Lamp-induced pest key identification feature image data acquisition platform based on machine vision
By adopting machine vision technology and vibration mechanism in the lamp-induced pest monitoring system, we ensure the uniform distribution and appropriate posture of pests on the transparent shooting plane, the problem of inaccurate collection of pest recognition features in traditional systems is solved, and the accuracy of monitoring data is improved and reliability of early warning information.
Patent Information
- Application Number
- CN202410872513.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-01
- Publication Date
- 2024-12-20
AI Technical Summary
The traditional lamp-induced pest monitoring and early warning system cannot fully and accurately and automatically collect the key identification characteristics of pests on the conveyor belt, resulting in sparse image data acquisition and unsatisfactory recognition accuracy, affecting the accuracy of monitoring data and the reliability of early warning information.
A platform for collecting key feature identification features of lamp-induced pests based on machine vision was designed, using a top-view insect image data acquisition module, a random stacked distributed insect swarm uniform vibration dispersion module, and a tilted elevation angle pest back recognition feature stereoscopic shooting module, through The sliding and vibration mechanism ensures even distribution and proper posture of the pest on the transparent shooting plane, ensuring that industrial cameras can capture key identification features of the pest’s back.
It improves the detection and recognition accuracy of pest image data and the accuracy of sunlight data, enhances the reliability of monitoring and early warning information, and improves the prevention and control effect.
Smart Images

Figure CN119169240A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of automatic identification, monitoring and early warning of pests, and in particular to a machine vision-based image data acquisition platform for key identification features of light-attracted pests. Background Art
[0002] Taking advantage of the biological characteristic of insects' phototaxis, pest light luring technology has been widely used in the field of agricultural plant protection. It is not only used to lure and kill pests to control their population and reduce crop yields caused by pest feeding, but also to obtain the dynamics of pest population growth and decline by counting the number of pests that come into contact with lights every day, providing a key basis for dynamic prediction and early warning of future pest occurrences.
[0003] Traditional monitoring and early warning tasks of light-induced pests are highly dependent on manpower, requiring manual labor to remove the insect collection box at the bottom of the insect-attracting lamp and pour out the insects for classification and counting one by one to obtain daily data on the pests on the lamp. In recent years, the grassroots plant protection field has faced problems such as a small number of personnel and uneven qualifications, so the quality of the monitoring and early warning tasks of light-induced pests is seriously insufficient, and the data reliability and authenticity are poor.
[0004] In recent years, computer vision technology has been introduced into the field of light-induced pest monitoring and early warning. The industry has proposed intelligent insect monitoring and reporting lamps using computer vision technology. The upper part of the intelligent insect monitoring and reporting lamp is a light-induced insect module, and the internal insect heat drying device is used to make insects lose their ability to move under high temperatures. The insects that have been heat-dried will be placed on the conveyor belt directly below. The conveyor belt will move the insects to the bottom of the industrial camera. The industrial camera collects image data of the insects on the upper surface of the conveyor belt from a bird's-eye view, and uploads the image data to the server through the remote communication module. The computer vision model in the server will detect, identify and count the target pests in the image data, thereby obtaining daily data on pests on the lamp, analyzing the dynamics of the number of pests, and then further combining other pest prediction and forecasting factors to analyze and judge the future occurrence of pests, and determine the best prevention and control time node to improve the prevention and control effect. The intelligent insect monitoring and reporting lamps that have been used for field light-induced pest monitoring and early warning have the disadvantage of being unable to fully and accurately automatically collect the key identification features of pests on the conveyor belt. The reason for this problem is that the posture and distribution of pests are random when they fall on the conveyor belt. There are three main postures of pests: back facing up, back facing down, and back facing sideways. The key identification features of most pests are concentrated on the back. The random distribution causes the pests to overlap and block each other. These situations will cause the industrial cameras that collect pest image data from a bird's-eye view angle to only collect sparsely distributed key identification features of pests with their backs facing up. This will increase the difficulty of subsequent computer vision models to identify the pest image data collected by the intelligent insect monitoring lights, resulting in unsatisfactory model recognition accuracy, affecting the accuracy of monitoring data, and being unable to reflect the actual number of field light-trapped pests and their dynamic laws of growth and decline, resulting in low reliability of early warning information and poor prevention and control effects. Summary of the invention
[0005] The technical problem to be solved by the present invention is to provide a machine vision-based image data acquisition platform for key identification features of light-lured pests, so as to improve the detection and identification accuracy of field light-lured pest image data, the accuracy of pest daylight data, and the reliability of pest monitoring and early warning information.
[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: A machine vision-based image data acquisition platform for key identification features of light-trapped pests, the platform consists of a bird's-eye view insect image data acquisition module installed on the platform structure frame, a uniform dispersion module for randomly stacked and distributed insect groups, a three-dimensional mobile shooting module for identifying features of the backs of inclined-elevation pests, and a transparent shooting plane. The bird's-eye view insect image data acquisition module is located at the top of the platform structure frame and can slide objects on the transparent shooting plane within the shooting frame, the uniform dispersion module for randomly stacked and distributed insect groups acts on the transparent shooting plane to generate vibrations, and the three-dimensional mobile shooting module for identifying features of the backs of inclined-elevation pests is located below the transparent shooting plane and can adjust the posture of its shooting module according to objects on the transparent shooting plane.
[0007] The above-mentioned overhead view insect image data acquisition module includes a overhead view industrial camera, which is installed on a overhead view industrial camera fixed bracket, and the overhead view industrial camera fixed bracket is installed on a overhead view industrial camera mobile shooting one-way slide module and can be controlled to slide along both sides. The overhead view industrial camera mobile shooting one-way slide module is fixedly connected to the platform structure frame.
[0008] The above-mentioned transparent shooting plane includes a transparent vibration plate, which is embedded in the voice coil motor box, and the voice coil motor box is installed on the box base bracket. The inclined elevation pest back recognition feature three-dimensional moving shooting module is located in the box base bracket and below the transparent vibration plate.
[0009] The above-mentioned randomly stacked and distributed insect swarm uniform vibration dispersion module includes four voice coil motors on a voice coil motor box, and the four voice coil motors are in contact with the four corners of the transparent vibration plate.
[0010] The above-mentioned tilted elevation angle pest back identification feature stereoscopic mobile shooting module includes a tilted elevation angle industrial camera, which is installed on a vertical rotation axis. The vertical rotation axis drives the tilted elevation angle industrial camera to perform elevation angle transformation for vertical shooting, and is used to capture the back identification features of target pests with their backs facing down and to the side, which cannot be captured by the upper downward-angle industrial camera.
[0011] The above-mentioned vertical rotating shaft is installed on the horizontal rotating shaft, and the horizontal rotating shaft drives the vertical rotating shaft and the tilt-elevation industrial camera thereon to rotate horizontally, so as to adjust the horizontal orientation of the tilt-elevation industrial camera according to the different orientations of the target pest body, so that the tilt-elevation industrial camera can collect image data facing the back of the target pest.
[0012] The above-mentioned vertical rotating shaft is fixedly connected to the tilt-elevation industrial camera fixed bracket, the tilt-elevation industrial camera fixed bracket is installed on the horizontal rotating shaft, the horizontal rotating shaft is installed on the tilt-elevation industrial camera plane moving slide module and can be controlled to slide along both sides, the tilt-elevation industrial camera plane moving slide module drives the horizontal rotating shaft, the vertical rotating shaft and the tilt-elevation industrial camera thereon to perform two-dimensional plane movement.
[0013] The above-mentioned top-view industrial camera fixing bracket is provided with a fill light, and the fill light is controlled by a fill light controller.
[0014] A platform chassis is provided at the bottom of the platform structure frame, and the lower end of the box base bracket, the tilt-elevation industrial camera plane moving slide module connected to the tilt-elevation industrial camera, and the fill light controller are installed on the platform chassis.
[0015] The present invention provides a machine vision-based light-lured pest key identification feature image data acquisition platform. Through the coordination of a bird's-eye view insect image data acquisition module, a randomly stacked and distributed insect group uniform dispersion module, and an inclined elevation angle pest back key identification feature three-dimensional mobile shooting module, the device can accurately acquire the target pest back key identification features that are invisible from a bird's-eye view. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 The structure of the image data acquisition platform for key identification features of light-trapped insect pests provided by the present invention Figure 1 ; Figure 2 The structure of the image data acquisition platform for key identification features of light-trapped pests Figure 2 ; Figure 3 Schematic diagram of the structure of the randomly stacked and evenly distributed insect swarm dispersion module Figure 1 ; Figure 4 Schematic diagram of the structure of the randomly stacked and evenly distributed insect swarm dispersion module Figure 2 ; Figure 5 It is a structural schematic diagram of a three-dimensional mobile shooting module for identifying features of the back of pests at an inclined elevation angle; Figure 6 This is a partially enlarged schematic diagram of the stereoscopic mobile shooting module for identifying features on the back of pests at an inclined angle.
[0017] In the figure: 1. Top-down industrial camera; 2. Fill light; 3. Transparent vibration plate; 4. Voice coil motor box; 5. Box base bracket; 6. Electric control box; 7. Fill light controller; 8. Switch; 9. Plane moving slide module for tilt elevation industrial camera; 10. Fixed bracket for top-down industrial camera; 11. Tilt elevation industrial camera; 12. Fixed bracket for tilt elevation industrial camera; 13. One-way slide module for mobile shooting of top-down industrial camera; 14. Platform structure frame; 15. Platform chassis; 16. Horizontal rotation axis; 17. Vertical rotation axis; 18. Fixed L-shaped plate for tilt elevation industrial camera; 19. Voice coil motor; 20. Side panel of voice coil motor box. DETAILED DESCRIPTION
[0018] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and embodiments.
[0019] A machine vision-based image data acquisition platform for key identification features of light-trapped pests, the platform consists of a bird's-eye view insect image data acquisition module installed on a platform structure frame 14, a uniform dispersion module for randomly stacked and distributed insect groups, a three-dimensional moving shooting module for identifying features of the back of insects at an inclined elevation, and a transparent shooting plane. The bird's-eye view insect image data acquisition module is located at the top of the platform structure frame 14 and can slide objects on the transparent shooting plane within the shooting frame. The uniform dispersion module for randomly stacked and distributed insect groups acts on the transparent shooting plane to generate vibrations. The three-dimensional moving shooting module for identifying features of the back of insects at an inclined elevation is located below the transparent shooting plane and can adjust the posture of its shooting module according to objects on the transparent shooting plane.
[0020] The above-mentioned overhead view insect image data acquisition module includes a overhead view industrial camera 1, which is installed on a overhead view industrial camera fixed bracket 10. The overhead view industrial camera fixed bracket 10 is installed on a overhead view industrial camera mobile shooting one-way slide module 13 and can be controlled to slide along both sides. The overhead view industrial camera mobile shooting one-way slide module 13 is fixedly connected to the platform structure frame 14.
[0021] The above-mentioned transparent shooting plane includes a transparent vibration plate 3, which is embedded in a voice coil motor box 4, and the voice coil motor box 4 is installed on a box base bracket 5. The inclined elevation pest back identification feature three-dimensional moving shooting module is located in the box base bracket 5 and below the transparent vibration plate 3.
[0022] The above-mentioned randomly stacked and distributed insect swarm uniform vibration dispersion module includes four voice coil motors 19 on the voice coil motor box 4, and the four voice coil motors 19 are in contact with the four corners of the transparent vibration plate 3.
[0023] The above-mentioned tilted elevation angle pest back identification feature stereoscopic mobile shooting module includes a tilted elevation angle industrial camera 11, which is installed on a tilted elevation angle industrial camera fixed L-shaped plate 18, and the tilted elevation angle industrial camera fixed L-shaped plate 18 is fixedly connected to a vertical direction rotating shaft 17. The vertical direction rotating shaft 17 drives the tilted elevation angle industrial camera 11 to perform elevation angle transformation for vertical shooting, which is used to capture the back identification features of target pests with their backs facing down and their backs facing sideways that cannot be captured by the upper bird's-eye view industrial camera 1.
[0024] The above-mentioned vertical rotation shaft 17 is installed on the tilt-elevation industrial camera fixing bracket 12, and the tilt-elevation industrial camera fixing bracket 12 is fixedly connected to the horizontal rotation shaft 16. The horizontal rotation shaft 16 drives the tilt-elevation industrial camera fixing bracket 12 and the vertical rotation shaft 17 thereon and the tilt-elevation industrial camera 11 to rotate in the horizontal direction, which is used to adjust the horizontal orientation of the tilt-elevation industrial camera 11 according to the different orientations of the target pest body, so that the tilt-elevation industrial camera 11 can collect image data facing the back of the target pest.
[0025] The above-mentioned tilt-elevation industrial camera fixed L-shaped plate 18 is fixedly connected to the vertical rotating shaft 17, the vertical rotating shaft 17 is fixedly connected to the tilt-elevation industrial camera fixed bracket 12, the tilt-elevation industrial camera fixed bracket 12 is installed on the horizontal rotating shaft 16, the horizontal rotating shaft 16 is installed on the tilt-elevation industrial camera plane moving slide module 9 and can be controlled to slide in both directions, the tilt-elevation industrial camera plane moving slide module 9 drives the horizontal rotating shaft 16, the vertical rotating shaft 17 and the tilt-elevation industrial camera 11 thereon to perform two-dimensional plane movement.
[0026] The above-mentioned top-view industrial camera fixing bracket 10 is provided with a fill light 2 , and the fill light 2 is controlled by a fill light controller 7 .
[0027] A platform chassis 15 is provided at the bottom of the platform structure frame 14 , and the lower end of the box base bracket 5 , the tilt-elevation industrial camera plane moving slide module 9 connected to the tilt-elevation industrial camera 11 , and the fill light controller 7 are installed on the platform chassis 15 .
[0028] Example: like Figure 1As shown, the top-down industrial camera 1 is installed on a top-down industrial camera fixed bracket 10, and the bracket 10 is installed on a top-down industrial camera mobile shooting one-way slide module 13; the industrial camera mobile shooting one-way slide module 13 is fixed on the platform structure frame 14, and the fill light 2 is installed on the top-down industrial camera fixed bracket 10; the fill light controller 7 is installed on the platform base 15, and the switch 8 is installed on the platform base 15; the transparent vibration plate 3 is embedded in the voice coil motor box 4, and the four corners of the transparent vibration plate 3 are fixed to the four voice coil motors in the box 4, the voice coil motor is fixed at the four corners inside the voice coil motor box, and the voice coil motor box 4 is installed on the transparent flexible vibration plate base bracket 5; the transparent flexible vibration plate base bracket 5 is installed on the platform chassis 15; the inclined elevation industrial camera plane moving slide module 9 is installed on the platform base 15.
[0029] like Figure 2 As shown, the tilted elevation industrial camera 11 is installed on the vertical rotation shaft 17, and the vertical reverse rotation shaft drives the tilted elevation industrial camera 11 to change the vertical shooting elevation angle, which is used to capture the key identification features of the back of the target pests with their backs facing down and facing sideways that cannot be captured by the upper downward-angle industrial camera 1. The vertical rotation shaft 17 is installed on the horizontal rotation shaft 16, and the horizontal rotation shaft 16 drives the vertical rotation shaft 17 and the tilted elevation industrial camera 11 thereon to rotate in the horizontal direction, so as to adjust the horizontal orientation of the tilted elevation industrial camera 11 according to the different orientations of the target pest body, so that the tilted elevation industrial camera 11 can collect image data facing the back of the target pest, and the target pest orientation data is given by the target pest key point prediction model trained by the target pest image data set; the tilted elevation industrial camera 11 is installed on the tilted elevation industrial camera fixing bracket 12; the tilted elevation industrial camera fixing bracket 12 is installed on the tilted elevation industrial camera plane moving slide module 13, and the tilted elevation industrial camera plane moving slide module 13 drives the horizontal rotation shaft 16, the vertical rotation shaft 17 and the tilted elevation industrial camera 11 thereon to perform two-dimensional plane movement, so that the tilted elevation industrial camera 11 can move to the most suitable shooting position facing the back of the target pest; the industrial camera plane moving slide module 13 is installed on the platform base 15.
[0030] The working principle and use process of the present invention: After high-temperature treatment, the insect swarm that loses its ability to move falls onto the transparent vibration plate 3, and the four voice coil motors inside the voice coil motor box 4 are started, creating micro-vibrations of a specific waveform on the transparent vibration plate 3 in contact. This vibration makes the insect swarms that are stacked and crowded on the transparent vibration plate 3 evenly and spacedly distributed on the transparent vibration plate 3. The fill light controller turns on the fill light 2, and the top-down industrial camera 1 shoots the insects on the transparent vibration plate 3. The trained target pest key point prediction model based on YOLOv8-pose is used for detection and identification. The position and orientation data of the target pests with their backs facing down or sideways on the transparent vibration plate 3 given by the model are converted into the horizontal two-dimensional plane position, horizontal two-dimensional plane angle and vertical direction angle required for the tilt-elevation industrial camera 11 to shoot the target pests with their backs facing down or sideways. According to the calculated horizontal two-dimensional plane position, the vertical direction axis 17 on the plane moving slide module of the tilt-elevation industrial camera is guided to move to the set real space position, and according to the calculated horizontal two-dimensional plane angle, the horizontal direction axis 16 is guided to rotate to the specified angle, and according to the calculated vertical direction angle, the vertical direction axis 17 is guided to rotate to the specified angle. At this time, the tilt-elevation industrial camera 11 will reach the position and angle facing the key identification features on the back of the target pest to complete the shooting.
Claims
1. A machine vision-based image data acquisition platform for key identification features of light-attracted pests, characterized by: The platform comprises a bird's-eye view insect image data acquisition module, a randomly stacked and distributed insect swarm uniform vibration dispersion module, an inclined elevation angle pest back identification feature stereoscopic moving shooting module, and a transparent shooting plane, the bird's-eye view insect image data acquisition module is located at the top of the platform structure frame (14) and can slide to shoot objects on the transparent shooting plane in the frame, the randomly stacked and distributed insect swarm uniform vibration dispersion module acts on the transparent shooting plane to generate vibration, and the inclined elevation angle pest back identification feature stereoscopic moving shooting module is located below the transparent shooting plane and can adjust the posture of its shooting module according to the objects on the transparent shooting plane.
2. According to the machine vision-based light-lured insect pest key identification feature image data acquisition platform described in claim 1, it is characterized in that: The overhead view insect image data acquisition module comprises an overhead view industrial camera (1), wherein the overhead view industrial camera (1) is mounted on an overhead view industrial camera fixing bracket (10), wherein the overhead view industrial camera fixing bracket (10) is mounted on an overhead view industrial camera mobile shooting one-way slide module (13) and can be controlled to slide in two directions, and the overhead view industrial camera mobile shooting one-way slide module (13) is fixedly connected to a platform structure frame (14).
3. According to the machine vision-based light-lured insect pest key identification feature image data acquisition platform described in claim 2, it is characterized in that: The transparent shooting plane comprises a transparent vibration plate (3), the transparent vibration plate (3) is embedded in a voice coil motor box (4), the voice coil motor box (4) is mounted on a box base bracket (5), and the inclined elevation angle pest back identification feature three-dimensional moving shooting module is located inside the box base bracket (5) and below the transparent vibration plate (3).
4. The machine vision-based light-lured insect pest key identification feature image data acquisition platform according to claim 3, characterized in that: The randomly stacked and distributed insect swarm uniform vibration dispersion module comprises four voice coil motors (19) on a voice coil motor housing (4), and the four voice coil motors (19) are in contact with four corners of a transparent vibration plate (3).
5. The machine vision-based light-attracted insect pest key identification feature image data acquisition platform according to claim 4, characterized in that: The tilted elevation angle pest back identification feature stereoscopic mobile shooting module comprises a tilted elevation angle industrial camera (11), which is mounted on a vertical rotation shaft (17). The vertical rotation shaft (17) drives the tilted elevation angle industrial camera (11) to perform elevation angle transformation for vertical shooting, so as to capture the back identification features of target pests with their backs facing downward or sideways, which cannot be captured by the upper downward-angle industrial camera (1).
6. The machine vision-based light-attracted insect pest key identification feature image data acquisition platform according to claim 5, characterized in that: The vertical rotation shaft (17) is mounted on the horizontal rotation shaft (16), and the horizontal rotation shaft (16) drives the vertical rotation shaft (17) and the tilted elevation industrial camera (11) thereon to rotate in the horizontal direction, so as to adjust the horizontal orientation of the tilted elevation industrial camera (11) according to different orientations of the target pest body, so that the tilted elevation industrial camera (11) can collect image data facing the back of the target pest.
7. The machine vision-based light-attracted insect pest key identification feature image data acquisition platform according to claim 6, characterized in that: The vertical rotation axis (17) is fixedly connected to the tilt-elevation industrial camera fixing bracket (12); the tilt-elevation industrial camera fixing bracket (12) is mounted on the horizontal rotation axis (16); the horizontal rotation axis (16) is mounted on the tilt-elevation industrial camera plane moving slide module (9) and can be controlled to slide in both directions; the tilt-elevation industrial camera plane moving slide module (9) drives the horizontal rotation axis (16), the vertical rotation axis (17) and the tilt-elevation industrial camera (11) thereon to perform two-dimensional plane movement.
8. The machine vision-based light-attracted insect pest key identification feature image data acquisition platform according to claim 7, characterized in that: The top-view industrial camera fixing bracket (10) is provided with a fill light (2), and the fill light (2) is controlled by a fill light controller (7).
9. The machine vision-based light-attracted insect pest key identification feature image data acquisition platform according to claim 8, characterized in that: A platform chassis (15) is provided at the bottom of the platform structure frame (14), and the lower end of the box base bracket (5), the tilt-elevation industrial camera plane moving slide module (9) connected to the tilt-elevation industrial camera (11), and the fill light controller (7) are mounted on the platform chassis (15).
Citation Information
Patent Citations
Computer vision-based large field pest situation monitoring sampling device and sampling method
CN104918007A
Image acquisition control method and device, storage medium and system
CN109657576A
Vehicle-mounted movable field disease and pest situation automatic investigation system
CN112572313A
Visual inspection system of bogie
CN112857841A
Intelligent insect situation lamp forecasting device
CN215380914U
