Unmanned aerial vehicle hydrological flow and velocity measurement method

Abstract

The invention discloses a hydrological flow and speed measurement method, which relates to the technical field of hydrological measurement and comprises the following steps: selecting a dynamic anchor point in a target speed measurement area, and carrying out anchor point hovering on an unmanned aerial vehicle according to the dynamic anchor point; the unmanned aerial vehicle collects surrounding images of the unmanned aerial vehicle to obtain obstacle data, and a motion route is generated in real time; the unmanned aerial vehicle performs video acquisition on the target velocity measurement area to generate a three-waveband water surface video, and analyzes the three-waveband water surface video to generate an initial velocity distribution diagram; the unmanned aerial vehicle collects riverbed data and water surface fluctuation data of the target flow measurement area, and according to the riverbed data, a riverbed terrain is obtained, and water depth data and a section average flow velocity are obtained; and constructing a three-dimensional water flow model, collecting a movement track of the buoy group and gyroscope data, correcting the three-dimensional water flow model, and outputting a hydrological flow and velocity measurement result. The method has the effects of improving the measurement precision of flow measurement and speed measurement and improving the flexibility and universality when facing a complex water area.

Description

Technical Field

[0001] This application relates to the technical field of hydrological measurement, and particularly to a method for measuring the flow rate and velocity of water using an unmanned aerial vehicle (UAV). Background Art

[0002] China is a country with numerous river basins. Especially in the southern region, the water network is densely distributed, and there are many large rivers. As a result, the detection and monitoring of the water flow in these rivers have become very important work contents. In particular, the monitoring of the water flow velocity of rivers is of great significance for flood prevention early warning of rivers.

[0003] In the prior art, when detecting the water flow velocity of a river, the traditional method still adopts the fixed-point monitoring method, setting detection devices on the bank of the river or putting flow meters into the river, etc. This requires a large amount of time and manpower, and it takes a certain amount of time to collect data, and the measurement time is relatively long. At the same time, during the flow measurement process, affected by various factors, such as wind speed, waves, impurities in the water flow, etc., the accuracy of the measurement result is relatively low. In addition, the above methods are only applicable to rivers with a certain width and water depth. When facing waters with a more complex environment, the difficulty of installing equipment and equipment measurement is relatively large, and it does not have universality. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for measuring the flow rate and velocity of water using an unmanned aerial vehicle to solve the problems raised in the above background art.

[0005] This application provides a method for measuring the flow rate and velocity of water using an unmanned aerial vehicle. The method includes: Obtain the target flow measurement area of the water area to be detected, call the unmanned aerial vehicle to go to the target flow measurement area, select a dynamic anchor point in the target velocity measurement area, and perform anchor point hovering on the unmanned aerial vehicle according to the dynamic anchor point; The unmanned aerial vehicle collects the surrounding images of itself, obtains obstacle data according to the surrounding images, and generates a movement route in real time according to the obstacle data and the dynamic anchor point; The unmanned aerial vehicle performs video collection on the target velocity measurement area, obtains a visible light water surface video, an infrared light water surface video, and a polarized light water surface video, fuses them to generate a three-band water surface video, and analyzes the three-band water surface video to generate an initial flow velocity distribution map; The unmanned aerial vehicle collects the riverbed data and water surface fluctuation data of the target flow measurement area, obtains the riverbed topography according to the riverbed data, and obtains the water depth data and the cross-sectional average flow velocity according to the riverbed topography and the water surface fluctuation data; Based on the riverbed topography, the water depth data, and the initial flow velocity distribution map, a three-dimensional water flow model is constructed, and a drone is called to release a group of buoys. The movement trajectories and gyroscope data of the group of buoys are collected, and the three-dimensional water flow model is corrected according to the movement trajectories, the gyroscope data, and the cross-sectional average flow velocity, and the hydrological flow measurement and velocity measurement results are output.

[0006] Preferably, the step of selecting a dynamic anchor point in the target velocity measurement area and performing anchor point hovering of the drone according to the dynamic anchor point is specifically as follows: Call the drone to collect images of the target velocity measurement area to obtain a watershed image, extract water flow information from the watershed image to obtain a water flow image, and particleize the water flow image to obtain water flow image particles Based on the water flow image particles, extract multiple particle features in the water flow image particles, and record the obviousness value of each particle feature; Select the particle feature with the highest obviousness value as the target particle feature, and perform anchor point selection on the water flow image particles according to the target particle feature to obtain a dynamic anchor point; Take pictures of the dynamic anchor point at a preset time interval, and record the displacement of the dynamic anchor point in two pictures. Obtain the hovering speed according to the displacement and the time interval; Adjust the propeller power distribution of the drone according to the hovering speed so that the drone remains relatively stationary with the dynamic anchor point.

[0007] Preferably, the step of the drone collecting the surrounding images of itself, obtaining obstacle data according to the surrounding images, and generating a movement route in real time according to the obstacle data and the dynamic anchor point is specifically as follows: The drone collects images of the surrounding environment of itself to obtain surrounding images, and discriminates the content of the surrounding images to obtain obstacle images; Extract the visual distance feature and the external shape feature of the obstacle image. Obtain the obstacle distance based on the visual distance feature, obtain the obstacle spatial position based on the external shape feature, and generate initial obstacle data by combining the obstacle distance and the obstacle spatial position; According to the initial obstacle data and the external shape feature, predict the obstacles outside or after the surrounding images to obtain predicted obstacle data, and add the predicted obstacle data to the initial obstacle data to generate obstacle data; Obtain the current UAV spatial position of the UAV and the current anchor spatial position of the dynamic anchor, and combine the obstacle data, the UAV spatial position, and the anchor spatial position to generate the movement route of the UAV in real time.

[0008] Preferably, after the step of combining the obstacle data, the UAV spatial position, and the anchor spatial position to generate the movement route of the UAV in real time, it further includes: When the UAV moves on the movement route, the surrounding images are updated in real time, and the obstacle images are updated in real time to obtain updated obstacle images; Based on the updated obstacle images, obtain updated obstacle data, and judge whether the movement route is intercepted by the obstacle based on the updated obstacle data; If it is judged that the movement route is intercepted by the obstacle, extract the intercepted route segment in the movement route; Based on the route segment and the updated obstacle data, reconstruct the route segment to obtain an updated route segment, and add the updated route segment to the movement route.

[0009] Preferably, before the step of fusing the visible light water surface video, the infrared light water surface video, and the polarized light water surface video to generate a three-band water surface video, it further includes: Extract the light reflection area in the visible light water surface video, and extract the light amount of the light reflection area. Perform exposure reduction processing on the light reflection area according to the light amount to obtain a target light reflection area; Add the target light reflection area to the visible light water surface video to obtain a target visible light water surface video; Based on the infrared light water surface video, extract the temperature distribution area of the water surface and the infrared light scattering amount, and perform clarity processing on the infrared light water surface video based on the temperature distribution area and the infrared light scattering amount to obtain a target infrared light water surface video; Based on the polarized light water surface video, extract the polarization state transformation of the water surface, generate polarization compensation parameters according to the polarization state change, and perform parameter compensation on the polarized light water surface video according to the polarization compensation parameters to obtain a target polarized light water surface video.

[0010] Preferably, the step of fusing the visible light water surface video, the infrared light water surface video, and the polarized light water surface video to generate a three-band water surface video and analyzing the three-band water surface video to generate an initial flow velocity distribution map is specifically: Fuse the target visible light water surface video, the target infrared light water surface video, and the target polarized light water surface video to generate a three-band water surface video; Based on the target infrared light water surface video, eliminate the water mist influence and temperature influence that appear in the three-band water surface video; Based on the target polarized light water surface video, enhance the content clarity of the underwater picture in the three-band water surface video; Perform water surface frame recognition on the three-band water surface video to obtain water surface video frames, and record the visible light water surface differences and infrared light water surface differences between the water surface video frames; Mutually verify and identify the visible light water surface difference and the infrared light water surface difference to obtain the initial water surface flow velocity distribution; Perform underwater frame recognition on the three-band water surface video to obtain underwater video frames, and record the water ripple shape differences and water ripple color differences between the underwater video frames; Based on the water ripple shape difference and the water ripple color difference, obtain the underwater water flow change characteristics, and based on the underwater water flow change characteristics, obtain the initial underwater flow velocity distribution; According to the initial water surface flow velocity distribution and the initial underwater flow velocity distribution, generate an initial flow velocity distribution map of the target flow measurement area.

[0011] Preferably, the step of eliminating the water mist influence and temperature influence that appear in the three-band water surface video based on the target infrared light water surface video is specifically: Extract the infrared frame picture in the target infrared light water surface video, and obtain the infrared temperature distribution and the water surface infrared characteristics based on the infrared frame picture; Locate the water surface boundary based on the water surface infrared characteristics to obtain water surface boundary data; Perform layout segmentation on the infrared temperature distribution according to the water surface boundary data to obtain the water surface temperature distribution and the over-water temperature distribution respectively; Based on the over-water temperature distribution, obtain the water mist influence in the target flow measurement area, and based on the water surface temperature distribution, obtain the temperature influence in the target flow measurement area; Generate a water mist picture compensation based on the water mist influence, generate a temperature picture compensation based on the temperature influence, and eliminate the water mist influence and the temperature influence in the three-band water surface video according to the water mist picture compensation and the temperature picture compensation.

[0012] Preferably, the step of obtaining the riverbed topography according to the riverbed data and obtaining the water depth data and the cross-sectional average flow velocity according to the riverbed topography and the water surface fluctuation data is specifically: Model the riverbed of the target flow measurement area according to the riverbed data to obtain the riverbed topography; Based on the water surface fluctuation data, obtain the average water surface spatial position of the target flow measurement area, and obtain the water depth data according to the riverbed topography and the average water surface spatial position; According to the riverbed topography, obtain the cross-sectional spatial curve of the riverbed, and based on the cross-sectional spatial curve and the water depth data, construct the watershed cross-sectional image of the target flow measurement area; Based on the water surface fluctuation data and the water depth data, divide a plurality of deep flow velocity relationship diagrams on the watershed cross-sectional image; According to the initial flow velocity distribution diagram and the deep flow velocity relationship diagram, construct the cross-sectional flow velocity data on the watershed cross-sectional image, and obtain the cross-sectional average flow velocity according to the cross-sectional flow velocity data.

[0013] Preferably, the step of calling a drone to release a buoy group, collecting the movement trajectory and gyroscope data of the buoy group, and correcting the three-dimensional water flow model according to the movement trajectory, the gyroscope data and the cross-sectional average flow velocity is specifically as follows: Call a drone to release a buoy group, the buoy group moves on the water surface of the target flow measurement area, and collect the movement trajectory and gyroscope data of the buoy group; Based on the movement trajectory, obtain the basic water flow direction of the target flow measurement area, and according to the gyroscope data, obtain the water flow resistance data of the target flow measurement area; Construct the underwater water flow model in the three-dimensional water flow model based on the cross-sectional average flow velocity; Based on the basic water flow direction and the water flow resistance data, refine and correct the water surface water flow in the three-dimensional water flow model.

[0014] In summary, the present application includes at least one of the following beneficial technical effects: Inspect the target flow measurement area by using a drone. First, select a dynamic anchor point in the water flow of the target flow measurement area, and then keep the drone and the dynamic anchor point on the same vertical line to maintain relative stillness in the plane. During the process of the drone moving following the dynamic anchor point, collect the surrounding images of the drone itself, then obtain the surrounding obstacle data based on the surrounding images, and set the movement route of the drone itself in real time according to the obstacle data. Then, the drone respectively conducts visible light, infrared light, and polarized light video acquisitions on the water surface in the target velocity measurement area, and then preprocess and fuse these three videos respectively to generate a three-band water surface video. Eliminate the water mist influence and temperature influence on the flow velocity analysis according to the three-band water surface video to obtain the initial flow velocity distribution map of the water surface. Then collect the riverbed data and water surface fluctuation data of the target flow measurement area to obtain the water depth data, and then obtain the cross-sectional average flow velocity according to the water depth data and the initial flow velocity distribution. Construct a three-dimensional water flow model based on the riverbed topography, water depth data, and initial flow velocity distribution map, then call the drone to release a group of buoys, and the drone collects the movement trajectories and gyroscope data of the buoy group to obtain the basic water flow direction and water flow resistance data respectively. Refine or correct the water surface water flow in the three-dimensional water flow model according to these two data, and then construct an underwater water flow model in the three-dimensional water flow model according to the cross-sectional average flow velocity. Finally, output a hydrological flow measurement and velocity measurement report according to the corrected three-dimensional water flow model. The measurement accuracy of flow measurement and velocity measurement, as well as the flexibility and universality in the face of complex water areas, are improved. Brief Description of the Drawings

[0015] Figure 1 It is a flowchart of the steps of a method for measuring water flow and velocity by using a drone provided by an embodiment of the present application. Detailed Embodiment

[0016] The following combines Figure 1 to further elaborate on the present application in detail, but the implementation manner of the present invention is not limited thereto.

[0017] An embodiment of the present application discloses a method for measuring water flow and velocity by using a drone.

[0018] In this embodiment, a method for measuring water flow and velocity by using a drone includes: S100: Obtain the target flow measurement area of the water area to be detected, call the drone to go to the target flow measurement area, select a dynamic anchor point in the target velocity measurement area, and perform anchor point hovering on the drone according to the dynamic anchor point; S200: The drone collects the surrounding images of itself, obtains the obstacle data according to the surrounding images, and generates a movement route in real time according to the obstacle data and the dynamic anchor point; S300: The drone conducts video acquisition on the target velocity measurement area, obtains visible light water surface video, infrared light water surface video, and polarized light water surface video, fuses them to generate a three-band water surface video, and analyzes the three-band water surface video to generate an initial flow velocity distribution map; S400: The drone collects the riverbed data and water surface fluctuation data of the target flow measurement area, obtains the riverbed topography based on the riverbed data, and obtains the water depth data and cross-sectional average flow velocity based on the riverbed topography and water surface fluctuation data; S500: Based on the riverbed topography, water depth data, and initial flow velocity distribution map, construct a three-dimensional water flow model, call the drone to release a buoy group, collect the movement trajectories and gyroscope data of the buoy group, and correct the three-dimensional water flow model according to the movement trajectories, gyroscope data, and cross-sectional average flow velocity, and output the hydrological flow measurement and velocity measurement results.

[0019] It should be noted that the above process is only the basic step of this embodiment. In the specific implementation process, without affecting the overall implementation effect, some steps can be appropriately added, reduced, or modified.

[0020] The steps of selecting a dynamic anchor point in the target velocity measurement area and hovering the drone according to the dynamic anchor point are specifically as follows: Call the drone to conduct image acquisition on the target velocity measurement area to obtain a basin image, extract water flow information from the basin image to obtain a water flow image, and particleize the water flow image to obtain water flow image particles Based on the water flow image particles, extract multiple particle features in the water flow image particles and record the obviousness value of each particle feature; Select the particle feature with the highest obviousness value as the target particle feature, and select an anchor point for the water flow image particles according to the target particle feature to obtain a dynamic anchor point; Take pictures of the dynamic anchor point at a preset time interval, record the displacement of the dynamic anchor point in two pictures, and obtain the hovering speed according to the displacement and the time interval; Adjust the propeller power distribution of the drone according to the hovering speed so that the drone remains relatively stationary with the dynamic anchor point.

[0021] In operation, taking the target flow measurement area of a certain river as an example, a drone is called to collect images of the river, obtaining a basin image. The basin image shows multiple vortices and waves on the water surface. Water flow information is extracted from the image to obtain a water flow image, which shows that the main water flow direction is southeast. The water flow image is particleized, and each water molecule is displayed as an independent particle, obtaining a water flow image particle. Particle features are extracted, including particle size, movement speed, and color depth. The obviousness value of each feature is calculated, and the obviousness value of the particle movement speed feature is the highest (the value is 95%). The particle area corresponding to this feature is selected as the dynamic anchor point, and the anchor point position is the central area of the river. The dynamic anchor point is photographed every 5 seconds, and it is found that the displacement of the anchor point in two consecutive photographs is 2 meters. Based on the displacement of 2 meters and the time interval of 5 seconds, the hovering speed is calculated to be 0.4 m / s. The power distribution of the four propellers of the drone is adjusted: the power of the two front propellers is reduced to 70%, and the power of the two rear propellers is increased to 120% to keep the drone relatively stationary with respect to the dynamic anchor point.

[0022] The steps for the drone to collect the surrounding images of itself, obtain obstacle data based on the surrounding images, and generate a movement route in real time according to the obstacle data and the dynamic anchor point are as follows: The drone collects images of the surrounding environment of itself to obtain surrounding images, and discriminates the content of the surrounding images to obtain obstacle images; The visual distance feature and the external shape feature of the obstacle image are extracted. The obstacle distance is obtained based on the visual distance feature, and the obstacle spatial position is obtained based on the external shape feature. The initial obstacle data is generated by combining the obstacle distance and the obstacle spatial position; Based on the initial obstacle data and the external shape feature, obstacles outside or after the surrounding images are predicted to obtain predicted obstacle data, and the predicted obstacle data is added to the initial obstacle data to generate obstacle data; The current drone spatial position of the drone and the current anchor point spatial position of the dynamic anchor point are obtained, and the movement route of the drone is generated in real time by combining the obstacle data, the drone spatial position, and the anchor point spatial position.

[0023] During operation, taking the target flow measurement area of a certain river as an example, the UAV collects surrounding images. The images show that there are branch obstacles on the left and rocks on the right. The visual distance feature of the branch obstacles is extracted (the calculated distance through the lens focal length is 15 meters), and the shape feature is a slender strip; the visual distance of the rocks is 20 meters, and the shape feature is an irregular block. Generate initial obstacle data: branches (15 meters, coordinates X = 10, Y = 5), rocks (20 meters, coordinates X = 30, Y = 8). According to the block feature of the rocks, it is predicted that there may be hidden obstacles behind them (such as stones submerged in water), and the coordinates of the predicted obstacle data are added as X = 32, Y = 10. Obtain the current position of the UAV (X = 50, Y = 0, Z = 100) and the dynamic anchor point position (X = 55, Y = 5, Z = 0). Generate a movement route: from the current position, bypass the branches and rocks in the southeast direction, and the path points are (X = 52, Y = 3) → (X = 58, Y = 7) → dynamic anchor point.

[0024] After the step of generating the movement route of the UAV in real time by combining the obstacle data, the UAV spatial position, and the anchor point spatial position, it further includes: When the UAV moves on the movement route, it updates the surrounding images in real time and updates the obstacle images in real time to obtain updated obstacle images; Based on the updated obstacle images, obtain updated obstacle data, and judge whether the movement route is intercepted by obstacles based on the updated obstacle data; If it is judged that the movement route is intercepted by obstacles, extract the intercepted route segment in the movement route; Based on the route segment and the updated obstacle data, reconstruct the route segment to obtain an updated route segment, and add the updated route segment to the movement route.

[0025] During operation, taking the target flow measurement area of a certain river as an example, when the UAV is flying along the initially planned route, a new obstacle is discovered. The UAV continuously captures the surrounding images and finds that a floating tree trunk appears near the originally planned path point (X = 58, Y = 7). The length of the tree trunk is about 6 meters, 15 meters away from the UAV, and it is moving downstream at a speed of 0.3 meters per second. At the same time, through the infrared sensor, an underwater reef is detected at the position (X = 60, Y = 5) on the right, only 0.8 meters away from the water surface. The system immediately updates the obstacle data and marks the coordinates of the tree trunk and the reef. The UAV is currently located at (X = 52, Y = 3), and the dynamic anchor point is at (X = 55, Y = 5). The system recalculates the route and finds that the section from (X = 58, Y - 7) to (X = 55, Y = 5) of the original path will be blocked by the tree trunk. The system generates three new routes: the first route bypasses to the right, passing through (X = 60, Y = 10) and (X = 62, Y = 6), increasing the flight distance by 13 meters; the second route climbs to a height of 120 meters to the left, but will consume 18% more power; the third route passes under the tree trunk 2 meters below, but there is a risk of collision. According to the remaining battery power of 75% and the urgency of the task, the first bypass route is selected. The UAV adjusts the power of the propellers, increasing the power of the left propeller to 130% and decreasing the power of the right propeller to 80%, and starts to turn to the new path point. During the flight, the position of the tree trunk is updated once per second. After finding that the tree trunk has drifted to X = 59, Y = 8, the system fine-tunes the path again, and finally safely arrives above the dynamic anchor point after 1 minute and 20 seconds.

[0026] Before the steps of fusing the visible light water surface video, the infrared light water surface video, and the polarized light water surface video to generate a three-band water surface video, it further includes: Extract the light reflection area in the visible light water surface video, and extract the light quantity of the light reflection area. Perform exposure reduction processing on the light reflection area according to the light quantity to obtain the target light reflection area; Add the target light reflection area to the visible light water surface video to obtain the target visible light water surface video; Based on the infrared light water surface video, extract the temperature distribution area of the water surface and the infrared light scattering quantity. Perform clarity processing on the infrared light water surface video based on the temperature distribution area and the infrared light scattering quantity to obtain the target infrared light water surface video; Based on the polarized light water surface video, extract the polarization state transformation of the water surface, generate polarization compensation parameters according to the polarization state change, and perform parameter compensation on the polarized light water surface video according to the polarization compensation parameters to obtain the target polarized light water surface video.

[0027] During operation, taking the target flow measurement area of a certain river as an example, in the visible light video captured by the UAV, strong reflection appears in the area (X = 15 - 20, Y = 30 - 35), and the light quantity reaches 2500 lumens. The system performs exposure reduction processing on this area, adjusts the light quantity to 800 lumens, and makes the water surface ripples clearly visible. The infrared light video shows that the central temperature of the water surface is 25°C, and the edge area is 28°C. The infrared scattering amount in the central area is 45% higher than that in the edge area. The system performs clarity processing on the central area, magnifies the temperature difference by 3 times, and makes the heat convection phenomenon more obvious. The polarized light video detects that the polarization angle of the water surface fluctuates between 30° and 45°, and generates a polarization compensation parameter of +5°. The fused display of the processed three-band video shows that details of water surface bubbles with a diameter of 0.5 mm in the original reflective area can be identified, and the clarity of the stone contour at a depth of 1.2 meters underwater is increased by 85%. Then, the system analyzes the influence of water mist in the infrared light video and finds that the water mist causes the water surface boundary to be blurred by about 20%. Through the temperature distribution data, the water mist area (temperature 22 - 24°C) is marked in the video, and the contrast of this area is increased by 50%. In the finally generated three-band video, the measurement error of the water surface flow velocity is reduced from the original 12% to 3%.

[0028] The steps of fusing the visible light water surface video, the infrared light water surface video, and the polarized light water surface video to generate a three-band water surface video and analyzing the three-band water surface video to generate an initial flow velocity distribution map are specifically as follows: Fuse the target visible light water surface video, the target infrared light water surface video, and the target polarized light water surface video to generate a three-band water surface video; Based on the target infrared light water surface video, eliminate the influence of water mist and temperature in the three-band water surface video; Based on the target polarized light water surface video, enhance the content clarity of the underwater picture in the three-band water surface video; Perform water surface frame recognition on the three-band water surface video to obtain water surface video frames, and record the visible light water surface differences and infrared light water surface differences between the water surface video frames; Mutually verify and identify the visible light water surface differences and infrared light water surface differences to obtain the initial water surface flow velocity distribution; Perform underwater frame recognition on the three-band water surface video to obtain underwater video frames, and record the water ripple shape differences and water ripple color differences between the underwater video frames; Based on the water ripple shape differences and water ripple color differences, obtain the underwater water flow change characteristics, and based on the underwater water flow change characteristics, obtain the initial underwater flow velocity distribution; Generate the initial flow velocity distribution map of the target flow measurement area according to the initial water surface flow velocity distribution and the initial underwater flow velocity distribution.

[0029] During operation, taking the target flow measurement area of a certain river as an example, the fused three-band video is split into 30 frames per second for analysis. The water surface frame shows that the wave height at the position of (X = 40, Y = 50) rises from 0.2 meters to 0.5 meters within 0.3 seconds, and the infrared difference shows that the temperature in this area drops by 0.6 °C per second. By comparing the visible light and infrared data, the system confirms the existence of a vortex here and calculates the water surface flow velocity to be 1.4 m / s. In the underwater frame analysis, it takes 2 seconds for the water pattern color at the position of (X = 38, Y = 55) to change from dark blue to light blue, and the shape stretching amount reaches 18%. Based on these data, the system calculates the flow velocity at a depth of 0.8 meters underwater to be 0.9 m / s. Then, the system superimposes the water surface and underwater data to generate an initial flow velocity distribution map. The map shows that the flow velocity in the 0 - 0.5 meter layer of the water surface is 1.2 - 1.6 m / s, the flow velocity in the middle layer of 0.5 - 1.5 meters is 0.8 - 1.2 m / s, and the flow velocity in the bottom layer of 1.5 - 3 meters is 0.5 - 0.8 m / s. At the same time, the system detects an abnormally low-speed area (flow velocity of 0.3 m / s) in the area of (X = 50, Y = 60). Combining the polarized light data, it is found that there is aquatic weed accumulation here, which is marked as a special observation point. After integrating all the data, the system generates an initial distribution map containing 12 flow velocity zones for subsequent 3D modeling.

[0030] Steps for eliminating the water mist influence and temperature influence in the three-band water surface video based on the target infrared light water surface video, specifically: Extract the infrared frame images from the target infrared light water surface video, and obtain the infrared temperature distribution and the water surface infrared characteristics based on the infrared frame images; Locate the water surface boundary based on the water surface infrared characteristics to obtain the water surface boundary data; Perform layout segmentation on the infrared temperature distribution according to the water surface boundary data to obtain the water surface temperature distribution and the above-water temperature distribution respectively; Based on the above-water temperature distribution, obtain the water mist influence in the target flow measurement area, and based on the water surface temperature distribution, obtain the temperature influence in the target flow measurement area; Generate a water mist image compensation based on the water mist influence, generate a temperature image compensation based on the temperature influence, and eliminate the water mist influence and temperature influence in the three-band water surface video according to the water mist image compensation and the temperature image compensation.

[0031] In operation, taking the target flow measurement area of a certain river as an example, the system extracts a frame from the infrared video and analyzes to obtain that the water surface boundary line is located in the coordinate band of (Y = 50 - 70). According to this boundary, the temperature distribution is segmented into a water surface area (25 - 28 °C) and an aerial water mist area (22 - 24 °C). It is detected that the water mist increases the blurriness of the water surface video by 25%, and the system applies a contrast enhancement algorithm to this area, increasing the edge sharpening parameter by 60%. At the same time, the water surface temperature distribution shows that there is an abnormally high temperature point (29 °C) in the area of (X = 45, Y = 55), which is found to be a misjudgment caused by direct sunlight. The system generates a temperature compensation formula and corrects the flow velocity data in this area to 85% of the original value. In the processed video, the flow velocity at (X = 45, Y = 55) is corrected from 1.5 m / s to 1.28 m / s. In addition, the system detects that the infrared velocity measurement deviation occurs in the water surface edge area (Y = 50 - 55) due to low temperature, recalculates this area with a weight coefficient of 0.7 increased, and finally controls the overall velocity measurement error within ±2%.

[0032] The steps of obtaining the riverbed topography based on the riverbed data and obtaining the water depth data and the cross-section average flow velocity based on the riverbed topography and the water surface fluctuation data are specifically as follows: Model the riverbed of the target flow measurement area based on the riverbed data to obtain the riverbed topography; Based on the water surface fluctuation data, obtain the average water surface spatial position of the target flow measurement area, and obtain the water depth data based on the riverbed topography and the average water surface spatial position; Based on the riverbed topography, obtain the cross-section spatial curve of the riverbed, and based on the cross-section spatial curve and the water depth data, construct the basin cross-section image of the target flow measurement area; Based on the water surface fluctuation data and the water depth data, divide and obtain multiple deep flow velocity relationship diagrams on the basin cross-section image; Based on the initial flow velocity distribution diagram and the deep flow velocity relationship diagrams, construct the cross-section flow velocity data on the basin cross-section image, and obtain the cross-section average flow velocity based on the cross-section flow velocity data.

[0033] In operation, taking the target flow measurement area of a certain river as an example, the drone obtains the riverbed data through sonar, showing that the deepest point is located at (X = 40, Y = 60) with a depth of 3.2 meters. Through the analysis of the water surface fluctuation data, the average wave height is obtained as 0.25 meters, and the system calculates the average water surface height as 100.3 meters above sea level. Combining with the riverbed topography, a water depth distribution map is generated: the water depth in the area (X = 30 - 50) is 2.1 - 3.2 meters, and the water depth in the area (X = 50 - 70) is 1.5 - 2.0 meters. Then, the system draws the riverbed cross-section curve according to the parabolic formula y = 0.02x², and divides the flow measurement area into 6 cross-sections. Each cross-section is further stratified at 0.5-meter intervals, generating a total of 12 flow velocity measurement layers. After combining the water surface fluctuation data with the initial flow velocity distribution map, the average cross-section flow velocity is calculated as 0.87 m / s. For example, in the third cross-section (X = 45 - 50), the flow velocity of the water surface layer is 1.2 m / s, the middle layer is 0.9 m / s, and the bottom layer is 0.6 m / s. After weighted averaging, the flow velocity of this cross-section is 0.92 m / s. After integrating all the cross-section data, the system outputs the average flow velocity of the entire basin as 0.84 m / s, and marks (X = 55, Y = 58) as a flow velocity anomaly point (1.8 m / s), indicating that there may be an undercurrent.

[0034] The steps of calling the drone to release a buoy group, collecting the movement trajectories and gyroscope data of the buoy group, and correcting the three-dimensional water flow model based on the movement trajectories, gyroscope data, and cross-section average flow velocity are as follows: Call the drone to release a buoy group, and the buoy group moves on the water surface of the target flow measurement area, collecting the movement trajectories and gyroscope data of the buoy group; Based on the movement trajectories, obtain the basic water flow direction of the target flow measurement area, and based on the gyroscope data, obtain the water flow resistance data of the target flow measurement area; Construct an underwater water flow model in the three-dimensional water flow model based on the cross-section average flow velocity; Based on the basic water flow direction and water flow resistance data, refine and correct the water surface flow in the three-dimensional water flow model.

[0035] In use, taking the target flow measurement area of ​​a river as an example, the drone released 20 buoys, 15 of which drifted in the southeast direction and 5 were stranded due to vortices. The buoy trajectory shows that the main water flow direction is 12 degrees south-east, with an average drift speed of 1.1 meters per second. Gyroscope data shows that the buoy is subjected to the largest impact force (6.2N) of the lateral water flow in the area (X=50, Y=60), causing the motion trajectory to deviate by 20 degrees. The system inputs these data into the three-dimensional water flow model, first correcting the flow direction of the water surface layer to 12 degrees south-east, and adjusting the flow velocity in the area from the model-predicted 1.3 meters per second to 1.15 meters per second. Then, based on the average flow velocity of the cross section of 0.84 meters per second, the flow velocity gradients of each underwater layer are constructed: 0.9 meters per second for the 0-1 meter layer, 0.7 meters per second for the 1-2 meter layer, and 0.5 meters per second for the 2-3 meter layer. Finally, for the abnormally high-speed area (X=55, Y=58), the model increases the local turbulence coefficient and corrects the flow velocity at this point to 1.6 m / s. The corrected three-dimensional water flow model shows that the maximum flow rate of the entire basin is 920 cubic meters per second, an increase of 8% over the initial model. All data are transmitted wirelessly to the control center to generate a final flow measurement report containing a flow velocity thermal map and dangerous area markings.

[0036] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for measuring water flow velocity of an unmanned aerial vehicle, characterized in that, It includes the following steps: Obtain the target flow measurement area of the water area to be detected, call the drone to fly to the target flow measurement area, select a dynamic anchor point in the target speed measurement area, and perform anchor point hovering on the drone according to the dynamic anchor point; The drone collects the surrounding images of itself, obtains obstacle data based on the surrounding images, and generates a movement route in real time according to the obstacle data and the dynamic anchor point; The drone performs video collection on the target speed measurement area, obtains a visible light water surface video, an infrared light water surface video, and a polarized light water surface video, fuses them to generate a three-band water surface video, and analyzes the three-band water surface video to generate an initial flow velocity distribution map; The drone collects the riverbed data and water surface fluctuation data of the target flow measurement area, obtains the riverbed topography based on the riverbed data, and obtains the water depth data and the cross-sectional average flow velocity based on the riverbed topography and the water surface fluctuation data; Based on the riverbed topography, the water depth data, and the initial flow velocity distribution map, construct a three-dimensional water flow model, call the drone to release a buoy group, collect the movement trajectories and gyroscope data of the buoy group, and correct the three-dimensional water flow model according to the movement trajectories, the gyroscope data, and the cross-sectional average flow velocity, and output the hydrological flow measurement and speed measurement results.

2. The method for hydrological flow measurement and velocity measurement of an unmanned aerial vehicle according to claim 1, wherein The step of selecting a dynamic anchor point in the target speed measurement area and performing anchor point hovering on the drone according to the dynamic anchor point is specifically as follows: Call the drone to perform image collection on the target speed measurement area to obtain a watershed image, extract water flow information from the watershed image to obtain a water flow image, and particleize the water flow image to obtain water flow image particles Based on the water flow image particles, extract multiple particle features in the water flow image particles, and record the obviousness value of each particle feature; Select the particle feature with the highest obviousness value as the target particle feature, and perform anchor point selection on the water flow image particles according to the target particle feature to obtain a dynamic anchor point; Take pictures of the dynamic anchor point at a preset time interval, and record the displacement of the dynamic anchor point in two pictures, and obtain the hovering speed according to the displacement and the time interval; Adjust the propeller power distribution of the drone according to the hovering speed so that the drone remains relatively stationary with the dynamic anchor point.

3. The method for measuring the water flow velocity of a drone according to claim 2, characterized in that, The step of the drone collecting the surrounding images of itself, obtaining obstacle data based on the surrounding images, and generating a movement route in real time according to the obstacle data and the dynamic anchor point is specifically as follows: The drone performs image collection on the surrounding environment of itself to obtain surrounding images, and discriminates the content of the surrounding images to obtain obstacle images; Extract the visual distance feature and the external shape feature of the obstacle image, obtain the obstacle distance based on the visual distance feature, obtain the obstacle spatial position based on the external shape feature, and generate initial obstacle data by combining the obstacle distance and the obstacle spatial position; Predict obstacles outside or after the peripheral image based on the initial obstacle data and the external shape features to obtain predicted obstacle data, and add the predicted obstacle data to the initial obstacle data to generate obstacle data; Obtain the current drone spatial position of the drone and the current anchor spatial position of the dynamic anchor, and generate the movement route of the drone in real time in combination with the obstacle data, the drone spatial position, and the anchor spatial position.

4. The method for measuring water flow velocity of an unmanned aerial vehicle according to claim 3, wherein, After the step of generating the movement route of the drone in real time in combination with the obstacle data, the drone spatial position, and the anchor spatial position, it further includes: When the drone moves on the movement route, update the peripheral image in real time and update the obstacle image in real time to obtain an updated obstacle image; Obtain updated obstacle data based on the updated obstacle image, and determine whether the movement route is intercepted by the obstacle based on the updated obstacle data; If it is determined that the movement route is intercepted by the obstacle, extract the intercepted route segment in the movement route; Based on the route segment and the updated obstacle data, reconstruct the route segment to obtain an updated route segment, and add the updated route segment to the movement route.

5. A method for measuring water flow velocity of an unmanned aerial vehicle according to claim 1, characterized in that, Before the step of fusing the visible light water surface video, the infrared light water surface video, and the polarized light water surface video to generate a three-band water surface video, it further includes: Extract the light reflection area in the visible light water surface video, and extract the light amount of the light reflection area. Perform exposure reduction processing on the light reflection area according to the light amount to obtain a target light reflection area; Add the target light reflection area to the visible light water surface video to obtain a target visible light water surface video; Based on the infrared light water surface video, extract the temperature distribution area and the infrared light scattering amount of the water surface, and perform clarity processing on the infrared light water surface video based on the temperature distribution area and the infrared light scattering amount to obtain a target infrared light water surface video; Based on the polarized light water surface video, extract the polarization state change of the water surface, generate polarization compensation parameters according to the polarization state change, and perform parameter compensation on the polarized light water surface video according to the polarization compensation parameters to obtain a target polarized light water surface video.

6. The method for measuring water flow velocity of an unmanned aerial vehicle according to claim 5, wherein The step of fusing the visible light water surface video, the infrared light water surface video, and the polarized light water surface video to generate a three-band water surface video and analyzing the three-band water surface video to generate an initial flow velocity distribution map is specifically: Fuse the target visible light water surface video, the target infrared light water surface video, and the target polarized light water surface video to generate a three-band water surface video; Based on the target infrared light water surface video, eliminate the water mist influence and temperature influence appearing in the three-band water surface video; Based on the target polarized light water surface video, enhance the content clarity of the underwater image in the three-band water surface video; Perform water surface frame recognition on the three-band water surface video to obtain water surface video frames, and record the visible light water surface difference and the infrared light water surface difference between the water surface video frames; Mutually verify and identify the visible light water surface difference and the infrared light water surface difference to obtain an initial water surface velocity distribution; Perform underwater frame recognition on the three-band water surface video to obtain underwater video frames, and record the water ripple shape difference and water ripple color difference between the underwater video frames; Based on the water ripple shape difference and water ripple color difference, obtain the underwater water flow change characteristics, and based on the underwater water flow change characteristics, obtain the initial underwater velocity distribution; Generate an initial velocity distribution map of the target flow measurement area according to the initial water surface velocity distribution and the initial underwater velocity distribution; 7. A method for measuring water flow velocity of an unmanned aerial vehicle according to claim 6, characterized in that, The step of eliminating the water mist influence and temperature influence occurring in the three-band water surface video based on the target infrared light water surface video is specifically as follows: Extract the infrared frame images in the target infrared light water surface video, and obtain the infrared temperature distribution and the water surface infrared characteristics based on the infrared frame images; Locate the water surface boundary based on the water surface infrared characteristics to obtain water surface boundary data; Perform layout segmentation on the infrared temperature distribution according to the water surface boundary data to respectively obtain the water surface temperature distribution and the above-water temperature distribution; Based on the above-water temperature distribution, obtain the water mist influence in the target flow measurement area, and based on the water surface temperature distribution, obtain the temperature influence in the target flow measurement area; Generate a water mist image compensation based on the water mist influence, generate a temperature image compensation based on the temperature influence, and eliminate the water mist influence and the temperature influence in the three-band water surface video according to the water mist image compensation and the temperature image compensation; 8. A method for hydrological flow measurement and velocity measurement of an unmanned aerial vehicle according to claim 7, characterized in that, The step of obtaining the riverbed topography according to the riverbed data and obtaining the water depth data and the cross-sectional average velocity according to the riverbed topography and the water surface fluctuation data is specifically as follows: Model the riverbed of the target flow measurement area according to the riverbed data to obtain the riverbed topography; According to the water surface fluctuation data, obtain the average water surface spatial position of the target flow measurement area, and obtain the water depth data according to the riverbed topography and the average water surface spatial position; According to the riverbed topography, obtain the cross-sectional spatial curve of the riverbed, and based on the cross-sectional spatial curve and the water depth data, construct the watershed cross-sectional image of the target flow measurement area; Based on the water surface fluctuation data and the water depth data, divide a plurality of deep velocity relationship diagrams on the watershed cross-sectional image; Construct the cross-sectional velocity data on the watershed cross-sectional image according to the initial velocity distribution map and the deep velocity relationship diagrams, and obtain the cross-sectional average velocity according to the cross-sectional velocity data; 9. A method for hydrological flow measurement and velocity measurement of an unmanned aerial vehicle according to claim 8, characterized in that, The step of calling a drone to release a buoy group, collecting the movement trajectories and gyroscope data of the buoy group, and correcting the three-dimensional water flow model according to the movement trajectories, the gyroscope data, and the cross-sectional average velocity is specifically as follows: Call a drone to release a buoy group, and the buoy group moves on the water surface of the target flow measurement area, and collect the movement trajectories and gyroscope data of the buoy group; Based on the movement trajectories, obtain the basic water flow direction of the target flow measurement area, and based on the gyroscope data, obtain the water flow confrontation data of the target flow measurement area; Construct the underwater flow model in the three-dimensional flow model based on the cross-section average flow velocity; Refine and correct the surface flow in the three-dimensional flow model based on the basic flow direction and the flow resistance data.

Images