Police service inspection method based on unmanned aerial vehicle

Abstract

The invention discloses a police service inspection method based on an unmanned aerial vehicle, and belongs to the technical field of unmanned aerial vehicle control. The police service inspection method comprises the following steps: preprocessing a multispectral image, a thermal image and point cloud data to obtain a spectral characteristic distribution diagram, a temperature change diagram and a three-dimensional space structure; performing space-time alignment on the spectral feature distribution diagram, the temperature change diagram and the three-dimensional space structure, then calculating a corresponding relation between pixel points and point clouds, and generating a comprehensive data set; calculating a slope parameter, a height parameter and an obstacle distribution parameter through the topographic feature parameters in the integrated data set to obtain an adaptability score of the current topography, determining a flight height adjustment value and a speed adjustment value of the unmanned aerial vehicle in combination with a wind speed parameter and a humidity parameter, and generating an adjustment parameter set; and updating the flight attitude of the UAV according to the adjustment parameter set. According to the police service inspection method based on the unmanned aerial vehicle, the problem that an existing unmanned aerial vehicle control mode is difficult to meet all-weather and all-terrain inspection requirements is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of UAV control, and particularly relates to a police patrol method based on UAVs. Background Art

[0002] As an important technical field of modern police management, UAV police patrol plays an irreplaceable key role in maintaining public safety, emergency response and complex environment monitoring with its high-efficiency coverage and real-time monitoring capabilities. With the acceleration of urbanization and the increasing complexity of security needs, the application of UAV remote sensing technology has become an important breakthrough in improving police efficiency. However, there are still significant limitations in the technical implementation of existing patrol methods. Traditional solutions mostly rely on single sensors or limited environmental adaptability, resulting in insufficient data acquisition capabilities under changing terrains, wind speeds or humidity, and it is difficult to meet the patrol requirements for all weather and all terrains. Summary of the Invention

[0003] In order to overcome the defects existing in the prior art, the present invention provides a police patrol method based on UAVs to solve the above problems.

[0004] The technical solution adopted by the present invention to solve its technical problems is: a police patrol method based on UAVs, including the following steps: S1: Obtain multispectral images, thermal images and point cloud data through a multispectral camera, an infrared thermal imager and a lidar respectively, and perform preprocessing on the multispectral images, thermal images and point cloud data to obtain a spectral feature distribution map, a temperature change map and a three-dimensional spatial structure; S2: Perform spatio-temporal alignment on the spectral feature distribution map, the temperature change map and the three-dimensional spatial structure, and then calculate the corresponding relationship between the pixel points of the spectral feature distribution map and the temperature change map and the point cloud in the three-dimensional spatial structure to generate a comprehensive data set; S3: Calculate the slope parameter S, the height parameter H and the obstacle distribution parameter D through the terrain feature parameters in the comprehensive data set to obtain the adaptability score A of the current terrain, and combine the wind speed parameter V and the humidity parameter M in the externally input weather data to determine the UAV flight height adjustment value H1 and the speed adjustment value V1, and generate an adjustment parameter set; S4: Update the UAV flight attitude according to the adjustment parameter set.

[0005] Preferably, in the step S1, the preprocessing process is: perform normalization processing on the multispectral images, thermal images and point cloud data; perform spectral feature extraction on the normalized multispectral images to obtain a spectral feature distribution map; perform temperature field simulation on the normalized thermal images to obtain a temperature change map; perform three-dimensional reconstruction on the normalized point cloud data to obtain a three-dimensional spatial structure.

[0006] Optionally, in step S2, the time and space consistency of the spectral feature distribution map, the temperature change map, and the three-dimensional spatial structure is adjusted through spatio-temporal alignment technology to obtain a preliminary alignment dataset.

[0007] Specifically, in step S2, the Scale-Invariant Feature Transform (SIFT) algorithm is used to extract feature points from the spectral feature distribution map and the temperature change map in the preliminary alignment dataset to obtain a feature point set; The Iterative Closest Point (ICP) algorithm is used to perform spatial mapping on the feature point set and the three-dimensional spatial structure, determine the correspondence between the pixel points in the feature point set and the point cloud in the three-dimensional spatial structure, obtain mapping result data, and use the mapping result data as a comprehensive dataset.

[0008] It should be noted that in step S2, fusion features are extracted from the mapping result data, and a fusion error value is obtained by calculating the deviation of the fusion features; It is determined whether the fusion error value is lower than a preset threshold. If it is lower, the comprehensive dataset is output; otherwise, after adjusting the parameters of the ICP algorithm, the steps of obtaining the mapping result data and obtaining the fusion error value are repeatedly executed until the fusion error value is lower than the preset threshold.

[0009] Preferably, in step S3, a terrain analysis tool in a geographic information system is used to calculate the slope parameter S, the elevation parameter H, and the obstacle distribution parameter D of the terrain feature parameters to obtain terrain feature description data; For the terrain feature description data, the slope parameter S, the height parameter H, and the obstacle distribution parameter D are fused using the linear weighted method to determine the adaptability score A.

[0010] Optionally, in step S3, the external wind speed parameter V and the humidity parameter M are obtained, and in combination with the adaptability score A, a bilinear interpolation algorithm is used to calculate the flight height adjustment value and the speed adjustment value to obtain an adjustment parameter set.

[0011] Specifically, in step S3, the process of the bilinear interpolation algorithm is as follows: four neighboring points required for interpolation are determined according to the ranges of the historical wind speed and historical humidity in the current area; According to the adaptability score A, the corresponding initial height adjustment value and initial speed adjustment value for each neighboring point are obtained from a predefined lookup table; The predicted height adjustment value and predicted speed adjustment value under the current wind speed parameter V and humidity parameter M are calculated using the bilinear interpolation formula; height adjustment value = initial height adjustment value + predicted height adjustment value, speed adjustment value = initial speed adjustment value + predicted speed adjustment value.

[0012] The beneficial effects of the present invention are as follows: The drone-based police patrol method first obtains the original data collected by a multispectral camera, an infrared thermal imager, and a lidar, and generates a comprehensive data set through standardization processing and spatio-temporal alignment. Subsequently, based on the terrain feature parameters in the comprehensive data set and combined with external weather data, the terrain adaptability score is calculated, and the adjustment values of the drone flight altitude and speed are determined. Through multi-source data processing and terrain adaptability analysis, the present invention realizes real-time optimization and adjustment of the drone flight parameters, improves the flight safety and task execution efficiency of the drone under complex terrain and changing weather conditions, and thus meets the patrol requirements for all weather and all terrains. Description of the Drawings

[0013] Figure 1 It is a flowchart of the drone-based police patrol method in an embodiment of the present invention. Detailed Embodiments

[0014] The following further describes the detailed embodiments of the present invention with reference to the drawings. It should be noted here that the description of these embodiments is for helping to understand the present invention, but does not constitute a limitation to the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0015] As Figure 1 shown, a drone-based police patrol method includes the following steps: S1: Respectively obtain a multispectral image, a thermal image, and point cloud data through a multispectral camera, an infrared thermal imager, and a lidar. After preprocessing the multispectral image, the thermal image, and the point cloud data, a spectral feature distribution map, a temperature change map, and a three-dimensional spatial structure are obtained; the data formats are standardized according to the data interface protocol between devices to obtain a spectral feature distribution map, a temperature change map, and a three-dimensional spatial structure in a unified format; S2: Perform spatio-temporal alignment on the spectral feature distribution map, the temperature change map, and the three-dimensional spatial structure, and then calculate the corresponding relationship between the pixel points of the spectral feature distribution map and the temperature change map and the point cloud in the three-dimensional spatial structure to generate a comprehensive data set; S3: Calculate the slope parameter S, the height parameter H, and the obstacle distribution parameter D through the terrain feature parameters in the comprehensive data set to obtain the adaptability score A of the current terrain. Combine the wind speed parameter V and the humidity parameter M in the externally input weather data to determine the drone flight altitude adjustment value H1 and the speed adjustment value V1, and generate an adjustment parameter set; S4: Update the drone flight attitude according to the adjustment parameter set. Specifically, update the drone flight attitude according to the flight altitude adjustment value H1 and the speed adjustment value V1 in the adjustment parameter set.

[0016] The police inspection method based on drones first obtains the raw data collected by multispectral cameras, infrared thermal imagers and lidars, and generates a comprehensive data set through standardized processing and spatiotemporal alignment. Subsequently, the terrain characteristic parameters in the comprehensive data set are combined with external weather data to calculate the terrain adaptability score and determine the adjustment value of the drone's flight altitude and speed. The present invention realizes real-time optimization and adjustment of drone flight parameters through multi-source data processing and terrain adaptability analysis, and improves the flight safety and mission execution efficiency of drones under complex terrain and changeable weather conditions. Thereby meeting the inspection needs of all-weather and all-terrain.

[0017] It is worth noting that in step S1, the preprocessing process is: normalizing the multispectral image, thermal imaging image and point cloud data; extracting spectral features of the normalized multispectral image to obtain a spectral feature distribution map; simulating the temperature field of the normalized thermal imaging image to obtain a temperature change map; and performing three-dimensional reconstruction on the normalized point cloud data to obtain a three-dimensional spatial structure.

[0018] After obtaining the spectral feature distribution map, temperature change map and three-dimensional spatial structure, the data are comparable. The purpose of extracting spectral features from the normalized multispectral image is to mine information such as vegetation health; for example, the reflectivity difference between the red and near-infrared bands can be analyzed to calculate the vegetation index. For example, the reflectivity ratio of healthy areas is higher and that of withered areas is lower, forming a spectral feature distribution map. Temperature field simulation of normalized thermal images can reveal abnormally high temperature points; for example, if the temperature jumps from ambient temperature to a temperature near body temperature, a person can be identified and a temperature change map is generated. Three-dimensional reconstruction of normalized point cloud data can restore the spatial structure.

[0019] Assuming that a suburban or forest area is monitored, the spectral feature distribution map provides spectral information of vegetation, the temperature change map reflects the temperature distribution, and the point cloud data provides three-dimensional terrain information. These data have different formats and dimensions due to differences in equipment, and are difficult to integrate directly. Normalization according to preset standardization rules can unify the data range and improve the accuracy of subsequent analysis. In this embodiment, for multispectral images, the pixel values ​​of each band can be mapped to the 0-1 interval. For example, if the original value range of a band is 50-200, it can be linearly scaled to 0-1.

[0020] For thermal images, temperature values ​​such as 20°C-40°C can also be normalized to the same range. The coordinate values ​​of the point cloud data are translated and scaled to fit the same scale.

[0021] Preferably, in step S2, the temporal and spatial consistency of the spectral feature distribution graph, the temperature change graph and the three-dimensional spatial structure is adjusted by using a spatiotemporal alignment technique to obtain a preliminary aligned data set.

[0022] Suppose a suburban or mountain forest area is monitored. The spectral feature distribution map provides spectral information of vegetation, the temperature change map reflects the temperature distribution, and the three-dimensional spatial structure shows the terrain structure. The spatio-temporal alignment technology ensures the synchronization of these data in time and space. Specifically, the data collected by different devices can be aligned to the same moment by timestamp matching, such as adjusting the acquisition time difference between the drone multi-spectral camera and the thermal imager to within 0.5 seconds, and at the same time using GPS coordinates to calibrate the spatial position to control the deviation within 1 meter.

[0023] Specifically, in the step S2, the SIFT algorithm is used to extract feature points from the spectral feature distribution map and the temperature change map in the preliminary alignment dataset, obtaining a feature point set; The ICP algorithm is used to perform spatial mapping on the feature point set and the three-dimensional spatial structure, determine the correspondence between the pixel points in the feature point set and the point cloud in the three-dimensional spatial structure, obtain the mapping result data, and use the mapping result data as the comprehensive dataset.

[0024] When using the SIFT algorithm to extract feature points, the spectral feature distribution map can identify significant spectral changes at the edge of vegetation, such as the mutation points of the reflectance in the red light band, while the temperature change map can extract the boundary of the high-temperature area. SIFT generates a feature point set containing positions and directions by detecting these local invariant features, extracting approximately 200 key points per image. It should be noted that these feature points lay the foundation for subsequent spatial mapping.

[0025] For the spatial mapping of the ICP algorithm, the correspondence between the feature point set and the point cloud data is crucial. The ICP algorithm is used to perform initial registration on the feature point set and the three-dimensional spatial structure to obtain a rough registration result. According to the rough registration result, fine registration is performed on the feature point set and the three-dimensional spatial structure to calculate an accurate transformation matrix. The transformation matrix is applied to perform spatial mapping on the feature point set to establish an accurate correspondence between the pixel points in the feature point set and the point cloud. According to the established correspondence, the mapping result data is generated, and the mapping result data contains the point cloud coordinate information corresponding to each feature point.

[0026] Optionally, in the step S2, fusion features are extracted from the mapping result data, and by calculating the deviation of the fusion features, a fusion error value is obtained; It is judged whether the fusion error value is lower than a preset threshold. If it is lower, the comprehensive dataset is output; otherwise, after adjusting the parameters of the ICP algorithm, the steps of obtaining the mapping result data and obtaining the fusion error value are repeated until the fusion error value is lower than the preset threshold.

[0027] ICP aligns the pixel coordinates of the spectral feature distribution map and the temperature change map with the three-dimensional coordinates of the point cloud in the three-dimensional space structure by iteratively optimizing the distance between points. For example, the edge pixels of a certain vegetation may be mapped to a point with a height of 15 meters in the point cloud, and the initial error may be 0.8 meters. Adjusting ICP parameters, such as increasing the number of iterations to 50 times or reducing the distance threshold to 0.1 meters, can reduce the error. It can be understood that each pixel in the mapping result data forms a one-to-one correspondence with the point cloud points. When extracting the fusion features from the mapping result data, the joint distribution deviation of the spectral value, temperature value, and height value can be calculated. Specifically, by calculating the Euclidean distance between each fusion feature and its corresponding reference value, the feature deviation value is obtained. According to the feature deviation value, the fusion error value is calculated using the weighted average method.

[0028] It should be noted that in the step S3, the slope parameter S, elevation parameter H, and obstacle distribution parameter D of the terrain feature parameters are calculated using the terrain analysis tool in the geographic information system to obtain the terrain feature description data; For the terrain feature description data, the slope parameter S, the height parameter H, and the obstacle distribution parameter D are fused using the linear weighted method to determine the adaptability score A.

[0029] In the monitoring of a street area, the comprehensive dataset contains the fusion information of multi-spectral, thermal imaging, and point cloud, and terrain features such as slope, height, and obstacle distribution can be extracted from it. For example, the slope parameter S reflects the degree of terrain inclination, the height parameter H represents the surface undulation, and the obstacle distribution parameter D describes the density of vegetation. Specifically, the slope can be calculated through the three-dimensional coordinates of the point cloud data. Assuming that the point cloud height in a certain area changes from 10 meters to 12 meters and the horizontal distance is 5 meters, the slope can be characterized as a relatively gentle inclination. The terrain analysis tool in the geographic information system can be used to calculate these parameters. For example, the raster analysis method is used to process the point cloud data, and the suburban or mountain forest area is divided into 10-meter × 10-meter grids, and the average height and slope are calculated for each grid. The obstacle distribution parameter D can be obtained through the point cloud density analysis. If the number of point clouds in a certain grid exceeds 1000, it can be considered that there is more vegetation. Exemplarily, assuming that the weight of the slope parameter S is 0.4, the weight of the height parameter H is 0.3, and the weight of the obstacle distribution parameter D is 0.3, according to the terrain characteristics, if S is 15 degrees, H is 20 meters, and D is 0.8 in a certain area, the adaptability score A can be calculated as a comprehensive value through weighted calculation. This score reflects the complexity of the terrain and helps to judge the difficulty of UAV flight.

[0030] Preferably, in the step S3, the external wind speed parameter V and humidity parameter M are obtained, combined with the adaptability score A, and the bilinear interpolation algorithm is used to calculate the flight height adjustment value and speed adjustment value to obtain the adjustment parameter set.

[0031] After obtaining the external wind speed parameter V and humidity parameter M, the bilinear interpolation algorithm can further optimize and adjust the values. For example, when the wind speed V is 5 m / s and the humidity M is 70%, combined with the adaptability score A, the interpolation calculation shows that the flight altitude needs to be increased by 2 m and the speed needs to be reduced by 1 m / s. It can be understood that this method balances the external environment and terrain factors through two-dimensional interpolation, and the generated fourth adjustment parameter is closer to the actual needs.

[0032] Specifically, assume that the current wind speed parameter V is 15 m / s, the humidity parameter M is 70%, and the adaptability score A is 85; in step S3, the process of the bilinear interpolation algorithm is as follows: determine the four neighboring points required for interpolation according to the range of historical wind speed and historical humidity in the current area; for example, the wind speed range is 10 to 20 m / s, the humidity range is 60% to 80%, and the four neighboring points required for interpolation are (10, 60), (10, 80), (20, 60), and (20, 80); According to the adaptability score A, obtain the initial altitude adjustment value and initial speed adjustment value corresponding to each neighboring point from the predefined lookup table; assume that the altitude adjustment value corresponding to the neighboring point (10, 60) in the lookup table is -50 m and the speed adjustment value is -10 m / s; the altitude adjustment value corresponding to the neighboring point (10, 80) is -40 m and the speed adjustment value is -8 m / s; the altitude adjustment value corresponding to the neighboring point (20, 60) is -60 m and the speed adjustment value is -12 m / s; the altitude adjustment value corresponding to the neighboring point (20, 80) is -50 m and the speed adjustment value is -10 m / s; Use the bilinear interpolation formula to calculate the predicted altitude adjustment value and predicted speed adjustment value under the current wind speed parameter V and humidity parameter M; altitude adjustment value = initial altitude adjustment value + predicted altitude adjustment value, speed adjustment value = initial speed adjustment value + predicted speed adjustment value; in this embodiment, the bilinear interpolation formula is: ; where represents the predicted altitude adjustment value or predicted speed adjustment value, is the current wind speed parameter, is the current humidity parameter, is the lower limit value of the wind speed range, is the upper limit value of the wind speed range, is the lower limit value of the humidity range, is the upper limit value of the humidity range; for the height adjustment value, first perform linear interpolation between wind speeds of 10 m / s and 20 m / s, and then perform linear interpolation between humidities of 60% and 80% to obtain the predicted height adjustment value; similarly, calculate the predicted speed adjustment value; then obtain the height adjustment value through the initial height adjustment value and the predicted height adjustment value, and obtain the speed adjustment value through the initial speed adjustment value and the predicted speed adjustment value. Finally, the height adjustment value at a current wind speed of 15 m / s and a humidity of 70% is -55 m, and the speed adjustment value is -15 m / s.

[0033] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, without departing from the principle and spirit of the present invention, various changes, modifications, substitutions, and variations made to these embodiments still fall within the protection scope of the present invention.

Claims

1. A police patrol method based on drones, characterized in that, It includes the following steps: S1: Obtain multispectral images, thermal images, and point cloud data through a multispectral camera, an infrared thermal imager, and a lidar respectively. After preprocessing the multispectral images, thermal images, and point cloud data, obtain a spectral feature distribution map, a temperature change map, and a three-dimensional spatial structure; S2: Perform spatio-temporal alignment on the spectral feature distribution map, the temperature change map, and the three-dimensional spatial structure, and then calculate the correspondence between the pixel points of the spectral feature distribution map and the temperature change map and the point cloud in the three-dimensional spatial structure to generate a comprehensive dataset; S3: Calculate the slope parameter S, the height parameter H, and the obstacle distribution parameter D of the terrain feature parameters through the comprehensive dataset to obtain the adaptability score A of the current terrain. Combine the wind speed parameter V and the humidity parameter M in the externally input weather data to determine the drone flight height adjustment value H1 and the speed adjustment value V1, and generate an adjustment parameter set; S4: Update the drone flight attitude according to the adjustment parameter set.

2. The method for police patrol inspection based on an unmanned aerial vehicle according to claim 1, characterized in that: In the step S1, the preprocessing process is: perform normalization processing on the multispectral images, thermal images, and point cloud data; extract spectral features from the normalized multispectral images to obtain a spectral feature distribution map; perform a temperature field simulation on the normalized thermal images to obtain a temperature change map; perform three-dimensional reconstruction on the normalized point cloud data to obtain a three-dimensional spatial structure.

3. The method for police patrol based on an unmanned aerial vehicle according to claim 2, characterized in that: In the step S2, adjust the temporal and spatial consistency of the spectral feature distribution map, the temperature change map, and the three-dimensional spatial structure through spatio-temporal alignment technology to obtain a preliminary alignment dataset.

4. The method for police patrol inspection based on an unmanned aerial vehicle according to claim 3, wherein: In the step S2, use the SIFT algorithm to extract feature points from the spectral feature distribution map and the temperature change map in the preliminary alignment dataset to obtain a feature point set; Use the ICP algorithm to perform spatial mapping on the feature point set and the three-dimensional spatial structure, determine the correspondence between the pixel points in the feature point set and the point cloud in the three-dimensional spatial structure to obtain mapping result data, and use the mapping result data as a comprehensive dataset.

5. The method for police patrol based on an unmanned aerial vehicle according to claim 4, characterized in that: In the step S2, extract the fusion features from the mapping result data, and obtain a fusion error value by calculating the deviation of the fusion features; Judge whether the fusion error value is lower than a preset threshold. If it is lower, output the comprehensive dataset. Otherwise, after adjusting the parameters of the ICP algorithm, repeat the steps of obtaining the mapping result data and obtaining the fusion error value until the fusion error value is lower than the preset threshold.

6. The method for police patrol inspection based on an unmanned aerial vehicle according to claim 5, wherein: In the step S3, use the terrain analysis tool in the geographic information system to calculate the slope parameter S, the elevation parameter H, and the obstacle distribution parameter D of the terrain feature parameters to obtain terrain feature description data; For the terrain feature description data, use the linear weighted method to fuse the slope parameter S, the height parameter H, and the obstacle distribution parameter D to determine the adaptability score A.

7. The method for police patrol inspection based on an unmanned aerial vehicle according to claim 6, wherein: In the step S3, obtain the external wind speed parameter V and the humidity parameter M, combine the adaptability score A, and use the bilinear interpolation algorithm to calculate the flight height adjustment value and the speed adjustment value to obtain an adjustment parameter set.

8. A method for police patrol inspection based on an unmanned aerial vehicle according to claim 7, characterized in that: In the step S3, the process of the bilinear interpolation algorithm is as follows: Determine four neighboring points required for interpolation according to the ranges of the historical wind speed and historical humidity in the current area; Obtain the corresponding initial height adjustment value and initial speed adjustment value for each neighboring point from a predefined lookup table according to the adaptability score A; Use the bilinear interpolation formula to calculate the predicted height adjustment value and predicted speed adjustment value under the current wind speed parameter V and humidity parameter M; Height adjustment value = Initial height adjustment value + Predicted height adjustment value, Speed adjustment value = Initial speed adjustment value + Predicted speed adjustment value.

Images