

[0006] Preferably, the method of taking the low-altitude remote sensing data with key frame similarities exceeding a threshold as overlapping low-altitude remote sensing data pairs comprises: The two low-altitude remote sensing data whose key frame similarity exceeds a threshold are regarded as adjacent low-altitude remote sensing data; Adjacent remote sensing data are spliced ​​to obtain three-dimensional low-altitude remote sensing data.

[0007] Preferably, the collaborative processing of the three-dimensional low-altitude remote sensing data and Beidou positioning data to obtain precise positioning data includes: Identify key features in three-dimensional low-altitude remote sensing data; Calculate the similarity between the key features and Beidou positioning data; If the similarity is less than the preset value, Beidou positioning data is used to correct the three-dimensional low-altitude remote sensing data.

[0008] Preferably, the key features of identifying three-dimensional low-altitude remote sensing data include: Use the YOLOv8 network to detect key features in remote sensing data and output multiple feature points; The YOLOv8 network includes: an input layer, a Backbone layer, a Neck layer, a Head layer and an output layer; The positive sample allocation strategy is adopted as the loss function.

[0009] Preferably, if the similarity is less than a preset value, using Beidou positioning data to correct the three-dimensional low-altitude remote sensing data includes: Calculate positioning correction factors based on Beidou positioning data and three-dimensional low-altitude remote sensing data; The three-dimensional low-altitude remote sensing data is corrected according to the positioning correction factor.

[0010] Preferably, the calculating the similarity between the key feature and Beidou positioning data comprises: The similarity between the key features and the Beidou positioning data is determined based on the distance between the key features and the Beidou positioning data.

[0011] Preferably, the step of taking two low-altitude remote sensing data whose key frame similarity exceeds a threshold as adjacent low-altitude remote sensing data comprises: Calculate the score of each remote sensing data according to the view angle, overlap area and scale factor; According to the score of each remote sensing data, it is determined which two pairs of remote sensing data are adjacent.

[0012] The collaborative perception network construction system based on Beidou and low-altitude remote sensing data includes: Data acquisition module, used to obtain low-altitude remote sensing data and Beidou positioning data; A data processing module performs collaborative processing on the low-altitude remote sensing data and Beidou positioning data to obtain precise positioning data; A path planning module is used to construct a navigation path according to the precise positioning data.

[0013] An electronic device includes: a chip, a processor and a memory, wherein the memory is used to store computer program code, and the computer program code includes computer instructions. When the chip executes the computer instructions, the electronic device executes a method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data.

[0014] A computer-readable storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes a method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data.

[0015] The beneficial effects of the present invention are: 1. The present invention first processes a set of remote sensing data taken by multiple drones, screens out remote sensing data pairs that can be spliced, and fuses multiple remote sensing data pairs to obtain preliminary three-dimensional low-altitude remote sensing data; 2. The present invention uses Beidou positioning data to coordinately process three-dimensional low-altitude remote sensing data, and finally obtains accurate positioning data, which can assist in land and resources surveys, agricultural resources surveys, and daily remote monitoring and daily patrols in cities; 3. The present invention does not need to synchronize the time of multiple sensors, but instead fuses low-altitude remote sensing data and Beidou data by establishing a score for each low-altitude remote sensing data and its similarity to Beidou data. The algorithm is simpler and the work efficiency is high. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0018] Figure 1 A flow chart of a method for constructing a collaborative sensing network based on Beidou and low-altitude remote sensing data according to the present invention; Figure 2 A flow chart of the present invention for collaboratively processing three-dimensional low-altitude remote sensing data and Beidou positioning data to obtain precise positioning data; Figure 3 The figure is a schematic diagram of the hardware structure of an electronic device of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0021] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their 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 addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0022] Collaborative Perception is a technology that enhances perception through multi-sensor collaboration, involving data processing, distributed computing, and artificial intelligence (AI). It integrates the data collected by sensors through the mutual collaboration between multiple sensors to produce more accurate and complete perception results. Collaborative perception solves the problems of long-distance occlusion and sparse data in single perception. The algorithms and technologies usually involved include data processing, distributed computing, and artificial intelligence. In the fusion of multi-UAV low-altitude remote sensing data and Beidou data, traditional technology usually requires time synchronization of multiple sensors and then fusion of the data. Time synchronization is a complex algorithm and requires a lot of calculations, resulting in low efficiency in collaborative perception of Beidou data and low-altitude remote sensing data.

[0023] The present invention first processes a set of remote sensing data taken by multiple drones, screens out remote sensing data pairs that can be spliced, and fuses multiple remote sensing data pairs to obtain preliminary three-dimensional low-altitude remote sensing data; the present invention uses Beidou positioning data to coordinately process the three-dimensional low-altitude remote sensing data, and finally obtains accurate positioning data, which can assist in land and resources surveys, agricultural resources surveys, and daily remote monitoring and daily patrols in cities.

[0024] Example 1 A collaborative perception network construction method based on Beidou and low-altitude remote sensing data, refer to Figure 1 ,include: S100 uses multiple drones to shoot the same location to obtain multiple low-altitude remote sensing data; Before shooting low-altitude remote sensing data, you need to set the shooting area and clarify the boundaries and key points. Set the flight altitude, speed, and overlap rate (usually 70%-80%) of the drone to ensure data integrity. And choose the right time to avoid weather or light effects. Select the camera or multispectral, thermal infrared and other sensors carried by the drone according to your needs. Finally, obtain the captured remote sensing data.

[0025] S200, taking the low-altitude remote sensing data with the number of key frames between each pair exceeding a threshold as an overlapping low-altitude remote sensing data pair; Due to the overlap of drone shots, the remote sensing data captured cannot be used directly and need to be spliced ​​and reconstructed to finally obtain a complete three-dimensional low-altitude remote sensing data.

[0026] S300, fusing each pair of overlapping low-altitude remote sensing data to obtain three-dimensional low-altitude remote sensing data; ‌Three-dimensional low-altitude remote sensing data‌ refers to data obtained from the air using low-altitude platforms (such as drones, light aircraft, etc.) equipped with various sensors to obtain information about the earth's surface. These data usually have high resolution and clear details, and can provide more accurate surface information.

[0027] S400, obtain Beidou positioning data; Beidou positioning data refers to the location information data obtained through the Beidou satellite navigation system. Beidou positioning data mainly includes the user's latitude and longitude coordinates, altitude and other information, which are used to determine the specific location of an object. The positioning principle of the Beidou positioning system is based on the concept of "arrival time difference" (delay). Each satellite is equipped with a precise atomic clock and radio transmitter. The receiver (such as a mobile phone or car navigation system) receives signals from three or more satellites and calculates the receiver's position by measuring the time difference between the signal from the satellite to the receiver.

[0028] S500 coordinates the processing of three-dimensional low-altitude remote sensing data and Beidou positioning data to obtain precise positioning data.

[0029] First, use the GDAL tool to convert Beidou positioning data (WGS84 coordinate system) and remote sensing data (local coordinate system) to the same coordinate system, then accurately align the three-dimensional remote sensing data with the Beidou positioning data, and fuse the aligned data to generate a three-dimensional model or point cloud data with precise geographic coordinates. In this process, it is necessary to ensure the coordinate system 1: Ensure that all data is processed in the same coordinate system to avoid errors caused by inconsistent coordinates. In the data processing process, strictly control the accuracy of each step to ensure the accuracy of the final result.

[0030] In the prior art, the fusion of remote sensing data and Beidou positioning data requires time synchronization to be able to fuse the remote sensing data and Beidou positioning data to avoid errors caused by time deviation. Time synchronization requires multi-sensor data fusion, which greatly increases the complexity and amount of calculation of the algorithm. In the embodiment of the present invention, the remote sensing data and Beidou positioning data can be fused only based on the similarity between the remote sensing data and the Beidou positioning data, which also has the effect of high precision, and the algorithm is simpler and more efficient.

[0031] Preferably, S200, taking the low-altitude remote sensing data with key frame similarities exceeding a threshold as overlapping low-altitude remote sensing data pairs includes: S210, taking two low-altitude remote sensing data whose key frame similarity exceeds a threshold as adjacent low-altitude remote sensing data; ‌Keyframes are state data of remote sensing data at a specific point in time. By setting keyframes on the timeline, the software can automatically calculate and interpolate the changes between these keyframes.

[0032] S220, splicing adjacent remote sensing data to obtain three-dimensional low-altitude remote sensing data.

[0033] Because there needs to be a certain overlap rate when using drones or other platforms to shoot the target area from different angles, it is necessary to extract the key frames in the remote sensing data, match the same key frames between different remote sensing data, and establish the correspondence between remote sensing data. Find adjacent remote sensing data for splicing, and finally obtain three-dimensional low-altitude remote sensing data.

[0034] Preferably, reference Figure 2 , S500, coordinates the processing of three-dimensional low-altitude remote sensing data and Beidou positioning data to obtain precise positioning data including: S510, identifying key features in three-dimensional low-altitude remote sensing data; Key features include points with significant characteristics in three-dimensional low-altitude remote sensing data, such as corner points, edge points, and areas with rich textures. Before extracting key points, the three-dimensional low-altitude remote sensing data can also be preprocessed, including: denoising the image, histogram equalization and other operations to improve the feature extraction effect. After extracting key points, detect key points at multiple scales to enhance scale invariance. Remove duplicate key points and retain the most representative points. Use PCA, hashing and other methods to reduce the descriptor dimension and improve matching efficiency.

[0035] S520, calculating the similarity between the key feature and the Beidou positioning data; The similarity between the key features and the Beidou positioning data reflects the accuracy of the three-dimensional low-altitude remote sensing data. If the similarity between the key features and the Beidou positioning data is high, it means that the gap between the three-dimensional low-altitude remote sensing data and the actual detection is not large, and the accuracy is low. If the similarity between the key features and the Beidou positioning data is low, it means that there is a certain gap between the three-dimensional low-altitude remote sensing data and the actual detection, and the three-dimensional low-altitude remote sensing data needs to be corrected.

[0036] S530: If the similarity is less than a preset value, the three-dimensional low-altitude remote sensing data is corrected using Beidou positioning data.

[0037] After correcting the three-dimensional low-altitude remote sensing data, it is also necessary to adaptively adjust the parameters of the reconstruction and splicing in order to improve the accuracy of the next low-altitude remote sensing detection.

[0038] Preferably, S510, identifying key features in three-dimensional low-altitude remote sensing data includes: Use the YOLOv8 network to detect key features in remote sensing data and output multiple feature points; The YOLOv8 network includes: input layer, Backbone layer, Neck layer, Head layer and output layer; Backbone: Adopts C2f module, and improves feature extraction capability through Bottleneck Block and SPPF modules. C2f module reduces the amount of calculation by reducing the number of channels of input Tensor, while increasing gradient flow, improving convergence speed and effect. Neck: Adopts PAN-FAN structure, and performs multi-scale feature fusion through PAN and FAN. Compared with YOLOv5, YOLOv8 removes the 1*1 downsampling layer. Head: Responsible for target detection, adopts decoupled head structure (Decoupled-Head), separates classification and detection heads, and changes from Anchor-Based to Anchor-Free mechanism.

[0039] The positive sample allocation strategy is adopted as the loss function.

[0040] The positive sample allocation strategy helps the model learn the target features more efficiently by clarifying which samples (such as anchor boxes, pixels) belong to positive samples (i.e., targets) and which belong to negative samples (i.e., background). It avoids the model's excessive attention to a large number of negative samples during training, thereby concentrating resources on learning the features of positive samples. In target detection tasks, negative samples (background) are usually far more than positive samples (targets), leading to class imbalance. The positive sample allocation strategy ensures that the model pays balanced attention to positive and negative samples during training by reasonably allocating positive and negative samples, avoiding the model's bias towards predicting the background. The positive sample allocation strategy ensures that each target has enough positive samples (such as anchor boxes) to match, thereby improving the model's recall rate for the target. By dynamically adjusting the positive sample allocation (such as ATSS, OTA), it can better adapt to targets of different scales and shapes and improve detection accuracy. The positive sample allocation strategy can dynamically adjust the allocation of positive and negative samples according to the characteristics of the target (such as scale, shape, and occlusion), making the model more robust to complex scenes. For example, when the target is partially occluded, the strategy can assign more positive samples to the visible part to ensure that the model can correctly learn the characteristics of the target. In the absence of a positive sample allocation strategy, the model may assign multiple positive samples (such as multiple anchor boxes) to the same target, resulting in ambiguity during training. The positive sample allocation strategy reduces this ambiguity through clear rules (such as IoU threshold, center point distance), making model training more stable. The positive sample allocation strategy enables the model to learn the characteristics of the target faster by reducing the interference of invalid samples, thereby accelerating model convergence. For example, Focal Loss accelerates convergence by reducing the weight of easy-to-classify samples, allowing the model to pay more attention to difficult-to-classify samples. In complex scenes, targets may have different scales, shapes, and occlusions. The positive sample allocation strategy can dynamically adjust the sample allocation according to the characteristics of the target, so that the model can better adapt to complex scenes.

[0041] Methods for extracting key features also include: scale space extreme value detection: using Gaussian difference pyramid to detect key points. Key point positioning: accurately positioning key points and removing low contrast and edge response points. Direction assignment: assigning main directions to key points to achieve rotation invariance. Feature descriptor generation: extracting gradient information of the area around the key points and generating 128-dimensional descriptors.

[0042] Preferably, in S530, if the similarity is less than a preset value, correcting the three-dimensional low-altitude remote sensing data using Beidou positioning data includes: The positioning correction factor is calculated based on Beidou positioning data and three-dimensional low-altitude remote sensing data, which is expressed as: ; in, is the coordinate of Beidou positioning data, is the coordinate of three-dimensional remote sensing data.

[0043] The three-dimensional low-altitude remote sensing data is corrected according to the positioning correction factor.

[0044] ‌Positioning correction factor‌ refers to the correction value introduced to compensate for various errors when correcting 3D low-altitude remote sensing data. These correction factors are mainly used to adjust various errors in 3D low-altitude remote sensing data to ensure the accuracy of positioning. Positioning correction factors can improve the positioning accuracy of the target by fine-tuning the initial positioning results to reduce positioning errors. In target detection, the correction factor can adjust the position and size of the bounding box to make it more closely fit the real target. In complex scenes, the initial positioning of the target may be affected by noise, occlusion, or changes in illumination.

[0045] The positioning correction factor makes the model more robust to noise and interference by introducing additional constraints or optimization methods. And the corrected targets may have different scales and shapes, and the initial positioning results may not be able to fully adapt to these changes. The positioning correction factor can dynamically adjust the positioning results according to the characteristics of the target to better adapt to multi-scale targets. The positioning correction factor can also be jointly optimized with other parts of the model (such as classification branches and regression branches) to achieve end-to-end training, which can further improve the overall performance of the model. The positioning correction factor accelerates convergence by providing more accurate positioning information and reducing the search space of the model during training. The correction factor can reduce the search range of the bounding box, allowing the model to find the optimal solution faster.

[0046] Preferably, S520, calculating the similarity between the key feature and the Beidou positioning data includes: The similarity between the key features and the Beidou positioning data is determined based on the distance between the key features and the Beidou positioning data, which is expressed as: ; in, is the Euclidean distance between the i-th feature point and the Beidou positioning data; S is the pixel area of ​​the remote sensing data bounding box; is the visibility of the feature points; is the confidence of the feature point; is an exponential function, when The value is 1 when , otherwise it is 0.

[0047] Remote sensing data is aligned with Beidou positioning data through the similarity between key features and Beidou positioning data to generate remote sensing products (such as orthophotos, three-dimensional models) with precise geographic coordinates. In addition, Beidou positioning data and three-dimensional low-altitude remote sensing data can assist each other in reducing errors and improving positioning accuracy. By analyzing the similarity between remote sensing data and Beidou positioning data, target detection and recognition can be assisted. Beidou positioning data provides geographic location information, and remote sensing data provides visual or geometric information of the scene. The scene can be better understood through similarity analysis. The similarity analysis between remote sensing data and Beidou positioning data can support the collaborative processing of multi-source data and improve data utilization. In an embodiment of the present invention, by analyzing the similarity between remote sensing data and Beidou positioning data, errors in the data can be detected and quality control can be performed.

[0048] Preferably, S210, taking two low-altitude remote sensing data whose key frame similarity exceeds a threshold as adjacent low-altitude remote sensing data includes: Calculate the score of each remote sensing data according to the view angle, overlap area and scale factor; ; Among them, C is the score of each remote sensing data, is the angle factor, is the scale factor, is the overlapping area between the pre-selected remote sensing data pairs; ; in, , represents the area shared by the pre-selected remote sensing data pairs, Indicates the base area.

[0049] According to the score of each remote sensing data, it is determined which remote sensing data are adjacent.

[0050] The main function of stitching adjacent remote sensing data is to ensure the integrity and continuity of the data. Through stitching processing, satellite image data of different scenes can be merged into a complete and continuous image to ensure that the data is not misplaced, so as to achieve seamless continuity of geographic information in subsequent analysis and application. This processing is particularly important for applications such as map making, resource management and environmental monitoring. It can provide consistent and complete geographic information, which helps to better understand and utilize the information on the surface of the target ground. Stitching processing can eliminate various radiation distortions and geometric distortions, so that the processed images can more realistically show the true appearance of the original scene. By stitching strip data at different time points, dynamic monitoring of the same area can be achieved, revealing the process and trend of surface changes, which is of great significance to urban planning, disaster warning and other fields. Compared with traditional ground observation methods, the acquisition cost of low-altitude remote sensing data is low and is not restricted by ground conditions. By stitching multiple strip data, large-scale observations can be achieved in a short time, improving work efficiency.

[0051] In an embodiment of the present invention, low-altitude remote sensing data and Beidou positioning data are processed in a coordinated manner, which can assist in urban management and geological disaster prediction. For example, remote sensing technology can obtain macro information of urban traffic, such as road distribution, general situation of traffic flow, etc. Beidou positioning data can provide accurate location and real-time dynamic information of vehicles, such as specific driving trajectory and speed of electric vehicles and cars. The combination of the two can enable traffic planners to have a more comprehensive understanding of urban traffic conditions, so as to rationally plan roads, set up bus routes and stations, etc., and also help traffic management departments to adjust traffic signals and optimize traffic organization in a timely manner. Remote sensing data can monitor the topography, vegetation coverage and other conditions of the surface on a large scale through satellite or aerial remote sensing images, and quickly discover areas where there may be geological disaster risks, such as landslides, mudslides and other prone areas. Beidou positioning data can accurately locate geological disaster monitoring points, monitor the position changes of monitoring points in real time, and issue early warnings in time once the position of the monitoring points moves abnormally. Remote sensing data can intuitively reflect the current status of urban land use, such as the distribution of buildings, green spaces, roads, etc. Beidou positioning data can accurately determine the location of these features, helping planners to more accurately carry out land use planning and urban layout, and reasonably arrange the functional areas of the city. Remote sensing data can be used to monitor marine environmental factors such as ocean temperature, salinity, and ocean currents, as well as the distribution of fishery resources in the ocean. Beidou positioning data can provide accurate location information for marine monitoring equipment, fishing vessels, etc., facilitating the rational development and management of marine resources, and also contributing to the safe operation of fishing vessels and the sustainable use of fishery resources.

[0052] Example 2 The collaborative perception network construction system based on Beidou and low-altitude remote sensing data includes: Data acquisition module, used to obtain low-altitude remote sensing data and Beidou positioning data; The data processing module coordinates the low-altitude remote sensing data and Beidou positioning data to obtain precise positioning data; The path planning module is used to build a navigation path based on precise positioning data.

[0053] Example 3 An electronic device includes: a chip, a processor and a memory, the memory is used to store computer program code, the computer program code includes computer instructions, and when the chip executes the computer instructions, the electronic device executes a method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data.

[0054] Example 4 A computer-readable storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes a method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data.

[0055] refer to Figure 3 , the electronic device 2 includes a processor 21, a memory 22, an input device 23, and an output device 24. The processor 21, the memory 22, the input device 23, and the output device 24 are coupled via a connector, and the connector includes various interfaces, transmission lines, or buses, etc., which are not limited in the embodiments of the present invention. It should be understood that in various embodiments of the present invention, coupling refers to mutual connection in a specific manner, including direct connection or indirect connection through other devices, for example, through various interfaces, transmission lines, buses, etc.

[0056] The processor 21 may be one or more graphics processing units (GPUs). When the processor 21 is a GPU, the GPU may be a single-core GPU or a multi-core GPU. Optionally, the processor 21 may be a processor group consisting of multiple GPUs, and the multiple processors are coupled to each other via one or more buses. Optionally, the processor may also be other types of processors, etc., which are not limited in the embodiments of the present invention.

[0057] The memory 22 can be used to store computer program instructions and various computer program codes including program codes for executing the scheme of the present invention. Optionally, the memory includes but is not limited to random access memory (RAM), read-only memory (ROM), erasable programmable read only memory (EPROM), or portable read only memory (CD-ROM), which is used for related instructions and data.

[0058] The input device 23 is used to input data and / or signals, and the output device 24 is used to output data and / or signals. The output device 24 and the input device 23 may be independent devices or an integrated device.

[0059] The present invention first processes a set of remote sensing data taken by multiple drones, selects remote sensing data pairs that can be spliced, and fuses multiple remote sensing data pairs to obtain preliminary three-dimensional low-altitude remote sensing data; the present invention uses Beidou positioning data to coordinately process the three-dimensional low-altitude remote sensing data, and finally obtains accurate positioning data, which can assist in land and resources surveys, agricultural resource surveys, and daily remote monitoring and daily patrols in cities. The present invention does not need to synchronize the time of multiple sensors, but instead fuses low-altitude remote sensing data and Beidou data by establishing the score of each low-altitude remote sensing data and the similarity with Beidou data. The algorithm is simpler and the work efficiency is high.

[0060] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data, characterized in that: include: Use multiple drones to shoot the same location to obtain multiple low-altitude remote sensing data; The low-altitude remote sensing data with key frame similarity exceeding a threshold between each other are regarded as overlapping low-altitude remote sensing data pairs; Each pair of overlapping low-altitude remote sensing data is fused to obtain three-dimensional low-altitude remote sensing data; Get Beidou positioning data; The three-dimensional low-altitude remote sensing data and Beidou positioning data are collaboratively processed to obtain precise positioning data.

2. The method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data according to claim 1 is characterized in that: The method of taking the low-altitude remote sensing data with key frame similarities exceeding a threshold as overlapping low-altitude remote sensing data pairs includes: The two low-altitude remote sensing data whose key frame similarity exceeds a threshold are regarded as adjacent low-altitude remote sensing data; Adjacent remote sensing data are spliced ​​to obtain three-dimensional low-altitude remote sensing data.

3. The method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data according to claim 1 is characterized in that: The collaborative processing of the three-dimensional low-altitude remote sensing data and Beidou positioning data to obtain accurate positioning data includes: Identify key features in three-dimensional low-altitude remote sensing data; Calculate the similarity between the key features and Beidou positioning data; If the similarity is less than the preset value, Beidou positioning data is used to correct the three-dimensional low-altitude remote sensing data.

4. The method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data according to claim 3 is characterized in that: The key features of identifying three-dimensional low-altitude remote sensing data include: Use the YOLOv8 network to detect key features in remote sensing data and output multiple feature points; The YOLOv8 network includes: an input layer, a Backbone layer, a Neck layer, a Head layer and an output layer; The positive sample allocation strategy is adopted as the loss function.

5. The method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data according to claim 3 is characterized in that: If the similarity is less than a preset value, the three-dimensional low-altitude remote sensing data is corrected using Beidou positioning data, including: Calculate positioning correction factors based on Beidou positioning data and three-dimensional low-altitude remote sensing data; The three-dimensional low-altitude remote sensing data is corrected according to the positioning correction factor.

6. The method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data according to claim 3 is characterized in that: Calculating the similarity between the key feature and Beidou positioning data includes: The similarity between the key features and the Beidou positioning data is determined based on the distance between the key features and the Beidou positioning data.

7. The method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data according to claim 2 is characterized in that: The method of taking two low-altitude remote sensing data whose key frame similarity exceeds a threshold as adjacent low-altitude remote sensing data includes: Calculate the score of each remote sensing data according to the view angle, overlap area and scale factor; According to the score of each remote sensing data, it is determined which two pairs of remote sensing data are adjacent.

8. A collaborative perception network construction system based on Beidou and low-altitude remote sensing data, characterized in that: include: Data acquisition module, used to obtain low-altitude remote sensing data and Beidou positioning data; A data processing module performs collaborative processing on the low-altitude remote sensing data and Beidou positioning data to obtain precise positioning data; A path planning module is used to construct a navigation path according to the precise positioning data.

9. An electronic device, characterized in that: include: A chip, a processor and a memory, wherein the memory is used to store computer program code, and the computer program code includes computer instructions. When the chip executes the computer instructions, the electronic device executes the method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor of an electronic device, the processor executes the method for constructing a collaborative perception network based on Beidou and low-altitude remote sensing data as described in any one of claims 1 to 7.

Images