A laser biological defense method and system based on wide-area visual guidance

Abstract

The application discloses a biological prevention and treatment method and system based on wide-area laser scanning, and belongs to the fields of biological prevention and treatment and photoelectric technology. The system comprises: an image acquisition module, which is used for acquiring image information of a monitoring area and identifying the spatial coordinates of target organisms; a laser generation module, which is used for generating a laser beam; an optical deflection positioning module, which is arranged on a laser beam transmission path and is used for adjusting the projection angle of the laser beam according to the spatial coordinates of the target organisms; a second optical module, which is arranged on the light-emitting side of the laser beam and is connected with a control module, and is used for converting the laser beam into a regional spot with a specific diffusion angle; and the control module, which is used for controlling the optical deflection positioning module to cover the target organism area with the regional spot, and physically hits the target organism in a non-single-point accurate alignment manner. The application solves the problems of small coverage and insufficient dynamic tracking precision of the existing laser prevention and treatment technology, and improves the efficiency and accuracy of biological prevention and treatment in a large-area scene.

Description

Technical Field

[0001] This invention relates to the fields of biological control and optoelectronic technology, and in particular to a laser biological control method and system based on wide-area vision guidance. Background Technology

[0002] With the acceleration of urbanization and the development of large-scale agricultural production, the threat posed by pests such as mosquitoes and flies to public health and crop safety is becoming increasingly serious. Data from the World Health Organization shows that mosquito-borne diseases such as malaria and dengue fever cause more than 700,000 deaths globally each year. Traditional chemical control methods suffer from problems such as environmental pollution, pesticide residues, and pest resistance, while physical control methods, such as ultraviolet lamps and sticky traps, have limitations such as limited coverage, lack of specificity, and accidental harm to non-target organisms.

[0003] In recent years, laser-based biological control technology has gradually attracted attention. Lasers possess advantages such as good directionality, high energy density, and fast response speed, demonstrating unique advantages in precisely targeting harmful organisms. However, existing laser insect repellent or extermination technologies mainly suffer from the following shortcomings:

[0004] Limited Coverage: Existing technologies mostly employ a single-point precision aiming method, focusing the laser beam on an extremely small area (typically <1mm in diameter), requiring each target organism to be irradiated individually. In large open environments (such as farmland, pastures, storage areas, and residential gardens), the effectiveness of a single device is extremely low, failing to meet the needs of large-area coverage. Furthermore, the single-point aiming method demands extremely high target positioning accuracy; even a slight positioning deviation can lead to misses, further reducing the actual effectiveness of the defense.

[0005] Poor all-weather adaptability: Most existing vision systems rely on single visible light imaging. In nighttime or low-light environments (such as cloudy days, dusk, or scenes with tree shade), visible light cameras cannot acquire clear target images, causing target recognition to fail and the system to be unable to work continuously in all weather conditions. Furthermore, harmful organisms such as mosquitoes are more active at night, and systems lacking nighttime operation capabilities are significantly less effective in practical applications.

[0006] Insufficient accuracy in dynamic target tracking: Flying pests such as mosquitoes fly at high speeds (typically 1-2 m / s) and their flight trajectories are irregular (frequent changes of direction, hovering, and rapid acceleration). Existing technologies have limited ability to predict the trajectory of flying targets, and there is a time delay between the laser aiming point and the actual target position, leading to deviations. Furthermore, in outdoor environments, wind disturbances can cause the actual flight trajectory of the target to deviate from the predicted trajectory. Existing technologies do not consider wind disturbance compensation, further increasing the miss rate.

[0007] Insufficient safety guarantees: Laser beams have high energy density, posing a potential hazard to human skin and eyes. Current technology lacks effective safety mechanisms for humans and pets, creating serious safety risks when used in areas with high human activity. If a person accidentally enters the laser irradiation area, the system may not be able to detect and stop laser emission in time, potentially causing personal injury. Summary of the Invention

[0008] The purpose of this invention is to propose a laser biological control system based on wide-area vision guidance, which aims to solve the technical problems of existing laser biological control technologies, such as small coverage, poor all-weather adaptability, insufficient dynamic target tracking accuracy, and insufficient safety assurance.

[0009] The present invention is implemented as follows: a laser-based biological control system based on wide-area vision guidance, the system comprising: The image acquisition module is used to acquire image information of the monitoring area and identify the spatial coordinates of the target organism in the image information; A laser generating module is used to generate at least one laser beam under the control of the control module; An optical deflection positioning module, connected to the control module, is positioned on the transmission path of the laser beam and is used to adjust the projection angle of the laser beam according to the instructions of the control module. The second optical module is located on the light-emitting side of the optical deflection and positioning module and is used to convert the laser beam into a regional light spot with a specific diffusion angle. The control module is connected to the image acquisition module, the laser generation module, and the optical deflection positioning module, respectively. It is used to control the optical deflection positioning module to adjust the projection position of the regional light spot according to the spatial coordinates output by the image acquisition module, so as to cover the target biological area.

[0010] Furthermore, the system also includes a dichroic mirror, which is disposed on the transmission path of the laser beam to reflect the laser beam to the optical deflection and positioning module, and simultaneously transmits the ambient light reflected by the target organism to the image capturing module, so that the laser emission optical axis is coaxial with the receiving optical axis of the image capturing module.

[0011] Furthermore, the system also includes The first optical module, located at the laser emission end and connected to the control module, is used to adjust the laser spot size and energy.

[0012] Another objective of this invention is to propose a laser-based biological control method based on wide-area vision guidance, the method comprising the following steps: Acquire image information of the monitoring area and identify the spatial coordinates of target organisms in the image information; Generate at least one laser beam; The projection angle of the laser beam is adjusted according to the spatial coordinates of the target organism; The laser beam is converted into a regional spot with a specific diffusion angle; The projection position of the regional light spot is controlled and adjusted to cover the target biological area.

[0013] The method also includes A dichroic mirror is placed along the transmission path of the laser beam to make the laser emission axis and the receiving axis coaxial.

[0014] Furthermore, the method also includes A dichroic mirror is placed along the transmission path of the laser beam to make the laser emission axis and the receiving axis coaxial.

[0015] Furthermore, the method also includes A first optical module is set at the laser output end to adjust the spot size and laser energy of the laser beam.

[0016] Beneficial effects of the present invention This invention discloses a biological control method and system based on wide-area laser scanning. The system includes: an image acquisition module for acquiring image information of the monitoring area and identifying the spatial coordinates of the target organism; a laser generation module for generating a laser beam; an optical deflection and positioning module disposed on the laser beam transmission path for adjusting the laser beam projection angle according to the spatial coordinates of the target organism; a second optical module disposed on the laser beam emission side for converting the laser beam into a regional spot with a specific diffusion angle; and a control module for controlling the optical deflection and positioning module to ensure that the regional spot covers the target organism area, enabling physical strikes through non-single-point precise aiming. This invention solves the problems of small coverage and insufficient dynamic tracking accuracy in existing laser control technologies, improving the efficiency and accuracy of biological control in large-area scenarios. Attached Figure Description

[0017] Figure 1 This is a structural diagram of a laser biological control system based on wide-area vision guidance according to a first preferred embodiment of the present invention; Figure 2 This is a structural diagram of a laser biological control system based on wide-area vision guidance according to a second preferred embodiment of the present invention; Figure 3 This is a structural diagram of a laser biological control system based on wide-area vision guidance according to a third preferred embodiment of the present invention; Figure 4 This is a flowchart of a laser biocontrol method based on wide-area vision guidance, according to a preferred embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. For ease of explanation, only the parts related to the embodiments of this invention are shown. It should be understood that the specific embodiments described herein are merely for explaining this invention and are not intended to limit this invention.

[0019] This invention proposes a biological control method and system based on wide-area laser scanning. The system embodiment includes: an image acquisition module for acquiring image information of the monitoring area and identifying the spatial coordinates of the target organism; a laser generation module for generating a laser beam; an optical deflection and positioning module, positioned along the laser beam transmission path, for adjusting the laser beam projection angle according to the spatial coordinates of the target organism; a second optical module, positioned on the laser beam emission side, for converting the laser beam into a regional light spot with a specific diffusion angle; and a control module for controlling the optical deflection and positioning module to ensure the regional light spot covers the target organism area, enabling physical strikes through non-single-point precise aiming. This invention solves the problems of small coverage and insufficient dynamic tracking accuracy in existing laser control technologies, improving the efficiency and accuracy of biological control in large-area scenarios.

[0020] Figure 1 This is a structural diagram of a laser biological control system based on wide-area vision guidance according to a preferred embodiment of the present invention; the system includes an image acquisition module, a laser generation module, an optical deflection and positioning module, a second optical module, and a control module; The image acquisition module is used to acquire image information of the monitoring area and identify the spatial coordinates of the target organism in the image information; A laser generating module is used to generate at least one laser beam under the control of the control module; An optical deflection positioning module, connected to the control module, is positioned on the transmission path of the laser beam and is used to adjust the projection angle of the laser beam according to the instructions of the control module. The second optical module is disposed on the light-emitting side of the laser beam and is used to convert the laser beam into a regional light spot with a specific diffusion angle; The control module is connected to the image acquisition module, the laser generation module, and the optical deflection positioning module, respectively. It is used to control the optical deflection positioning module to adjust the projection position of the regional light spot according to the spatial coordinates output by the image acquisition module, so as to cover the target biological area.

[0021] Figure 2This is a structural diagram of a laser biological control system based on wide-area visual guidance according to a second preferred embodiment of the present invention; further, the system also includes a dichroic mirror, which is disposed on the transmission path of the laser beam to reflect the laser beam to the optical deflection and positioning module, and simultaneously transmits the ambient light reflected by the target organism to the image capturing module, so that the laser emission optical axis and the receiving optical axis of the image capturing module are coaxial.

[0022] In some optional embodiments, the laser generation module and the image capturing module can be designed with a coaxial optical path using a dichroic mirror, and the specific optical path architecture is as follows: The laser beam (e.g., 532nm green light) generated by the laser generation module is directed towards a dichroic mirror. The dichroic mirror is positioned at a preset tilt angle (e.g., 45°) along the laser beam's transmission path, and its surface has high reflectivity (e.g., reflectivity ≥95%) for the laser wavelength, reflecting the laser beam onto the reflective surface of the optical deflection and positioning module. The optical deflection and positioning module rotates the reflective surface according to instructions from the control module, reflecting the laser beam towards the target organism in space, thus projecting the laser beam onto the target organism.

[0023] Ambient light (containing visible light components) reflected from the surface of the target organism is directed towards the reflective mirror of the optical deflection and positioning module in the opposite direction to the laser emission path. After being reflected by the galvanometer, it is directed towards the dichroic mirror. The dichroic mirror has high transmittance (e.g., transmittance ≥ 90%) in the visible light band, transmitting the ambient light reflected from the target organism to the image acquisition unit (e.g., camera) sensor of the image acquisition module for imaging, which is then used by the deep learning model unit for target recognition and coordinate calculation.

[0024] The laser emission direction and the image acquisition module's receiving direction share the same galvanometer reflector, and after being combined by the dichroic mirror, their optical axes are completely aligned. That is, the center of the image captured by the image acquisition module always corresponds to the laser's pointing point, eliminating the need for software parallax compensation. This coaxial design ensures consistency between the laser projection point and the visual monitoring center, avoiding aiming deviations caused by optical axis offset and improving the system's projection accuracy onto the target organism.

[0025] Figure 3 This is a structural diagram of a laser-based biological control system based on wide-area vision guidance according to a third preferred embodiment of the present invention; further, the system also includes... The first optical module, located at the laser emission end and connected to the control module, is used to adjust the laser spot size and energy. Specifically, it is positioned between the laser generation module and the dichroic mirror. In some optional implementations, the first optical module can employ an adjustable collimating lens group (with a stepper motor driving the lens displacement) + an electric laser attenuation component. Its control logic is as follows: the image acquisition module calculates the target's three-dimensional distance, and the control module outputs instructions based on the distance range to drive the motor to adjust the collimating lens spacing to match the corresponding spot reference size. Simultaneously, the attenuation component is adjusted to match the energy density required for the distance, ensuring that the kill threshold (e.g., 0.5 J / cm²) is maintained within the spot. The collimating lens group consists of at least two spherical lenses and can be designed as a 4f system to compress the laser beam divergence angle to <0.5 mrad, ensuring that the beam diameter expansion at the maximum effective distance (e.g., 50 m) does not exceed the preset maximum beam diameter (e.g., 25 mm).

[0026] In some alternative embodiments, the first optical module may also be disposed between the laser generating module and the optical deflection positioning module (in which case there is no dichroic mirror), as shown in the attached figures; In some alternative embodiments, the laser generating module employs a semiconductor laser diode, which may have the following technical parameters: The working wavelength is set to a visible light band that has a sensitive stimulating effect on the insect visual system, such as 532nm (green light), 450nm (blue light), or 940nm (infrared light). In an optional embodiment of the present invention, the working wavelength is set to 532nm (green light). This wavelength has a strong stimulating effect on the insect visual system. The compound eyes of insects are most sensitive to the green light band. The 532nm laser can effectively interfere with the flight control system of mosquitoes, causing them to lose flight attitude stability, disrupt wing vibration frequency, and ultimately lead to wing damage or loss of flight ability. The output power is set to a level sufficient to destroy the wings of the target organism and the power density is within a safe range after the light spot is expanded. For example, the output power is 5W. This power level ensures that it is sufficient to destroy the wings of mosquitoes (requiring an energy density of about 0.5J / cm²). After the light spot is expanded to 5mm-30mm by the second optical module, the power density in the light spot area is controlled within a safe range to avoid thermal damage to the surrounding environment. Beam quality: Beam quality factor M² < 1.5, divergence angle < 1.5 mrad, ensuring that the laser beam maintains good directionality during transmission; Driving method: Constant current source driving is adopted, and the stability of the driving current is ±0.1%, ensuring that the output power fluctuation is <±2%. The modulation frequency can reach a preset high frequency level, such as 20kHz, and supports pulse working mode. The laser emission timing can be adjusted as needed.

[0027] In some alternative embodiments, the optical deflection positioning module employs a biaxial galvanometer scanner, and may use the following technical parameters: The scanning angle range is set to a preset angle range (e.g., ±20°, optical angle) to ensure that the scanning range covers the preset monitoring area (e.g., ±17.6m) at the maximum effective distance (e.g., 50m). The scanning speed significantly exceeds the maximum flight speed of the target organism; for example, a scanning speed of 5000° / s corresponds to a linear velocity of 436m / s at the maximum effective distance of 50m, which is far greater than the flight speed of the target organism (such as a mosquito) (about 1-2m / s), ensuring that the laser beam can quickly track and cover the target organism. The positioning accuracy meets the preset accuracy requirements for precise projection of regional light spots; for example, a positioning accuracy of 0.005° corresponds to a positioning accuracy of 4.4mm at the maximum effective distance of 50m (better than the preset minimum light spot size), which meets the precise projection requirements of regional light spots (5mm-30mm). The small signal response time is within the preset millisecond-level response time threshold to ensure real-time response to fast-moving targets; for example, the response time for 1% of the full stroke is 0.2ms to ensure real-time response to fast-moving targets.

[0028] In this embodiment of the invention, the optical deflection positioning module calculates the deflection angles of the X-axis and Y-axis galvanometers based on the three-dimensional spatial coordinates of the target organism output by the image acquisition module using a coordinate transformation algorithm. The coordinate transformation process is as follows: First, the world coordinates (X, Y, Z) of the target organism are converted into spherical coordinates (θ, φ, r) relative to the laser emission point, where θ is the horizontal deflection angle, φ is the vertical deflection angle, and r is the distance. Then, the spherical coordinates (θ, φ) are mapped to the driving voltages of the X-axis and Y-axis galvanometers, driving the galvanometers to deflect to the corresponding angles. The calculation delay of the coordinate transformation is <0.1ms, ensuring the overall system response speed.

[0029] In some alternative embodiments, the second optical module is an aspherical diffuser lens or a holographic diffuser. The aspherical diffuser lens is manufactured using a precision molding process, and its surface shape is optimized by optical design software, enabling it to precisely diffuse an incident parallel laser beam into a spot of a specific shape. The holographic diffuser achieves beam diffusion by recording a specific diffraction pattern on an optical substrate. The diffraction pattern is optimized using an Iterative Fourier Transform (IFTA) algorithm, allowing for the creation of spots of arbitrary shape and energy distribution.

[0030] The second optical module expands the originally single-point laser beam into a spot with a predetermined geometry, which can be circular, elliptical, or rectangular. A circular spot is suitable for isotropic target coverage scenarios; an elliptical spot is suitable for scenarios where coverage is stretched along the target's flight direction; and a rectangular spot is suitable for area scanning scenarios. The spot shape can be switched by changing different second optical modules or adjusting the spacing of the diffusion lenses.

[0031] In some alternative embodiments, the cross-sectional diameter of the regional light spot is 5 mm to 30 mm. The choice of light spot diameter depends on the distance between the target organism and the system: a smaller light spot (5 mm to 10 mm) is used for close range (5 m to 15 m) to obtain higher energy density; a medium light spot (10 mm to 20 mm) is used for medium range (15 m to 30 m); and a larger light spot (20 mm to 30 mm) is used for long range (30 m to 50 m) to ensure sufficient coverage area. The energy distribution of the light spot exhibits a flat-top profile or a Gaussian distribution. A flat-top profile ensures that the energy density at each point within the light spot is basically uniform, which is beneficial for uniform strikes; a Gaussian distribution results in the highest energy density at the center of the light spot, which is suitable for scenarios requiring concentrated energy. Within the coverage area of ​​the light spot, the energy density must meet a threshold (approximately 0.5 J / cm²) sufficient to destroy the target organism's wings.

[0032] Furthermore, the image acquisition module also includes a multimodal sensing unit and a light source detection logic unit. A multimodal sensing unit is used to acquire image information of the monitoring area in visible light mode or infrared mode, and obtain image data of the target organism. A light source detection logic unit, connected to the multimodal sensing unit, is used to detect ambient illuminance in real time. When the ambient illuminance is lower than a preset threshold, the multimodal sensing unit is triggered to automatically switch its working mode. The working modes include visible light mode and infrared mode.

[0033] In an optional embodiment, the multimodal sensing unit defaults to visible light mode; the multimodal sensing unit includes a visible light camera module and an infrared camera module. For example, the visible light camera module has a resolution of 2-8 MHz, a frame rate of 30 fps, and a field of view of 90°×60°, used to acquire high-resolution texture images of the target area under good lighting conditions. The infrared camera module operates in the 850nm-940nm wavelength range, has a frame rate of 25 fps, and a temperature resolution of 0.05℃, used to acquire thermal radiation images of the target organism in low-light environments.

[0034] In some specific implementations, the hardware composition of the multimodal sensing unit can be a lens + image sensor CMOS; the lens part can be a (single) low-light night vision lens (low-light environment, infrared + visible light), or a binocular lens (dual, visible light / infrared). In an optional embodiment, the light source detection logic unit monitors the ambient illuminance in real time using an ambient light sensor. For example, the ambient light sensor has an accuracy of ±5%, a range of 0-200,000 lux, and a sampling frequency of 10 Hz. When the ambient illuminance is below a preset threshold (e.g., 50 lux), the light source detection logic automatically triggers a mode switch, switching the multimodal sensing unit from visible light mode to infrared mode. Simultaneously, an active infrared supplemental light is activated. For example, this supplemental light has a working wavelength of 850 nm, a power of 20 W, and an illumination angle of 60°, providing uniform infrared supplemental lighting within a range of 5 m-30 m. Through this mechanism, the system can effectively acquire the spatial coordinates of the target organism under various lighting conditions, including daytime, nighttime, cloudy days, and tree shade.

[0035] Furthermore, the image capturing module also includes: The deep learning model unit, connected to the multimodal sensing unit, is used to perform target recognition on the acquired image information through the deep learning model, distinguish target organisms (such as mosquitoes) from non-target dynamic objects, and calculate the three-dimensional centroid coordinates of the target organism in the monitoring area as a projection reference point.

[0036] In this embodiment of the invention, the deep learning model is trained to distinguish between target organisms (such as mosquitoes) and non-target dynamic objects (such as swaying leaves, flying birds, falling objects, etc.), and calculates the three-dimensional centroid coordinates of the target organism in the monitoring area as a projection reference point.

[0037] The deep learning model may employ an improved YOLO architecture, specifically including: Model structure: The backbone network (Backneck) can use CSPDarknet53, the neck network (Neck) can use a multi-scale feature fusion network (such as PANet+FPN) to perform multi-scale feature fusion, and the detection head (Head) contains multiple (such as 3) detection layers of different scales, corresponding to small targets (mosquitoes, body length 2mm-6mm), medium targets and large targets respectively; Training dataset: A labeled image dataset of a predetermined size (e.g., 15,000 images), where target biological images constitute a predetermined proportion, covering different poses, lighting conditions, and background conditions; non-target dynamic object images constitute a predetermined proportion, including various easily confused interference object types (e.g., leaves, birds, falling objects, etc.). Data augmentation strategies include random rotation (predetermined angle range, e.g., ±15°), brightness jitter (predetermined brightness variation range), Gaussian noise addition, random cropping, etc.

[0038] Training parameters: An adaptive learning rate optimizer (e.g., Adam optimizer) is used, with an initial learning rate of a preset value (e.g., 0.001), a batch size of a preset value (e.g., 16), and a preset number of training epochs (e.g., 300 epochs). The loss function consists of three parts: bounding box regression loss, confidence loss, and classification loss, with the weights of the three parts set to a preset ratio (e.g., 1:1:1).

[0039] Model performance: The detection accuracy (mAP@0.5) on the test set reaches the preset accuracy threshold, the recall rate reaches the preset recall rate threshold, and the single-frame inference time is within the preset inference time threshold on the embedded computing platform, meeting the real-time requirements.

[0040] Based on the pixel coordinates (u,v) of the center point of the 2D bounding box output by the deep learning model in the image, combined with the camera calibration intrinsic parameter matrix K and extrinsic parameter matrix [R|t], the 3D centroid coordinates (X,Y,Z) of the target organism are calculated using the principle of triangulation. The calculation formula is: [X,Y,Z]ᵀ=Z_c×K⁻¹×[u,v,1]ᵀ, where Z_c is the depth information. When using a monocular camera, Z_c is estimated by the ratio of the known body length of the target organism (approximately 3mm-5mm for mosquitoes) to the pixel size; when using binocular stereo vision, Z_c is obtained through parallax calculation, resulting in higher accuracy. The 3D centroid coordinates (X,Y,Z) serve as the projection reference point for the optical deflection positioning module, and the coordinate data is updated at a frequency of 100Hz to ensure real-time tracking of fast-moving targets.

[0041] Furthermore, the image capturing module also includes: A non-target large biometric identification logic unit, connected to the deep learning model unit and the control module, is used to identify whether a non-target large organism appears in the monitoring area. When such an organism is identified, a safety cut-off signal is output to the control module. The non-target large organisms include humans and pets (such as cats and dogs). The control module is also used to immediately forcibly cut off the power supply of the laser generation module or drive the optical deflection positioning module to project the light beam into the safe light absorption area when it receives a safety cut-off signal sent by the non-target large biometric logic unit.

[0042] In this embodiment of the invention, the non-target large biometric logic unit further includes: The identification model subunit is used to classify each detected target using a convolutional neural network (CNN)-based target classification model. It takes all detection results from the deep learning model as input and classifies each target to determine whether it is a non-target large organism. For example, the training dataset for the classification model contains 5000 human images and 3000 pet images, covering different poses, distances, and lighting conditions, achieving a classification accuracy of 99.5%.

[0043] The safety distance determination subunit is used to calculate the distance to the laser emission point based on the three-dimensional spatial coordinates of a non-target large organism when it is detected. When this distance is less than the safety distance threshold (e.g., 10m), a safety protection action is triggered.

[0044] The safety protection subunit is used to immediately and forcibly cut off the power supply to the laser generating module based on the command sent by the safety distance determination subunit to trigger the safety protection action, such as a cutoff response time of <1ms. Simultaneously, it drives the optical deflection and positioning module to deflect the laser beam to a safe absorption zone. This safe absorption zone is located at the default deflection position and uses a high-absorption-rate material (e.g., absorptivity ≥99.5%) to ensure that the laser beam will not cause harm to the surrounding environment in an emergency shutdown state.

[0045] The recovery mechanism subunit is used to automatically restore the power supply to the laser generation module and normal control operations after a preset time (e.g., 5 seconds) has elapsed since a non-target large organism leaves the monitoring area and safety has been confirmed.

[0046] Furthermore, the system also includes The trajectory prediction module, connected to the image acquisition module and the optical deflection positioning module, is used to calculate the flight vector of the target organism based on the continuous frame data acquired by the image acquisition module, and to direct the optical deflection positioning module to project a regional light spot onto the expected position at the front end of the flight vector. Furthermore, the trajectory prediction module also includes a target trajectory buffer, a flight vector calculation module, and a front-end expected position calculation module. The trajectory buffer stores the target position data P1, P2, ..., P for the most recent N frames (N is a natural number, e.g., N=10, corresponding to a time window of approximately 333ms). n Where Pᵢ=(Xᵢ,Yᵢ,Zᵢ) are the three-dimensional barycenter coordinates of the i-th frame; the flight vector calculation module is used to perform linear fitting on the position data in the buffer using the least squares method to obtain the flight vector V=( , , The formulas for calculating each component of the flight vector are as follows: =Σ(Xᵢ-X̄)(tᵢ-t̄) / Σ(tᵢ-t̄)², , Similarly, where X̄ is the position mean, t̄ is the time mean, and tᵢ is the acquisition timestamp of the i-th frame image;

[0047] The front-end expected position calculation module is used to calculate the expected position of the target organism after time Δt, P_aim = P_current + V × Δt, based on the flight vector V and the total response delay Δt. P_current represents the current position of the target organism, i.e., the three-dimensional centroid coordinates (X, Y, Z) calculated in the latest frame; the total response delay Δt includes image acquisition delay (approximately 33 ms), image processing delay (approximately 8 ms), galvanometer response delay (approximately 0.2 ms), and communication delay (approximately 1 ms). P_aim is used as the aiming coordinate of the optical deflection positioning module, which directs the galvanometer to project a regional light spot onto this front-end expected position.

[0048] Furthermore, after locking onto the target organism, the optical deflection positioning module controls the regional light spot to perform a small-range high-frequency jitter scan around the spatial coordinates to further expand the instantaneous impact coverage and reduce the miss rate caused by the target organism's sudden turning.

[0049] In some alternative embodiments, the high-frequency perturbation scanning strategy is as follows: Disturbance signal: A high-frequency sinusoidal disturbance signal is superimposed on the X-axis and Y-axis drive signals of the galvanometer. The disturbance frequency is 100Hz-500Hz (adjustable), and the disturbance amplitude is 0.5 times to 1 times the spot radius (adjustable). Scan trajectory: The perturbation scan trajectory is a Lissajous pattern. The ratio of the perturbation frequency fx on the X-axis to the perturbation frequency fy on the Y-axis, fx:fy, is set to an irrational ratio (e.g., 1:√2) to ensure that the perturbation scan trajectory uniformly covers the target area after time integration, leaving no scan blind spots. Effect: Through high-frequency perturbation scanning, the instantaneous impact coverage area is expanded to 2-4 times the area of ​​the laser spot. When the target organism suddenly turns during laser irradiation, the perturbation scanning spot still has a certain probability of covering the new position of the target, thereby reducing the miss rate. Experimental data shows that after enabling high-frequency perturbation scanning, the miss rate was reduced from approximately 15% before it was enabled to below 5%.

[0050] Corresponding to the wide-area vision-guided laser biocontrol system described in the above embodiments, Figure 4 This is a flowchart of a laser-based biological control method based on wide-area vision guidance, according to a preferred embodiment of the present invention. For ease of explanation, only the parts relevant to the embodiments of this application are shown. The method includes the following steps:

[0051] S1: Acquire image information of the monitoring area and identify the spatial coordinates of target organisms in the image information; In some alternative embodiments, an image capturing module can be used to capture image information of the monitoring area and identify the spatial coordinates of the target organism in the image information; S2: Generate at least one laser beam; In some alternative embodiments, at least one laser beam can be generated by a laser generating module, which employs a semiconductor laser diode and may use the following technical parameters: The working wavelength is set to a visible light band that has a sensitive stimulating effect on the insect visual system, such as 532nm (green light), 450nm (blue light), or 940nm (infrared light). In an optional embodiment of the present invention, the working wavelength is set to 532nm (green light). This wavelength has a strong stimulating effect on the insect visual system. The compound eyes of insects are most sensitive to the green light band. The 532nm laser can effectively interfere with the flight control system of mosquitoes, causing them to lose flight attitude stability, disrupt wing vibration frequency, and ultimately lead to wing damage or loss of flight ability. The output power is set to a level sufficient to destroy the wings of the target organism and the power density is within a safe range after the light spot is expanded. For example, the output power is 5W. This power level ensures that it is sufficient to destroy the wings of mosquitoes (requiring an energy density of about 0.5J / cm²). After the light spot is expanded to 5mm-30mm by the second optical module, the power density in the light spot area is controlled within a safe range to avoid thermal damage to the surrounding environment. Beam quality: Beam quality factor M² < 1.5, divergence angle < 1.5 mrad, ensuring that the laser beam maintains good directionality during transmission; Driving method: Constant current source driving is adopted, and the stability of the driving current is ±0.1%, ensuring that the output power fluctuation is <±2%. The modulation frequency can reach a preset high frequency level, such as 20kHz, and supports pulse working mode. The laser emission timing can be adjusted as needed.

[0052] S3: Adjust the projection angle of the laser beam according to the spatial coordinates of the target organism; In some alternative embodiments, an optical deflection positioning module can adjust the projection angle of the laser beam according to the spatial coordinates of the target organism; the optical deflection positioning module employs a biaxial galvanometer scanner and can use the following technical parameters: The scanning angle range is set to a preset angle range (e.g., ±20°, optical angle) to ensure that the scanning range covers the preset monitoring area (e.g., ±17.6m) at the maximum effective distance (e.g., 50m). The scanning speed significantly exceeds the maximum flight speed of the target organism; for example, a scanning speed of 5000° / s corresponds to a linear velocity of 436m / s at the maximum effective distance of 50m, which is far greater than the flight speed of the target organism (such as a mosquito) (about 1-2m / s), ensuring that the laser beam can quickly track and cover the target organism. The positioning accuracy meets the preset accuracy requirements for precise projection of regional light spots; for example, a positioning accuracy of 0.005° corresponds to a positioning accuracy of 4.4mm at the maximum effective distance of 50m (better than the preset minimum light spot size), which meets the precise projection requirements of regional light spots (5mm-30mm). The small signal response time is within the preset millisecond-level response time threshold to ensure real-time response to fast-moving targets; for example, the response time for 1% of the full stroke is 0.2ms to ensure real-time response to fast-moving targets.

[0053] In this embodiment of the invention, the deflection angles of the X-axis and Y-axis galvanometers are calculated using a coordinate transformation algorithm based on the output three-dimensional spatial coordinates of the target organism. The coordinate transformation process is as follows: First, the world coordinates (X, Y, Z) of the target organism are converted into spherical coordinates (θ, φ, r) relative to the laser emission point, where θ is the horizontal deflection angle, φ is the vertical deflection angle, and r is the distance. Then, the spherical coordinates (θ, φ) are mapped to the driving voltages of the X-axis and Y-axis galvanometers, driving the galvanometers to deflect to the corresponding angles. The calculation delay of the coordinate transformation is <0.1ms, ensuring the overall response speed of the system.

[0054] S4: Convert the laser beam into a regional light spot with a specific diffusion angle; In some alternative embodiments, the laser beam is converted into a regional spot with a specific diffusion angle by a second optical module; the second optical module is an aspherical diffusion lens or a holographic diffuser. The aspherical diffusion lens is manufactured using a precision molding process, and its surface shape is optimized by optical design software, enabling it to precisely diffuse the incident parallel laser beam into a spot of a specific shape. The holographic diffuser achieves beam diffusion by recording a specific diffraction pattern on an optical substrate; the diffraction pattern is optimized using an Iterative Fourier Transform (IFTA) algorithm, enabling the creation of spots of arbitrary shape and energy distribution.

[0055] The second optical module expands the originally single-point laser beam into a spot with a predetermined geometry, which can be circular, elliptical, or rectangular. A circular spot is suitable for isotropic target coverage scenarios; an elliptical spot is suitable for scenarios where coverage is stretched along the target's flight direction; and a rectangular spot is suitable for area scanning scenarios. The spot shape can be switched by changing different second optical modules or adjusting the spacing of the diffusion lenses.

[0056] In some alternative embodiments, the cross-sectional diameter of the regional light spot is 5 mm to 30 mm. The choice of light spot diameter depends on the distance between the target organism and the system: a smaller light spot (5 mm to 10 mm) is used for close range (5 m to 15 m) to obtain higher energy density; a medium light spot (10 mm to 20 mm) is used for medium range (15 m to 30 m); and a larger light spot (20 mm to 30 mm) is used for long range (30 m to 50 m) to ensure sufficient coverage area. The energy distribution of the light spot exhibits a flat-top profile or a Gaussian distribution. A flat-top profile ensures that the energy density at each point within the light spot is basically uniform, which is beneficial for uniform strikes; a Gaussian distribution results in the highest energy density at the center of the light spot, which is suitable for scenarios requiring concentrated energy. Within the coverage area of ​​the light spot, the energy density must meet a threshold (approximately 0.5 J / cm²) sufficient to destroy the target organism's wings.

[0057] S5: Control and adjust the projection position of the regional light spot to cover the target biological area.

[0058] In some alternative embodiments, the optical deflection positioning module is controlled to cover the area where the target organism is located with a regional light spot. After locking onto the target organism, the optical deflection positioning module controls the regional light spot to perform a small-range high-frequency jitter scan around the spatial coordinates to further expand the instantaneous impact coverage and reduce the miss rate caused by the target organism's sudden turning.

[0059] In some alternative embodiments, the high-frequency perturbation scanning strategy is as follows: Disturbance signal: A high-frequency sinusoidal disturbance signal is superimposed on the X-axis and Y-axis drive signals of the galvanometer. The disturbance frequency is 100Hz-500Hz (adjustable), and the disturbance amplitude is 0.5 times to 1 times the spot radius (adjustable). Scan trajectory: The perturbation scan trajectory is a Lissajous pattern. The ratio of the perturbation frequency fx on the X-axis to the perturbation frequency fy on the Y-axis, fx:fy, is set to an irrational ratio (e.g., 1:√2) to ensure that the perturbation scan trajectory uniformly covers the target area after time integration, leaving no scan blind spots. Effect: Through high-frequency perturbation scanning, the instantaneous impact coverage area is expanded to 2-4 times the area of ​​the laser spot. When the target organism suddenly turns during laser irradiation, the perturbation scanning spot still has a certain probability of covering the new position of the target, thereby reducing the miss rate. Experimental data shows that after enabling high-frequency perturbation scanning, the miss rate was reduced from approximately 15% before it was enabled to below 5%.

[0060] Furthermore, step S1 may also include: S11: Collect image information of the monitoring area and detect the ambient illuminance in real time through the light source detection logic; S12: When the ambient illuminance is lower than the preset threshold, automatically switch to infrared mode or start the active infrared fill light; In an optional embodiment, image information of the monitored area can be acquired through a multimodal sensing unit, which defaults to visible light mode. The multimodal sensing unit includes a visible light camera module and an infrared camera module. For example, the visible light camera module has a resolution of 2-8 MHz, a frame rate of 30 fps, and a field of view of 90°×60°, used to acquire high-resolution texture images of the target area under good lighting conditions. The infrared camera module operates in the 850nm-940nm band, has a frame rate of 25 fps, and a temperature resolution of 0.05℃, used to acquire thermal radiation images of the target organism in low-light environments.

[0061] In an optional embodiment, the light source detection logic unit monitors the ambient illuminance in real time using an ambient light sensor. For example, the ambient light sensor has an accuracy of ±5%, a range of 0-200,000 lux, and a sampling frequency of 10 Hz. When the ambient illuminance is below a preset threshold (e.g., 50 lux), the light source detection logic automatically triggers a mode switch, switching the multimodal sensing unit from visible light mode to infrared mode. Simultaneously, an active infrared supplemental light is activated. For example, this supplemental light has a working wavelength of 850 nm, a power of 20 W, and an illumination angle of 60°, providing uniform infrared supplemental lighting within a range of 5 m-30 m. Through this mechanism, the system can effectively acquire the spatial coordinates of the target organism under various lighting conditions, including daytime, nighttime, cloudy days, and tree shade.

[0062] Furthermore, step S1 may also include: S13: Use a deep learning model to identify targets in the collected image information, distinguish between target organisms (such as mosquitoes) and non-target dynamic objects, and calculate the three-dimensional centroid coordinates of the target organism in the monitoring area as a projection reference point. In this embodiment of the invention, the deep learning model is trained to distinguish between target organisms (such as mosquitoes) and non-target dynamic objects (such as swaying leaves, flying birds, falling objects, etc.), and calculates the three-dimensional centroid coordinates of the target organism in the monitoring area as a projection reference point.

[0063] The deep learning model adopts an improved YOLO architecture, specifically including: Model structure: The backbone network (Backneck) can use CSPDarknet53, the neck network (Neck) can use a multi-scale feature fusion network (such as PANet+FPN) to perform multi-scale feature fusion, and the detection head (Head) contains multiple (such as 3) detection layers of different scales, corresponding to small targets (mosquitoes, body length 2mm-6mm), medium targets and large targets respectively; Training dataset: A labeled image dataset of a predetermined size (e.g., 15,000 images), where target biological images constitute a predetermined proportion, covering different poses, lighting conditions, and background conditions; non-target dynamic object images constitute a predetermined proportion, including various easily confused interference object types (e.g., leaves, birds, falling objects, etc.). Data augmentation strategies include random rotation (predetermined angle range, e.g., ±15°), brightness jitter (predetermined brightness variation range), Gaussian noise addition, random cropping, etc.

[0064] Training parameters: An adaptive learning rate optimizer (e.g., Adam optimizer) is used, with an initial learning rate of a preset value (e.g., 0.001), a batch size of a preset value (e.g., 16), and a preset number of training epochs (e.g., 300 epochs). The loss function consists of three parts: bounding box regression loss, confidence loss, and classification loss, with the weights of the three parts set to a preset ratio (e.g., 1:1:1).

[0065] Model performance: The detection accuracy (mAP@0.5) on the test set reaches the preset accuracy threshold, the recall rate reaches the preset recall rate threshold, and the single-frame inference time is within the preset inference time threshold on the embedded computing platform, meeting the real-time requirements.

[0066] Based on the pixel coordinates (u,v) of the center point of the 2D bounding box output by the deep learning model in the image, combined with the camera calibration intrinsic parameter matrix K and extrinsic parameter matrix [R|t], the 3D centroid coordinates (X,Y,Z) of the target organism are calculated using the principle of triangulation. The calculation formula is: [X,Y,Z]ᵀ=Z_c×K⁻¹×[u,v,1]ᵀ, where Z_c is the depth information. When using a monocular camera, Z_c is estimated by the ratio of the known body length of the target organism (approximately 3mm-5mm for mosquitoes) to the pixel size; when using binocular stereo vision, Z_c is obtained through parallax calculation, resulting in higher accuracy. The 3D centroid coordinates (X,Y,Z) serve as the projection reference point for the optical deflection positioning module, and the coordinate data is updated at a frequency of 100Hz to ensure real-time tracking of fast-moving targets.

[0067] S14: Determine whether a non-target large biometric entity has appeared in the monitoring area through the non-target large biometric identification logic. If so, immediately cut off the power supply of the laser generation module or control the beam to be projected to the safe light absorption area. If not, output the three-dimensional center of gravity coordinates to step S3.

[0068] In this embodiment of the invention, the non-target large-scale biometric identification logic is as follows: A target classification model based on convolutional neural networks (CNNs) is employed. Using all detection results output by the deep learning model as input, each detected target is classified to determine whether it is a non-target large organism. For example, the training dataset for the classification model contains 5000 human images and 3000 pet images, covering different poses, distances, and lighting conditions, achieving a classification accuracy of 99.5%.

[0069] When a large, non-target organism is detected, the distance to the laser emission point is calculated based on its three-dimensional spatial coordinates. If this distance is less than a safe distance threshold (e.g., 10m), a safety protection action is triggered.

[0070] Based on the command sent by the safety distance determination subunit to trigger safety protection actions, the power supply to the laser generating module is immediately and forcibly cut off, with a cutoff response time of <1ms. Simultaneously, the beam is deflected to a safe absorption zone. This safe absorption zone is set at the default deflection position and uses a high-absorption-rate material (e.g., absorptivity ≥99.5%) to ensure that the laser beam will not cause harm to the surrounding environment in an emergency shutdown state.

[0071] Once a non-target large organism leaves the monitoring area, after a preset delay (e.g., 5 seconds) to confirm safety, the power supply to the laser generating module is automatically restored and normal control operations resume.

[0072] Furthermore, the laser biocontrol method based on wide-area vision guidance also includes a trajectory prediction step; Based on the acquired continuous frame data of the image, the flight vector of the target organism is calculated, and the regional light spot is directed to be projected onto the expected position at the front end of the flight vector. The method for calculating flight vectors is as follows: The position data in the target trajectory buffer are linearly fitted using the least squares method to obtain the target organism's flight vector V=( , , The formulas for calculating each component of the flight vector are as follows: =Σ(Xᵢ-X̄)(tᵢ-t̄) / Σ(tᵢ-t̄)², , Similarly, where X̄ is the position mean, t̄ is the time mean, tᵢ is the acquisition timestamp of the i-th frame image, and the trajectory buffer is used to store the target position data P1, P2, ..., P of the most recent N frames (e.g., N=10, corresponding to a time window of approximately 333ms). n , where Pᵢ=(Xᵢ,Yᵢ,Zᵢ) are the three-dimensional barycenter coordinates of the i-th frame;

[0073] The method for calculating the expected front-end position is as follows: Based on the flight vector V and the total response delay Δt, the expected position of the target organism after Δt time is calculated as P_aim = P_current + V × Δt. P_current represents the current position of the target organism, i.e., the three-dimensional centroid coordinates (X, Y, Z) calculated in the latest frame; the total response delay Δt includes image acquisition delay (approximately 33 ms), image processing delay (approximately 8 ms), galvanometer response delay (approximately 0.2 ms), and communication delay (approximately 1 ms). Using P_aim as the aiming coordinate, the galvanometer is directed to project a regional light spot onto the expected position at the front end.

[0074] Furthermore, the method also includes A dichroic mirror is placed along the transmission path of the laser beam to make the laser emission axis and the receiving axis coaxial. In a specific embodiment, the dichroic mirror implements a coaxial optical path design, and the specific optical path architecture is as follows:

[0075] A laser beam (e.g., 532nm green light) generated by the laser generation module is directed towards a dichroic mirror. The dichroic mirror is positioned at a preset tilt angle (e.g., 45°) along the laser beam's transmission path, and its surface has high reflectivity (e.g., reflectivity ≥95%) for the laser wavelength, reflecting the laser beam onto the reflective surface of the optical deflection and positioning module. The optical deflection and positioning module rotates the reflective surface according to instructions from the control module, reflecting the laser beam towards the target organism in space, thus projecting the laser beam onto the target organism.

[0076] Ambient light (containing visible light components) reflected from the surface of the target organism is directed towards the reflective mirror of the optical deflection and positioning module in the opposite direction to the laser emission path. After being reflected by the galvanometer, it is directed towards the dichroic mirror. The dichroic mirror has high transmittance (e.g., transmittance ≥ 90%) in the visible light band, transmitting the ambient light reflected from the target organism to the image acquisition unit (e.g., camera) sensor of the image acquisition module for imaging, which is then used by the deep learning model unit for target recognition and coordinate calculation.

[0077] The laser emission direction and the image acquisition module's receiving direction share the same galvanometer reflector, and after being combined by the dichroic mirror, their optical axes are completely aligned. That is, the center of the image captured by the image acquisition module always corresponds to the laser's pointing point, eliminating the need for software parallax compensation. This coaxial design ensures consistency between the laser projection point and the visual monitoring center, avoiding aiming deviations caused by optical axis offset and improving the system's projection accuracy onto the target organism.

[0078] Furthermore, the method also includes A first optical module is set at the laser output end to adjust the spot size and laser energy of the laser beam.

[0079] In some optional implementations, the first optical module can adopt an adjustable collimating lens group (stepper motor drives lens displacement) + electric laser attenuation component; its control logic is as follows: the image acquisition module calculates the three-dimensional distance of the target, the control module outputs instructions according to the distance range, drives the motor to adjust the collimating lens spacing to match the light spot reference size at the corresponding distance; the attenuation component is adjusted synchronously to match the energy density required for the distance, ensuring that the kill threshold (e.g., 0.5J / cm²) is maintained within the light spot; the collimating lens group consists of at least two spherical lenses, which can adopt a 4f system design to compress the divergence angle of the laser beam to <0.5mrad, ensuring that the beam diameter expansion at the maximum effective distance (e.g., 50m) does not exceed the preset maximum beam diameter (e.g., 25mm).

[0080] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by program instructions and related hardware. The program can be stored in a computer-readable storage medium, such as ROM, RAM, disk, optical disk, etc.

[0081] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A laser-based biological control system based on wide-area vision guidance, characterized in that, The system includes: The image acquisition module is used to acquire image information of the monitoring area and identify the spatial coordinates of the target organism in the image information; A laser generating module is used to generate at least one laser beam under the control of the control module; An optical deflection positioning module, connected to the control module, is positioned on the transmission path of the laser beam and is used to adjust the projection angle of the laser beam according to the instructions of the control module. The second optical module is located on the light-emitting side of the optical deflection and positioning module and is used to convert the laser beam into a regional light spot with a specific diffusion angle. The control module is connected to the image acquisition module, the laser generation module, and the optical deflection positioning module, respectively. It is used to control the optical deflection positioning module to adjust the projection position of the regional light spot according to the spatial coordinates output by the image acquisition module, so as to cover the target biological area.

2. The laser-based biological control system based on wide-area vision guidance as described in claim 1, characterized in that, The image capturing module also includes: A multimodal sensing unit is used to acquire image information of the monitoring area in visible light mode or infrared mode, and obtain image data of the target organism. A light source detection logic unit, connected to the multimodal sensing unit, is used to detect ambient illuminance in real time. When the ambient illuminance is lower than a preset threshold, the multimodal sensing unit is triggered to automatically switch its working mode. The working modes include visible light mode and infrared mode.

3. The laser-based biological control system based on wide-area vision guidance as described in claim 2, characterized in that, The image capturing module also includes: The deep learning model unit, connected to the multimodal sensing unit, is used to perform target recognition on the acquired image information through the deep learning model, distinguish between target organisms and non-target dynamic objects, and calculate the three-dimensional centroid coordinates of the target organism in the monitoring area as a projection reference point.

4. The laser-based biological control system based on wide-area vision guidance as described in claim 3, characterized in that, The image capturing module also includes: A non-target large biometric identification logic unit, connected to the deep learning model unit and the control module, is used to identify whether a non-target large organism appears in the monitoring area. When such an organism is identified, a safety cut-off signal is output to the control module. The non-target large organisms include humans and pets. The control module is also used to immediately forcibly cut off the power supply of the laser generation module or drive the optical deflection positioning module to project the light beam into the safe light absorption area when it receives a safety cut-off signal sent by the non-target large biometric logic unit.

5. The laser-based biological control system based on wide-area vision guidance as described in claim 4, characterized in that, The non-target large biometric logic unit also includes: The identification model subunit is used to adopt a target classification model based on a convolutional neural network. It takes all the detection results output by the deep learning model as input, classifies each detected target, and determines whether it is a non-target large organism. The safety distance determination subunit is used to calculate the distance to the laser emission point based on the three-dimensional spatial coordinates of a non-target large organism when it is detected. If the distance is less than the safety distance threshold, a safety protection action is triggered. The safety protection subunit is used to immediately and forcibly cut off the power supply of the laser generating module based on the command sent by the safety distance determination subunit to trigger the safety protection action, and at the same time drive the optical deflection positioning module to deflect the beam to a safe light absorption area. The recovery mechanism subunit is used to automatically restore the power supply to the laser generation module and resume normal control operations after a preset delay and confirmation of safety following the departure of a non-target large organism from the monitoring area.

6. The laser-based biological control system based on wide-area vision guidance as described in claim 1, characterized in that, The system also includes: The trajectory prediction module, connected to the image acquisition module and the optical deflection positioning module, is used to calculate the flight vector of the target organism based on the continuous frame data acquired by the image acquisition module, and output the expected position of the front end as the aiming coordinates to the optical deflection positioning module.

7. The laser-based biological control system based on wide-area vision guidance as described in claim 6, characterized in that, The trajectory prediction module also includes a target trajectory buffer, a flight vector calculation module, and a front-end expected position calculation module. The trajectory buffer is used to store the target position data P1, P2, ..., P of the most recent N frames. n Where Pᵢ=(Xᵢ,Yᵢ,Zᵢ) are the three-dimensional barycenter coordinates of the i-th frame; the flight vector calculation module is used to perform linear fitting on the position data in the buffer using the least squares method to obtain the flight vector V=( , , ); The front-end expected position calculation module is used to calculate the expected position P_aim of the target organism after time Δt based on the flight vector V and the total response delay time Δt. The total response delay Δt includes image acquisition delay, image processing delay, galvanometer response delay and communication delay. P_aim is used as the aiming coordinate of the optical deflection positioning module to direct the galvanometer to project a regional light spot onto the front-end expected position.

8. The laser-based biological control system based on wide-area vision guidance as described in any one of claims 1-7, characterized in that, The system also includes a dichroic mirror, which is placed on the transmission path of the laser beam to reflect the laser beam to the optical deflection and positioning module, and at the same time transmits the ambient light reflected by the target organism to the image capturing module, so that the laser emission optical axis and the receiving optical axis of the image capturing module are coaxial.

9. The laser-based biological control system based on wide-area vision guidance as described in claim 8, characterized in that, The system also includes The first optical module, located at the laser emission end and connected to the control module, is used to adjust the laser spot size and energy.

10. A laser-based biological control method based on wide-area vision guidance, characterized in that, The method includes the following steps: Acquire image information of the monitoring area and identify the spatial coordinates of target organisms in the image information; Generate at least one laser beam; The projection angle of the laser beam is adjusted according to the spatial coordinates of the target organism; The laser beam is converted into a regional spot with a specific diffusion angle; The projection position of the regional light spot is controlled and adjusted to cover the target biological area.

11. The laser-based biological control method based on wide-area vision guidance as described in claim 10, characterized in that, The step of "acquiring image information of the monitoring area and identifying the spatial coordinates of the target organism in the image information" further includes: Collect image information of the monitoring area and detect the ambient illuminance in real time through light source detection logic; When the ambient illuminance is lower than the preset threshold, it will automatically switch to infrared mode or activate the active infrared fill light. The collected image information is used to identify targets by using a deep learning model, distinguishing target organisms from non-target dynamic objects, and calculating the three-dimensional centroid coordinates of the target organism in the monitoring area as a projection reference point. Determine whether a non-target large organism is present in the monitoring area. If so, immediately cut off the laser power supply or project the beam into a safe light-absorbing area. If not, output the three-dimensional centroid coordinates.

12. The laser-based biological control method based on wide-area vision guidance as described in claim 11, characterized in that, The method also includes a trajectory prediction step; Based on the acquired continuous frame data of the image, the flight vector of the target organism is calculated, and the regional light spot is projected onto the expected position at the front end of the flight vector.

13. The laser-based biological control method based on wide-area vision guidance as described in any one of claims 10-12, characterized in that, The method also includes A dichroic mirror is placed along the transmission path of the laser beam to make the laser emission axis and the receiving axis coaxial.

14. The laser-based biological control system based on wide-area vision guidance as described in claim 13, characterized in that, The method also includes A first optical module is set at the laser output end to adjust the spot size and laser energy of the laser beam.

Images