UAV Video Flow Measurement Device Based on Image Recognition

Abstract

The invention discloses an unmanned aerial vehicle (UAV) video flow measurement device based on image recognition, which relates to the technical field of river water flow monitoring. The device includes a UAV body, on which an image acquisition module, an image recognition module and a control module are arranged, and the UAV body, the image acquisition module, the image recognition module and the control module are electrically connected. The image acquisition module is used to capture the water surface image of the river channel, the image recognition module is used to preprocess the water surface image and extract feature information, and a flow prediction model is preset in the control module, which can predict the flow of the river channel according to the feature information. The UAV video flow measurement device based on image recognition proposed by the invention can use the image recognition module to identify feature points such as water ripples in the video, extract the texture map of the feature points by frame-by-frame extraction of the video, then calculate the gradient of the texture map to obtain the water flow velocity, and finally calculate the flow of the river channel by combining information such as the width and depth of the river channel.

Description

Technical Field

[0001] The present invention relates to the technical field of river water flow monitoring, and more specifically, to an unmanned aerial vehicle (UAV) video flow measurement device based on image recognition. Background Art

[0002] There are numerous rivers in our country, and the comprehensive utilization of rivers plays an important role in the national economic and social development.

[0003] However, traditional river flow monitoring methods have problems such as low monitoring efficiency, high cost, and small coverage. Although UAVs have been applied in river flow monitoring, the degree of automation and monitoring efficiency still need to be further improved. Summary of the Invention

[0004] To overcome the defects of the prior art, the present invention proposes an unmanned aerial vehicle (UAV) video flow measurement device based on image recognition. After the image acquisition module acquires video images, the image recognition module can identify feature points such as water ripples in the video, and by extracting each frame of the video, a texture map can be synthesized with time as the abscissa from a number of feature points. Then, by calculating the gradient values formed by the feature points in the texture map, the flow velocity of the water can be obtained, and combined with information such as the width and depth of the river channel, the flow rate of the river channel can be calculated.

[0005] To achieve this purpose, the present invention adopts the following technical solutions:

[0006] The present invention provides an unmanned aerial vehicle (UAV) video flow measurement device based on image recognition, including a UAV body. An image acquisition module, an image recognition module, and a control module are provided on the UAV body, and the UAV body, the image acquisition module, the image recognition module, and the control module are electrically connected. The image acquisition module is used to capture the water surface image of the river channel, the image recognition module is used to preprocess the water surface image and extract feature information, and a flow rate prediction model is preset in the control module, which can predict the flow rate of the river channel according to the feature information.

[0007] In a preferred technical solution of the present invention, an attitude correction module and a height detection module are further provided on the UAV body.

[0008] In a preferred technical solution of the present invention, an auxiliary detection mechanism is further provided on the UAV body. The auxiliary detection mechanism includes a storage tank, tracer particles, and a sprinkling assembly. The storage tank is fixedly arranged at the bottom of the UAV body, the storage tank is filled with tracer particles, and a sprinkling assembly is provided on the storage tank for sprinkling the tracer particles onto the water surface.

[0009] In a preferred technical solution of the present invention, the sprinkling assembly includes a cylinder, a motor, a rotating block, a connecting rod, a spoon and a throwing pipe. The cylinder is fixedly arranged on one side of the storage box, and the cylinder is communicated with the storage box through a discharge hole. The motor is fixedly arranged on one side of the cylinder, and the output end of the motor extends into the cylinder and is connected with the rotating block. Connecting rods are circumferentially and uniformly arranged on the rotating block, the free end of the connecting rod is connected with a spoon, and the spoon abuts against the inner side wall of the cylinder. A throwing pipe is arranged on one side of the cylinder and is inclined downward along the tangent direction.

[0010] In a preferred technical solution of the present invention, a feeding port is further arranged on one side of the storage box, and a plugging block is threadedly connected to the feeding port.

[0011] In a preferred technical solution of the present invention, the tracer particles are made of a degradable material such as polylactic acid, polyadipic acid, butylene terephthalate or polybutylene succinate.

[0012] In a preferred technical solution of the present invention, a fluorescent coating is arranged on the surface of the tracer particles.

[0013] In a preferred technical solution of the present invention, the auxiliary detection mechanism further includes an illumination module. The illumination module includes a support plate, a rotating seat and an ultraviolet lamp. The support plate is fixedly arranged on the storage box, a rotating seat is arranged on the support plate, and the output end of the rotating seat is connected with an ultraviolet lamp.

[0014] In a preferred technical solution of the present invention, a positioning module is further arranged on the UAV body for positioning the position of the UAV body, and the control module is electrically connected with the positioning module.

[0015] The beneficial effects of the present invention are as follows:

[0016] 1. For a UAV video flow measurement device based on image recognition proposed by the present invention, after the image acquisition module acquires video images, the image recognition module can identify feature points such as water ripples in the video, and by extracting frames of the video one by one, a texture map can be synthesized from several feature points with time as the abscissa, and then the gradient value formed by the feature points in the texture map can be calculated to obtain the flow velocity of the water flow. Combining information such as the width and depth of the river channel, the flow rate of the river channel can be calculated, thereby improving the degree of automation, reducing the necessity of manual intervention, improving the monitoring efficiency, shortening the monitoring period, and improving the accuracy and reliability of the data.

[0017] 2. A UAV video flow measurement device based on image recognition proposed by the present invention can be widely applied to the protection and management of water resources in water areas such as river channels and lakes, and provides important data support for the protection and management of water resources. Description of the Drawings

[0018] Figure 1It is a schematic structural diagram of an unmanned aerial vehicle video flow measurement device based on image recognition provided by the specific embodiment of the present invention;

[0019] Figure 2 is Figure 1 a schematic structural diagram of the auxiliary detection mechanism in

[0020] Figure 3 It is a control principle block diagram of the device of the present invention;

[0021] Figure 4 It is a flow chart of a test method applicable to the device of the present invention.

[0022] In the figure:

[0023] 1. UAV body; 2. Image acquisition module; 3. Image recognition module; 4. Control module; 5. Attitude correction module; 6. Altitude detection module; 7. Auxiliary detection mechanism; 71. Storage tank; 72. Tracer particles; 73. Sprinkling assembly; 731. Cylinder; 732. Discharge hole; 733. Motor; 734. Rotating block; 735. Connecting rod; 736. Spoon; 737. Feeding pipe; 74. Feeding port; 75. Sealing block; 76. Lighting module; 761. Support plate; 762. Rotating seat; 763. Ultraviolet lamp; 8. Positioning module. Specific embodiment

[0024] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings and through specific embodiments.

[0025] Such as Figures 1-4As shown in the figure, in the embodiment, a drone video flow measurement device based on image recognition is provided, including a drone body 1, on which an image acquisition module 2, an image recognition module 3 and a control module 4 are arranged, and the drone body 1, the image acquisition module 2, the image recognition module 3 and the control module 4 are electrically connected. The image acquisition module 2 is used to capture the water surface image of the river channel, the image recognition module 3 is used to preprocess the water surface image and extract feature information, and a flow prediction model is preset in the control module 4, which can predict the flow of the river channel according to the feature information. In this embodiment, the drone body 1 is a multi-rotor drone with high stability and reliability. It can fly according to the instructions of the control module 4, capture the water surface image of the river channel, and a 4G gateway is arranged on the control module 4, which can perform remote wireless communication with the ground terminal, enabling users to remotely control the flight of the drone body 1 on the ground. The image acquisition module 2 is a high-definition camera, which can capture high-definition water surface images and transmit the image information to the image recognition module 3. Moreover, the image acquisition module 2 is installed on the gimbal at the bottom of the drone body 1, which can reduce the influence of device vibration on image stability. The image recognition module 3 uses deep learning algorithms to automatically recognize and analyze the captured video, extract the features of the river channel, such as width, depth, water flow direction, water level, rigid floating objects, water ripples, etc. Then, the flow prediction model calculates the flow of the river channel according to the extracted information. For example, after the image acquisition module 2 captures the video image, the image recognition module 3 recognizes the water ripple features in the video. By extracting each frame of the video, a texture map can be synthesized from several water ripple features with time as the abscissa. Finally, the gradient value formed by the water ripple characteristics in the texture map can be calculated to obtain the water flow velocity, and the flow of the river channel can be calculated by combining information such as the width and depth of the river channel. In addition, a database is also arranged in the control module 4, and different characteristic data represented by the surface water flow of the river under different weather conditions are pre-input into the database. The control module 4 compares the features extracted by the image recognition module 3 with the standard features in the database to verify the accuracy of the calculation results.

[0026] Specifically, an attitude correction module 5 and a height detection module 6 are also arranged on the drone body 1. In this embodiment, the attitude correction module 5 can obtain the attitude data of the drone, including pitch angle, roll angle and yaw angle, compare the attitude data with the preset attitude data range, and then judge whether the attitude of the drone is normal and automatically correct the attitude of the drone to ensure the accuracy of the detection results. Moreover, the attitude correction module 5 can also detect the orientation of the drone body 1, so that the front of the drone body 1 always faces the upstream of the river channel, and thus the water flow direction in the video captured by the image acquisition module 2 remains consistent. The arranged height detection module 6 is used to monitor the height of the drone body 1, and both the attitude correction module 5 and the height detection module 6 are electrically connected to the control module 4.

[0027] Specifically, an auxiliary detection mechanism 7 is further provided on the UAV body 1. The auxiliary detection mechanism 7 includes a storage tank 71, tracer particles 72, and a sprinkling assembly 73. The storage tank 71 is fixedly arranged at the bottom of the UAV body 1. The storage tank 71 is filled with tracer particles 72. A sprinkling assembly 73 is arranged on the storage tank 71 for sprinkling the tracer particles 72 onto the water surface. In this embodiment, the provided auxiliary detection mechanism 7 is used to assist the image recognition module 3 in feature extraction to improve the detection accuracy. Among them, the storage tank 71 is detachably installed at the bottom of the UAV body 1 through fastening bolts, which is convenient for disassembly and replacement. Due to the complex environment of the river water surface, the provided tracer particles 72 can float on the water surface as rigid floating objects, and then be extracted by the image recognition module 3 as river features. Moreover, the tracer particles 72 are in a granular structure and will not be entangled or adsorbed with waterweeds, etc., which helps to improve the accuracy of the flow measurement result. The provided sprinkling assembly 73 can sprinkle the tracer particles 72 towards the front side of the UAV body 1 so that they fall on the upstream of the river, thereby ensuring that the image acquisition module 2 can capture video images with tracer particles 72.

[0028] Specifically, the sprinkling assembly 73 includes a cylinder 731, a motor 733, a rotating block 734, a connecting rod 735, a spoon 736, and a throwing pipe 737. The cylinder 731 is fixedly arranged on one side of the storage tank 71, and the cylinder 731 is communicated with the storage tank 71 through a discharge hole 732. The motor 733 is fixedly arranged on one side of the cylinder 731. The output end of the motor 733 extends into the cylinder 731 and is connected to the rotating block 734. A plurality of connecting rods 735 are circumferentially and uniformly arranged on the rotating block 734. The free end of the connecting rod 735 is connected with a spoon 736, and the spoon 736 abuts against the inner side wall of the cylinder 731. A throwing pipe 737 is arranged on one side of the cylinder 731 and is inclined downward along the tangent direction. In this embodiment, the central axis of the cylinder 731 is horizontally arranged, and the discharge hole 732 is located at the bottom of the inner side wall of the cylinder 731. On the one hand, it enables the tracer particles 72 in the storage tank 71 to automatically enter the cylinder 731 through the discharge hole 732 under the action of gravity to achieve automatic feeding. On the other hand, due to the poor fluidity of the tracer particles 72, when the accumulated height of the tracer particles 72 in the cylinder 731 exceeds the height of the discharge hole 732, the discharge hole 732 will be blocked to achieve the purpose of automatically controlling the feeding amount. The motor 733 is arranged at the center of the cylinder 731 and can drive the rotating block 734 to rotate. The rotation of the rotating block 734 can drive the spoon 736 to rotate synchronously, and then throw the tracer particles 72 accumulated at the bottom of the cylinder 731 towards the front side of the UAV body 1 through the throwing pipe 737.

[0029] Specifically, a feeding port 74 is further provided on one side of the storage bin 71, and a blocking block 75 is threadedly connected to the feeding port 74. In this embodiment, the feeding port 74 is located at the top of one side of the storage bin 71, which is convenient for the staff to add the tracer particles 72 into the storage bin 71. The blocking block 75 can block the feeding port 74 after the feeding is completed to prevent water or dust from entering the storage bin 71.

[0030] Specifically, the tracer particles 72 are made of a degradable material such as polylactic acid, polyadipic acid, butylene terephthalate or polybutylene succinate. In this embodiment, the use of degradable materials can reduce environmental pollution.

[0031] Specifically, a fluorescent coating is provided on the surface of the tracer particles 72. In this embodiment, the provided fluorescent coating can produce a strong fluorescence effect when irradiated by sunlight or ultraviolet light, making the coating present a bright and vivid color, so that the tracer particles 72 can be more easily and clearly captured by the image acquisition module 2.

[0032] Specifically, the auxiliary detection mechanism 7 further includes an illumination module 76. The illumination module 76 includes a support plate 761, a rotating seat 762 and an ultraviolet lamp 763. The support plate 761 is fixedly provided on the storage bin 71. A rotating seat 762 is provided on the support plate 761, and the output end of the rotating seat 762 is connected to the ultraviolet lamp 763. In this embodiment, the provided illumination module 76 can irradiate the surface of the tracer particles 72 to make them emit light in dim light, which helps to improve the applicability of the device. Among them, the support plate 761 serves the purpose of fixedly installing the rotating seat 762. The provided rotating seat 762 can drive the ultraviolet lamp 763 to rotate to achieve a larger range of illumination; the provided ultraviolet lamp 763 can make the fluorescent coating emit light, so that the device can also perform image acquisition and flow measurement operations at dusk or at night.

[0033] Specifically, a positioning module 8 is further provided on the UAV body 1 for positioning the position of the UAV body 1. The control module 4 is electrically connected to the positioning module 8. In this embodiment, the provided positioning module 8 is used to position the UAV body 1 to achieve positioning and marking the area of river channel flow measurement, avoid repeated detection and improve work efficiency.

[0034] Working principle: During use, the operator controls the UAV body 1 to fly over the river channel. Then, the image acquisition module 2 carried on the UAV takes the water surface image of the river channel and sends the image data to the image recognition module 3. Then, the image recognition module 3 uses deep learning algorithms to automatically recognize and analyze the video taken by the UAV body 1, extract the features of the river channel, such as width, depth, water flow direction, water level, rigid floating objects, water ripples, etc. Then, the control module 4 calculates the flow rate of the river channel through hydraulic formulas based on the data obtained by the image acquisition module 2 and the recognition results of the image recognition module 3. At the same time, the sprinkling component 73 sprinkles the tracer particles 72 onto the water surface of the river channel to act as rigid floating objects, which are extracted as river channel features by the image recognition module 3 and emit light under the irradiation of sunlight or ultraviolet lamp 763, making it easier to be captured by the image acquisition module 2, so as to serve the purpose of assisting flow measurement and improving the accuracy of the detection results.

[0035] This invention is described by way of preferred embodiments. Those skilled in the art will appreciate that various changes or equivalent replacements can be made to these features and embodiments without departing from the spirit and scope of the invention. This invention is not limited by the specific embodiments disclosed herein, and other embodiments falling within the scope of the claims of this application belong to the scope of protection of this invention.

Claims

1. A drone video flow measurement device based on image recognition, comprising a drone body (1), characterized in that: An image acquisition module (2), an image recognition module (3) and a control module (4) are arranged on the drone body (1), and the drone body (1), the image acquisition module (2), the image recognition module (3) and the control module (4) are electrically connected. The image acquisition module (2) is used to capture a water surface image of a river channel, the image recognition module (3) is used to pre-process the water surface image and extract feature information, and the control module (4) is pre-installed with a flow prediction model, which can predict the flow of the river channel based on the feature information. The drone body (1) is also provided with an auxiliary detection mechanism (7), the auxiliary detection mechanism (7) comprising a storage box (71), tracer particles (72) and a throwing assembly (73), the storage box (71) being fixedly arranged at the bottom of the drone body (1), the storage box (71) containing tracer particles (72), and the storage box (71) being provided with a throwing assembly (73) for throwing the tracer particles (72) onto the water surface; The throwing assembly (73) comprises a cylinder (731), a motor (733), a rotating block (734), a connecting rod (735), a paddle (736) and a throwing tube (737); the cylinder (731) is fixedly mounted on one side of the material storage box (71), and the cylinder (731) is connected to the material storage box (71) through a discharge hole (732); the motor (733) is fixedly mounted on one side of the cylinder (731), and an output end of the motor (733) extends into the cylinder (731) and is connected to the rotating block (734). Then, connecting rods (735) are evenly arranged on the circumference of the rotating block (734), and a paddle (736) is connected to the free end of the connecting rod (735), and the paddle (736) abuts against the inner wall of the cylinder (731), and a throwing tube (737) is arranged on one side of the cylinder (731) in an inclined downward direction along the tangential direction; when the accumulation height of the tracer particles (72) in the cylinder (731) exceeds the height of the discharge hole (732), the discharge hole (732) will be blocked, so as to achieve the purpose of automatically controlling the discharge amount; The tracer particles (72) are provided with a fluorescent coating on their surfaces; The auxiliary detection mechanism (7) further comprises a lighting module (76), the lighting module (76) comprising a support plate (761), a rotating seat (762) and an ultraviolet lamp (763), the support plate (761) being fixedly mounted on the material storage box (71), the rotating seat (762) being arranged on the support plate (761), and the output end of the rotating seat (762) being connected to the ultraviolet lamp (763).

2. The drone video flow measurement device based on image recognition according to claim 1 is characterized by: The drone body (1) is also provided with a posture correction module (5) and a height detection module (6).

3. The drone video flow measurement device based on image recognition according to claim 1 is characterized by: A feeding port (74) is also provided on one side of the material storage box (71), and a sealing block (75) is threadedly connected to the feeding port (74).

4. The drone video flow measurement device based on image recognition according to claim 1 is characterized by: The tracer particles (72) are made of a degradable material such as polylactic acid, polyadipate, polybutylene terephthalate or polybutylene succinate.

5. The drone video flow measurement device based on image recognition according to claim 1 is characterized by: The drone body (1) is also provided with a positioning module (8) for locating the position of the drone body (1), and the control module (4) is electrically connected to the positioning module (8).

Images