A UAV system for river channel monitoring and its autonomous patrol method

Abstract

The present invention discloses a drone system for river channel monitoring and an autonomous line inspection method thereof. The drone system comprises a drone body, on which an embedded industrial computer and a lithium battery module are arranged, as well as a GPS module electrically connected to the embedded industrial computer, a drone flight control module, a fixed attitude detection module, a WIFI communication module, a radio station module, an embedded industrial computer, a two-dimensional pan-tilt camera, a depth camera, an infrared thermal imaging camera and four rotor modules; the lithium battery module supplies power to the above modules; the four rotor modules have the same structure and are evenly distributed along the periphery of the drone body. The drone system has a simple structure and is equipped with a variety of sensor modules, which can realize multi-task inspection. The present invention designs a vision-based drone autonomous line inspection method, which can realize a stable autonomous line inspection flight of the drone along the boundary line between the river channel and the river bank.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicles, and in particular relates to an unmanned aerial vehicle system for river channel monitoring and an autonomous line patrol method thereof. Background Art

[0002] River inspection is an important means to maintain river cleanliness and ensure river safety. The quality and efficiency of river inspection are crucial to river ecological protection. In the process of river inspection and management, it is necessary to grasp the dynamic changes of the river, river blockage, illegal discharge of pollutants, and monitor and warn of river disasters that may be caused by weather changes in real time.

[0003] At present, river patrols are generally conducted by staff on foot or in patrol cars using equipment. However, manual inspections require a lot of manpower. If you want to achieve fast inspections, manual inspections often lead to problems such as missed inspections or misjudgments.

[0004] Based on this, some researchers have proposed using unmanned patrol boats to inspect rivers. However, since unmanned patrol boats are traveling in rivers, the inspection area is small and the inspection efficiency is relatively low. With the development of drone control technology, the application field of drones has been continuously expanded, and more and more research on drones has been conducted at home and abroad. UAVs are equipped with visual sensors, laser sensors and other equipment to detect pollutants in rivers or conduct biological sampling and inspection in rivers. However, the current drone system structure is relatively complex, and the drone system used for inspection is mainly controlled by manual remote control. The control proficiency of the operator will also affect the effect of river monitoring. Since there are no obvious markers in the river, functions similar to lane tracking cannot be achieved. Summary of the invention

[0005] The present invention provides an unmanned aerial vehicle system for river channel monitoring and an autonomous line patrol method thereof, which overcome the technical defects and shortcomings of the prior art in unmanned aerial vehicle systems for river channel patrol.

[0006] In order to solve the above technical problems, the present invention adopts the following solutions:

[0007] A drone system for river channel monitoring comprises a drone body, on which an embedded industrial computer and a lithium battery module are arranged, as well as a GPS module electrically connected to the embedded industrial computer, a drone flight control module, a fixed attitude detection module, a WIFI communication module, a radio station module, an embedded industrial computer, a two-dimensional pan-tilt camera, a depth camera, an infrared thermal imaging camera and four rotor modules; the lithium battery module supplies power to the above modules; the four rotor modules have the same structure and are evenly distributed along the periphery of the drone body.

[0008] Further optimized, the drone body adopts a layered structure design, including a first layer, a second layer, a third layer, a fourth layer and a fifth layer; wherein the first layer is fixedly connected to the third layer through 8 long pillars, and the first layer is located above the third layer; the second layer is fixedly connected to the third layer through 4 short pillars, and the second layer is located between the first and third layers.

[0009] The third layer board and the fourth layer board are fixed together, and the third layer board is located above the fourth layer board, and the fifth layer board and the fourth layer board are fixedly connected by short pillars.

[0010] Further optimization, the GPS module is fixed on the upper surface of the first layer of board, the UAV flight control module is fixed on the upper surface of the second layer of board, and the attitude detection module, WIFI communication module and radio station module are fixed on the upper surface of the third layer of board; among them, the attitude detection module is mainly used to detect the three-dimensional attitude of the UAV, the WIFI communication module is used to transmit inspection videos, and the radio station module is used to communicate with the operator controller.

[0011] The embedded industrial computer is fixedly installed on the lower surface of the fourth layer board, and the lithium battery module is fixedly installed on the upper surface of the fifth layer board; the two-dimensional pan-tilt camera is fixedly installed on the lower surface of the fifth layer board for realizing river channel monitoring; the depth camera and the infrared thermal imaging camera are fixed on the fifth layer board through a bent mounting plate, and the shooting angle is tilted downward at 45 degrees; the depth camera is used to realize drone line patrol monitoring, and the infrared thermal imaging camera is used to detect the river water temperature.

[0012] Further optimized, the rotor module includes a rotor support rod, two drone support legs, a brushless motor and a propeller; wherein the rotor support rod includes an upper and a lower support plate, the two support plates are fixedly connected as a whole at one end away from the drone body, and are connected at one end close to the drone body, and both are provided with mounting holes;

[0013] The four corners of the third and fourth layers are provided with mounting bosses, the mounting holes of the two support plates are matched with the corresponding mounting bosses and are fixedly connected by bolts; the two UAV support legs are fixedly installed on both sides of the rotor support rod, and the brushless motor and propeller are fixedly installed on the end of the rotor support rod away from the UAV body.

[0014] A vision-based UAV river autonomous patrol method, based on the above-mentioned UAV system, includes the following steps:

[0015] S1: The depth camera is calibrated by Zhang Zhengyou's chessboard to eliminate or reduce the impact of camera distortion on line patrol detection;

[0016] S2: Preprocess the image captured by the depth camera; select the color space for the original image captured by the camera, because the HSV space is the most consistent with the human eye's visual color perception, and the HSV space is a uniform color space. H represents hue, S represents saturation, and V represents brightness. Analyze the image in the HSV space and separate the H, S, and V channel images. Since the color contrast between the river and the shore is sharp, and the H channel component is less affected by light, select the H channel image to perform the operation as in step 3.

[0017] S3: Segmentation of images captured by the depth camera:

[0018] The image of the H channel obtained in step S2 is segmented using a K-means-based clustering algorithm. Since the segmented area is the river water and a single bank, the number of clusters is selected as 2, and the river water and the bank in the image are segmented into two categories, the black area is the bank, and the white area is the river water;

[0019] S4: performing noise elimination on the image segmented in the above step S3, and eliminating the noise formed by these erroneous clustering by erosion and expansion operations in image processing;

[0020] The K-means clustering segmentation in the H channel has a good segmentation effect, and the shore and river water have been segmented into two categories. However, due to the shadows formed by the illumination, some areas in the river water are incorrectly clustered as shore areas. The noise formed by these incorrect clustering is eliminated through the erosion and expansion operations in image processing.

[0021] S5: Use the contour extraction algorithm in OpenCV to extract the contour line of the boundary between the bank and the river water, and provide the outer contour and the inner contour;

[0022] S6: Use the straight line model to scan the extracted contour line to obtain the two edge points required for edge detection. The middle value of the edge point coordinates is the coordinate of the drone patrol line;

[0023] S7: The line coordinates obtained by the straight line scan are used as the center of the ellipse scan, and then the line coordinates obtained by the ellipse scan are used as the center of the next ellipse scan. A straight line scan is performed once for each frame of the image, and six ellipse scans are iterated to obtain the accurate coordinate position of the center of the contour line, so that the dynamic shore line can be recognized;

[0024] S8: Execute steps S2 to S7 for each frame of image, and six coordinate values ​​can be obtained for each frame of image; when the next image is acquired, steps S2 to S7 will be repeatedly executed.

[0025] Further optimization, in step S3, the K-means clustering algorithm is specifically as follows:

[0026] S31: Select 2 points as initial cluster centers C = {c1, c2};

[0027] S32: Calculate the distance between the pixel point sj and the cluster center ci. If the distance is less than the threshold 1, the pixel point sj is classified into the cluster ci.

[0028] S33: recalculate the cluster center of class ci;

[0029] S34: Iterate and repeat the above S32 and S33 until the cluster center remains unchanged and then stop the calculation.

[0030] Further optimization, in step S6, the linear model adopts a hierarchical scanning method, and the specific steps are as follows:

[0031] S61: Perform a linear scanning model on the image, scan the entire row of pixels, and obtain the grayscale value of the pixels in this row;

[0032] S62: Differentiate the grayscale values ​​of adjacent pixels to obtain a maximum value and a minimum value, and the coordinates of these two values ​​are the line edge coordinates;

[0033] S63: The midpoint of the two edge coordinates is used as the coordinate of the drone line patrol.

[0034] Further optimization, in step S7, the specific operation of the ellipse scanning model is as follows:

[0035] S71: Using the line center coordinates obtained by straight line scanning as the center of the ellipse scanning;

[0036] S72: To improve scanning efficiency, pixel values ​​of the upper half of the ellipse are obtained at intervals of 5°, and values ​​of 36 pixel points can be obtained by scanning 180°;

[0037] S73: Differencing the adjacent pixel values ​​on the half circumference of the ellipse, a maximum value and a minimum value can be obtained, which are the coordinates of the two edges of the target line;

[0038] S74: The 36 pixel values ​​obtained by scanning are stored in an array in sequence. When the edge coordinates are obtained, the index position of the edge coordinates in the entire array can also be known. The coordinates of the point at the middle value of the two index positions are approximately used as the center coordinates of the target line. The UAV calculates the flight trajectory based on the coordinate values.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. The drone system of the present invention has a simple structure, is equipped with a variety of modules, can realize multi-task inspection, and can provide a stable and accurate drone autonomous line patrol function.

[0041] 2. In the present invention, since the depth camera on the drone is installed at the center line position of the front end of the drone, if the coordinates in the image are converted into actual coordinates for drone control, a relatively large conversion error will be introduced. Therefore, the drone in the present invention directly uses the image collected by the depth camera for line patrol control, which reduces the error and improves the accuracy.

[0042] 3. The present invention uses a typical closed-loop controller as the drone line patrol controller, and the closed-loop controller runs in the embedded industrial computer carried by the drone. In order to ensure the stable line patrol flight of the drone, the line patrol coordinates are required to always remain on the center line of the image. Therefore, the horizontal axis coordinate of the center line is used as the input value of the closed-loop controller, and the horizontal axis coordinate of the target line after image processing is used as the feedback value of the closed-loop controller. The rotation speed of the four rotors of the drone is controlled according to the interpolation of the two. When the closed-loop controller reaches stability, the drone can patrol along the boundary line between the river bank and the river channel. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the three-dimensional structure of the UAV system of the present invention;

[0044] Figure 2 It is a schematic diagram of the three-dimensional structure of the drone body of the present invention;

[0045] Figure 3 This is a schematic diagram of the structure of the right side of the drone body of the present invention;

[0046] Figure 4 A top view of the drone body of the present invention;

[0047] Figure 5 This is a schematic diagram of the rotor module structure of the present invention;

[0048] Figure 6 This is a flow chart of the vision-based autonomous river patrol method of a UAV according to the present invention;

[0049] Figure 7 This is a block diagram of the line patrol controller of the drone system described in the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are 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.

[0051] Embodiment 1:

[0052] See also Figure 1 A drone system for river channel monitoring mainly comprises a drone body 1 and four rotor modules (2, 3, 4, 5). The drone body 1 adopts a layered structure and is mainly used for fixing and installing a variety of sensors. The four rotors (2, 3, 4, 5) are installed in a diagonal form and distributed around the drone body 1.

[0053] See also Figure 2 The structure of the drone body 1 of the present invention is composed of a first layer 101, a second layer 102, a third layer 103, a fourth layer 104 and a fifth layer 105. The first layer 101 is fixed to the third layer 103 through 8 long pillars, the second layer 102 is fixed to the third layer 103 through 4 short pillars, and the second layer 102 is between the first layer 101 and the third layer 103. The third layer 103 is fixed to the fourth layer 104, and the fifth layer 105 is connected to the fourth layer 104.

[0054] See also Figure 2 , Figure 3 , Figure 4 , GPS module 111 is fixed on the upper surface of the first layer board 101, UAV flight control module 116 is fixed on the upper surface of the second layer board 102, and attitude detection module 112, WIFI communication module 107 and radio module 115 are fixed on the third layer board 103, wherein attitude detection module 112 is mainly used to detect the three-dimensional attitude of the UAV, WIFI communication module 107 is used to transmit inspection video, and radio module 115 is used to communicate with the operator controller. The above three modules are fixedly installed on the upper surface of the third layer board.

[0055] The embedded industrial computer 108 is fixedly mounted on the lower surface of the fourth layer 104, and the lithium battery module 113 is fixedly mounted on the upper surface of the fifth layer 105 for powering the drone. A two-dimensional pan-tilt camera 106 is fixedly mounted on the lower surface of the fifth layer 105, which is mainly used to monitor the river. The depth camera 110 and the infrared thermal imaging camera 109 are fixed to the front end of the fifth layer 105 through a 45-degree bent mounting plate 114. The depth camera 110 is mainly used to monitor the drone patrol line, and the infrared thermal imaging camera 109 is used to detect the river water temperature, etc.

[0056] See also Figure 5The four rotor modules (2, 3, 4, 5) have the same structure. Taking rotor module 2 as an example, it mainly includes a rotor support rod 203, two drone support legs (201, 202), a brushless motor 204 and a propeller 205. The rotor support rod 203 is sandwiched by the bosses on the four corners of the third layer 103 and the fourth layer 104, the two drone support legs 201, 202 are fixedly installed on both sides of the rotor support rod 203, and the brushless motor 204 and the propeller 205 are fixedly installed at the end of the rotor support rod 203.

[0057] Embodiment 2:

[0058] Generally, there is a clear dividing line between the river channel and the river bank. Therefore, based on the above-mentioned UAV system, a vision-based UAV river autonomous patrol method includes the following steps, see Figure 6 :

[0059] S1: The depth camera 110 needs to be calibrated by Zhang Zhengyou's chessboard to eliminate or reduce the impact of camera distortion on line patrol detection;

[0060] S2: Preprocess the image captured by the depth camera 110. Select the color space for the original image captured by the camera, because the HSV space is the most consistent with the human eye's visual color perception, and the HSV space is a uniform color space. H represents hue, S represents saturation, and V represents brightness. Analyze the image in the HSV space, and separate the H, S, and V channel images. Since the color contrast between the river water and the shore is sharp, and the H channel component is less affected by light, select the H channel image to perform the operation as in step three.

[0061] S3: Segmentation of the image captured by the depth camera 110. Segment the image of the H channel obtained in the above step 2. Perform image segmentation on the image of the H channel based on the K-means clustering algorithm. Since the segmented area is the river water and a single bank, the number of clusters is selected as 2, and the river water and the bank in the image are divided into two categories, the black area is the bank, and the white area is the river water. The K-means clustering algorithm is specifically as follows:

[0062] S31: Select 2 points as the initial cluster center C = {c1, c2};

[0063] S32: Calculate the distance between the pixel point sj and the cluster center ci. If the distance is less than the threshold 1, the pixel point sj is classified into the cluster ci.

[0064] S33: recalculate the cluster center of class ci;

[0065] S34: Iterate and repeat the above (2) and (3) until the cluster center remains unchanged and then stop calculating;

[0066] S4: The segmented image is filtered. The image segmented in step 3 above is subjected to noise elimination. K-means clustering segmentation in the H channel has a good segmentation effect, and the shore and river water have been segmented into two categories. However, due to the shadows formed by the illumination, some areas in the river water are incorrectly clustered into the shore area. The noise formed by these incorrect clustering is eliminated through the erosion and dilation operations in image processing.

[0067] S5: Use the contour extraction algorithm in OpenCV to extract the contour line of the boundary between the bank and the river water to provide the outer contour and the inner contour.

[0068] S6: Use the straight line model to scan the extracted contour line to obtain the two edge points required for edge detection. The middle value of the edge point coordinates is the coordinate of the drone patrol line. The straight line model adopts a hierarchical scanning method. The specific steps are as follows:

[0069] S61: Perform a linear scanning model on the image, scan the entire row of pixels, and obtain the grayscale value of the pixels in this row;

[0070] S62: Differentiate the grayscale values ​​of adjacent pixels to obtain a maximum value and a minimum value, and the coordinates of these two values ​​are the line edge coordinates;

[0071] S63: The midpoint of the two edge coordinates is used as the coordinate of the drone patrol line;

[0072] S7: To improve the efficiency of line inspection, elliptical scanning is performed on the basis of linear scanning. The line coordinates obtained by linear scanning are used as the center of elliptical scanning, and then the line coordinates obtained by elliptical scanning are used as the center of the next elliptical scanning. A linear scan is performed on each frame of image and six elliptical scans are iterated to obtain the accurate coordinate position of the center of the contour line, so as to realize the recognition of dynamic shore line. The specific operation of the elliptical scanning model is as follows:

[0073] S71: Using the line center coordinates obtained by straight line scanning as the center of the ellipse scanning;

[0074] S72: To improve scanning efficiency, pixel values ​​of the upper half of the ellipse are obtained at intervals of 5°, and values ​​of 36 pixel points can be obtained by scanning 180°;

[0075] S73: Differencing the adjacent pixel values ​​on the half circumference of the ellipse, a maximum value and a minimum value can be obtained, which are the coordinates of the two edges of the target line;

[0076] S74: The 36 pixel values ​​obtained by scanning are stored in an array in sequence. When the edge coordinates are obtained, the index position of the edge coordinates in the entire array can also be known. The coordinates of the point at the middle value of the two index positions are approximately used as the center coordinates of the target line. The UAV calculates the flight trajectory based on the coordinate values.

[0077] S8: Steps 2 to 7 are executed for each frame of image, and six coordinate values ​​can be obtained for each frame of image. When the next image is acquired, steps 2 to 7 are repeated.

[0078] See also Figure 7 , since the depth camera 110 on the drone is installed at the center line position of the front end of the drone body 1, if the coordinates in the image are converted into actual coordinates for drone control, a relatively large conversion error will be introduced. Therefore, the drone in the present invention directly uses the image collected by the depth camera 110 for line patrol control. A typical closed-loop controller is used as the drone line patrol controller, and the closed-loop controller runs in the embedded industrial computer 108 carried by the drone. In order to ensure the stable line patrol flight of the drone, the line patrol coordinates are required to always remain on the center line of the image. Therefore, the horizontal axis coordinate of the center line is used as the input value of the closed-loop controller, and the horizontal axis coordinate of the target line after image processing is used as the feedback value of the closed-loop controller. The rotation speed of the four rotors of the drone is controlled according to the interpolation of the two. When the closed-loop controller reaches stability, the drone can patrol along the boundary line between the river bank and the river channel.

[0079] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A drone system for river monitoring, characterized in that: The invention comprises a drone body, on which an embedded industrial computer and a lithium battery module are arranged, as well as a GPS module electrically connected to the embedded industrial computer, a drone flight control module, a fixed attitude detection module, a WIFI communication module, a radio station module, an embedded industrial computer, a two-dimensional pan-tilt camera, a depth camera, an infrared thermal imaging camera and four rotor modules; the lithium battery module supplies power to the above modules; the four rotor modules have the same structure and are evenly distributed along the periphery of the drone body; The drone body adopts a layered structure design, including a first layer, a second layer, a third layer, a fourth layer and a fifth layer; wherein the first layer is fixedly connected to the third layer through 8 long struts, and the first layer is located above the third layer; the second layer is fixedly connected to the third layer through 4 short struts, and the second layer is located between the first layer and the third layer; The third layer board and the fourth layer board are fixed together, and the third layer board is located above the fourth layer board, and the fifth layer board and the fourth layer board are fixedly connected by short pillars; The GPS module is fixed on the upper surface of the first layer, the UAV flight control module is fixed on the upper surface of the second layer, and the attitude detection module, the WIFI communication module and the radio station module are fixed on the upper surface of the third layer; wherein the attitude detection module is mainly used to detect the three-dimensional attitude of the UAV, the WIFI communication module is used to transmit the inspection video, and the radio station module is used to communicate with the operator controller; The embedded industrial computer is fixedly installed on the lower surface of the fourth layer board, and the lithium battery module is fixedly installed on the upper surface of the fifth layer board; the two-dimensional pan-tilt camera is fixedly installed on the lower surface of the fifth layer board for realizing river channel monitoring; the depth camera and the infrared thermal imaging camera are fixed on the fifth layer board through a bent mounting plate, and the shooting angle is tilted downward at 45 degrees; the depth camera is used to realize drone line patrol monitoring, and the infrared thermal imaging camera is used to detect the river water temperature.

2. The UAV system for river monitoring according to claim 1, characterized in that: The rotor module comprises a rotor support rod, two drone support legs, a brushless motor and a propeller; wherein the rotor support rod comprises an upper support plate and a lower support plate, the two support plates are fixedly connected as a whole at one end away from the drone body, and are connected at one end close to the drone body, and both are provided with mounting holes; The four corners of the third and fourth layers are provided with mounting bosses, the mounting holes of the two support plates are matched with the corresponding mounting bosses and are fixedly connected by bolts; the two UAV support legs are fixedly installed on both sides of the rotor support rod, and the brushless motor and propeller are fixedly installed on the end of the rotor support rod away from the UAV body.

3. A vision-based UAV river autonomous patrol method, characterized in that: The UAV system according to any one of claims 1 to 2 comprises the following steps: S1: The depth camera is calibrated by Zhang Zhengyou's chessboard to eliminate or reduce the impact of camera distortion on line patrol detection; S2: preprocess the image captured by the depth camera; Select the color space of the original image collected by the camera, analyze the image in the HSV space, and separate the H, S, and V channel images; S3: Segmentation of images captured by the depth camera: The image of the H channel obtained in step S2 is segmented using a K-means-based clustering algorithm. Since the segmented area is the river water and a single bank, the number of clusters is selected as 2, and the river water and the bank in the image are segmented into two categories, the black area is the bank, and the white area is the river water; S4: performing noise elimination on the image segmented in the above step S3, and eliminating the noise formed by these erroneous clustering by erosion and dilation operations in image processing; S5: Use the contour extraction algorithm in OpenCV to extract the contour line of the boundary between the bank and the river water, and provide the outer contour and the inner contour; S6: Use the straight line model to scan the extracted contour line to obtain the two edge points required for edge detection. The middle value of the edge point coordinates is the coordinate of the drone patrol line; S7: The line coordinates obtained by the straight line scan are used as the center of the ellipse scan, and then the line coordinates obtained by the ellipse scan are used as the center of the next ellipse scan. A straight line scan is performed once for each frame of the image, and six ellipse scans are iterated to obtain the accurate coordinate position of the center of the contour line, so that the dynamic shore line can be recognized; S8: Execute steps S2 to S7 for each frame of image, and six coordinate values ​​can be obtained for each frame of image; when the next image is acquired, steps S2 to S7 will be repeatedly executed.

4. The vision-based UAV river autonomous patrol method according to claim 3 is characterized in that: In step S3, the K-means clustering algorithm is specifically as follows: S31: Select 2 points as initial cluster centers ; S32: Calculate the distance between the pixel point sj and the cluster center ci. If the distance is less than the threshold 1, the pixel point sj is classified into the cluster ci. S33: recalculate the cluster center of class ci; S34: Iterate and repeat the above S32 and S33 until the cluster center remains unchanged and then stop the calculation.

5. The vision-based UAV river autonomous patrol method according to claim 4 is characterized in that: In step S6, the linear model adopts a hierarchical scanning method. The specific steps are as follows: S61: Perform a linear scanning model on the image, scan the entire row of pixels, and obtain the grayscale value of the pixels in this row; S62: Differentiate the grayscale values ​​of adjacent pixels to obtain a maximum value and a minimum value, and the coordinates of these two values ​​are the line edge coordinates; S63: The midpoint of the two edge coordinates is used as the coordinate of the drone line patrol.

6. The vision-based UAV river autonomous patrol method according to claim 5 is characterized in that: In step S7, the specific operation of the ellipse scanning model is as follows: S71: Using the line center coordinates obtained by straight line scanning as the center of the ellipse scanning; S72: Obtain pixel values ​​of the upper half of the ellipse at intervals of 5°, and scan 180° to obtain values ​​of 36 pixel points; S73: Differencing the adjacent pixel values ​​on the half circumference of the ellipse, a maximum value and a minimum value can be obtained, which are the coordinates of the two edges of the target line; S74: The 36 pixel values ​​obtained by scanning are stored in an array in sequence. When the edge coordinates are obtained, the index position of the edge coordinates in the entire array can also be known. The coordinates of the point at the middle value of the two index positions are approximately used as the center coordinates of the target line. The UAV calculates the flight trajectory based on the coordinate values.

Images