Laser bird repelling device and use method

Abstract

The invention provides an automatic bird repelling device and a using method, and belongs to the technical field of laser bird repelling. Comprising a supporting platform fixedly arranged relative to the ground. The wide-angle camera is arranged on the supporting platform and is used for carrying out continuous imaging on a set monitoring risk area so as to realize landing detection of a target; the two-axis holder is arranged on the supporting platform and rotates in the horizontal direction and the vertical direction relative to the supporting platform; the zoom camera and the laser bird repeller are both arranged at the movable end of the two-axis holder, and the zoom camera performs secondary identification on a target in a monitoring risk area where the fallen birds are located; the laser expeller emits a laser beam to an area where the target is secondarily identified as the harmful bird; the control unit is integrated on the supporting platform and used for receiving the landing detection result of the target of the wide-angle camera and the secondary recognition result of the zoom camera on the target and further driving the two-axis holder or the laser expeller to act.

Description

Technical Field

[0001] The present invention relates to the technical field of laser bird repelling technology, and in particular to a laser bird repelling device and a use method thereof. Background Art

[0002] High-value aquaculture and fruit planting areas are often attacked by birds, causing significant economic losses to farmers and growers, especially during the maturity period of the products, when a large number of birds come to peck at them, which is hard to prevent. Moreover, birds have great adaptability, and conventional methods can only have a short-term partial effect, and it is difficult to maintain a long-term repelling effect. Common bird repelling methods are as follows: 1. Artificial use of long poles or gestures, which has little effect on birds at a distance; 2. Air-blast bombs are fired at the flock of birds, and the loud sound produced by the air-blast is used to scare the birds. This method is cumbersome to implement, and after the air-blast, the birds will quickly return, and the long-term use effect is limited; 3. Use ultrasound to stimulate the birds, causing them to feel physiological discomfort, so as to achieve the purpose of repelling them. However, this method has a limited range of action, and large-scale use requires the deployment of multiple ultrasonic devices, which is difficult to install and maintain. On the other hand, this method only has a good effect on specific birds and has poor universality; 4. Use the strong light of the laser to illuminate the bird's eyes, causing the birds to feel uncomfortable, so as to achieve the purpose of repelling them. This method requires the laser to be aimed at the bird's eyes, which is very difficult to operate at a long distance and may cause damage to the bird's eyes.

[0003] Therefore, it is very necessary to provide a laser bird-repelling device and a method of use, which can identify whether the target entering the defense area is a harmful bird, predict the movement trajectory and resting position of the harmful bird, and use a laser virtual rod to drive away the resting harmful bird without causing harm to the bird, thereby improving the actual effect and efficiency of the drive away. Summary of the invention

[0004] In view of this, the present invention proposes a laser bird repellent device and a method for using the device, which performs a target detection on a preset area, identifies the target as having a risk of falling, then enables secondary target recognition and further performs laser simulation stick repelling.

[0005] In one aspect, the present invention provides a laser bird-repelling device, comprising:

[0006] A support platform is fixedly arranged relative to the ground;

[0007] A wide-angle camera, arranged on the supporting platform, is used to continuously image the set monitoring risk area, establish a unified visual reference, and realize the landing detection of the target;

[0008] A two-axis gimbal, arranged on the support platform, and capable of performing horizontal and vertical rotational motion relative to the support platform;

[0009] A zoom camera, arranged on the active end of the two-axis gimbal, is used to focus on the monitoring risk area where the landed harmful bird is located, so as to realize secondary identification of the target;

[0010] A laser repeller is arranged at the active end of the two-axis gimbal, and is used to emit a laser beam to the area where the zoom camera secondary identifies the target as a harmful bird, wherein the laser repeller is arranged in parallel with the optical path of the zoom camera and the pointing direction is synchronized;

[0011] The control unit is integrated on the supporting platform and is respectively connected to the wide-angle camera, the two-axis gimbal, the zoom camera and the laser drive away for communication. The control unit is used to receive the landing detection result of the target of the wide-angle camera and the secondary recognition result of the target of the zoom camera. The control unit also drives the two-axis gimbal or the laser drive away to operate.

[0012] On the basis of the above technical solution, preferably, the set monitoring risk area is to select an area at a certain height above the ground or water surface within the field of view of the wide-angle camera as the monitoring risk area.

[0013] Preferably, the wide-angle camera realizes the landing detection of the target, including the following contents:

[0014] The wide-angle camera pre-processes the acquired images;

[0015] Inter-frame difference processing: perform difference processing on two consecutive pre-processed frames, calculate the grayscale values ​​of the corresponding pixels in the two frames, subtract them, calculate the absolute value of the grayscale value subtraction result, and use the absolute value of the grayscale value subtraction result to generate a differential grayscale image;

[0016] Change pixel extraction: perform change detection on the differential grayscale image and extract the change area between the two frames of images;

[0017] Noise reduction processing: operate the change area through morphological algorithm, count the size of the change area, set the pixel area threshold of the change area, and set the pixel area threshold of the change area according to the number of pixels occupied by the farthest typical harmful bird in the field of view; filter out the change area that is not within the pixel area threshold range;

[0018] Target extraction: determine whether the remaining change area after noise reduction processing is within the set monitoring risk area. If the change area is within the set monitoring risk area, target extraction is performed on the change area. The center of gravity of the change area is calculated using the weighted centroid method. The center of gravity represents the target corresponding to the change area, and the target attribute information is stored in the target queue. If the change area is not within the set monitoring risk area, target extraction is not performed.

[0019] Target search, identifying and recording the corresponding relationship of each target in continuous images, and building a trajectory queue based on each target;

[0020] Trajectory tracking: track each target’s trajectory queue separately to obtain the trajectory data of each target: if the target’s trajectory is outside the set monitoring risk area, delete the target from the target queue; if the target’s trajectory is still within the set monitoring risk area and moving toward the ground, output the target’s trajectory data moving toward the ground to the landing trajectory queue;

[0021] Landing detection: perform landing detection on the targets in the landing trajectory queue, calculate the changes in the adjacent position information of the trajectory to determine whether the target has landed. If it is determined that the target has not landed, further track the trajectory; if it is determined that the target has landed, trigger the bird repelling mechanism; and

[0022] After the information is sent and the bird-repelling mechanism is triggered, the wide-angle camera sends the position of the landed target in a unified visual reference to the control unit, and the control unit drives the zoom camera and the laser repeller to further move and perform the repeller action.

[0023] Further preferably, the wide-angle camera preprocesses the acquired image, specifically, statistics the overall brightness and variance of the image, and calculates the clarity of the image using a gradient difference method. If the clarity of the image is less than a set clarity threshold, the gain of the wide-angle camera is adjusted, and the image after gain adjustment is subjected to histogram equalization to make the grayscale probability distribution of the image after histogram equalization uniform, thereby obtaining a preprocessed image; if the clarity of the image is not less than the set clarity threshold, the image is not preprocessed.

[0024] More preferably, the target search, identifying and recording the corresponding relationship of each target in the continuous images, and constructing a trajectory queue based on each target, specifically adopts a three-dimensional target search based on Manhattan distance, and on the basis of statistically analyzing the two-dimensional Manhattan distance between the current target and another target in the image at a certain moment, analyzes the relationship between the current target and the target in the image at the previous moment, and finds the continuous corresponding relationship of the current target in the time dimension; let the current target be k, and the other target be j, is the two-dimensional Manhattan distance between the current target k and the target j at the i-th moment; is the two-dimensional Manhattan distance between the current target k at the i-th moment and the target j at the i-1th moment; min(·) is the minimum value of the two-dimensional Manhattan distance between the current target k and another target in the i-th moment image; search the target queue at the i-1th moment, when the current target k at the i-th moment and the target j at the i-1th moment satisfy the formula , the information of the current target k and target j is saved in the trajectory queue, and the information of the current target k and target j in the continuous images is recorded.

[0025] Further preferably, the landing detection performs landing detection on the targets in the landing trajectory queue, calculates the changes in adjacent position information of the trajectory to determine whether the target has landed. Specifically, within a given time range, if the position change of the center of gravity of the target in the height direction within the set monitoring risk area does not exceed twice the distance of the longest path between any two points on the contour boundary corresponding to the initial stop moment, then it is determined that the target has landed.

[0026] Further preferably, after the control unit receives the position of the landed target in a unified visual reference, it obtains the deviation between the current position of the target and the current pointing direction of the zoom camera based on the calibration relationship between the zoom camera and the wide-angle camera, and adjusts the servo motor of the two-axis gimbal according to the deviation, so that the zoom camera and the laser bird repellent point to the current position of the target, and the zoom camera confirms the landed target as a harmful bird.

[0027] More preferably, the zoom camera confirms the harmful bird for the landed target, that is, the zoom camera obtains continuous images of the landed target, performs differential change detection on the continuous images again, and normalizes the results of the differential change. When the change range of the landed target exceeds a set threshold, the zoom camera determines that there is an active target in the field of view as a harmful bird, wherein the set threshold of the change range of the landed target is set according to the number of pixels occupied by the farthest typical harmful bird in the field of view; the zoom camera sends a repelling instruction to the laser bird repellent; after the laser bird repellent is started, the laser spot is enlarged and shot in the current pointing direction of the zoom camera, and a virtual rectangle surrounding the target center of gravity is constructed with the target center of gravity as the center, and a number of trajectories passing through the center of gravity are fitted in the rectangle, and the endpoints of each trajectory are located on two different sides of the virtual rectangle, and the output spot of the laser bird repellent moves back and forth along the endpoints of different trajectories.

[0028] In another aspect, the present invention provides a method for using a laser bird repellent device, comprising the following steps:

[0029] S1: configure the above-mentioned laser bird-repelling device; arrange a wide-angle camera and a two-axis gimbal at different positions of the supporting platform, respectively; the built-in servo motor of the two-axis gimbal realizes the pitch and rotation movement of the active end of the two-axis gimbal; the active end of the two-axis gimbal is configured with a zoom camera and a laser bird-repelling device; the laser bird-repelling device is arranged in parallel with the optical path of the zoom camera, and the pointing direction is synchronized; a unified visual reference is established based on the wide-angle camera, that is, the corresponding relationship between the camera coordinate system of the zoom camera and the camera coordinate system of the wide-angle camera is calibrated, so that the wide-angle camera obtains an image or the image of the zoom camera or is converted to a corresponding position in the world coordinate system; configure a control unit on the supporting platform;

[0030] S2: Within the field of view of the wide-angle camera, an area at a certain height above the ground or water surface is selected as the monitoring risk area. The wide-angle camera and zoom camera ignore objects outside the monitoring risk area.

[0031] S3: When a target enters the monitoring risk area, the wide-angle camera detects the target's landing, including the following:

[0032] S31: The wide-angle camera pre-processes the acquired image;

[0033] S32: inter-frame difference processing, performing difference processing on two consecutive pre-processed frames respectively, calculating the grayscale values ​​of corresponding pixel points in the two frames of images for subtraction, calculating the absolute value of the grayscale value subtraction result, and generating a differential grayscale image using the absolute value of the grayscale value subtraction result;

[0034] S33: extracting changed pixels, performing change detection on the differential grayscale image, and extracting the changed area between the two frames of images;

[0035] S34: noise reduction processing, operating the changed area through a morphological algorithm, counting the size of the changed area, setting a pixel area threshold of the changed area, and filtering out the changed area that is not within the pixel area threshold range;

[0036] S35: Target extraction, judging whether the remaining change area after the noise reduction process is within the set monitoring risk area: if the change area is within the set monitoring risk area, target extraction is performed on the change area, the center of gravity of the change area is calculated using the weighted centroid method, the target corresponding to the change area is represented by the center of gravity, and the attribute information of the target is stored in the target queue; if the change area is not within the set monitoring risk area, target extraction is not performed;

[0037] S36: target search, identifying and recording the corresponding relationship of each target in the continuous images, and constructing a trajectory queue based on each target;

[0038] S37: Tracking: Tracking each target’s track queue separately to obtain the track data of each target: if the track of the target track is outside the set monitoring risk area, the target is deleted from the target queue; if the track of the target track is still within the set monitoring risk area and moves toward the ground, the target track data moving toward the ground is output to the landing track queue;

[0039] S38: Landing detection, performing landing detection on the target in the landing trajectory queue, calculating the change of the adjacent position information of the trajectory to determine whether the target has landed, if it is determined that the target has not landed, further tracking the trajectory; if it is determined that the target has landed, triggering the bird repelling mechanism; and

[0040] S39: information sending, after the bird-repelling mechanism is triggered, the zoom camera sends the position of the landed target in the unified visual reference to the control unit;

[0041] S4: After receiving the position of the landed target in the unified visual reference, the control unit drives to change the current position of the zoom camera so that the zoom camera and the laser bird repellent point to the target position;

[0042] S5; the zoom camera performs a secondary confirmation on the landed target. When the change range of the landed target exceeds the set threshold, the zoom camera sends the content that the target is a harmful bird to the control unit. At this time, the control unit starts the laser bird repellent. The laser bird repellent has a built-in laser generator and a beam expander. The laser generator is used to generate a laser beam. The beam expander thickens the laser beam. The thickened laser beam is directed to the current pointing direction of the zoom camera, and a number of trajectories passing through the center of the target are constructed. The thickened laser beam sweeps along the trajectory and appears around the harmful bird, so that the bird has a visual illusion that the thickened laser beam is a physical stick. In order to avoid being hit by the stick, the bird will quickly flee, thereby achieving timely expulsion of the hunting harmful bird;

[0043] S6: The zoom camera continuously monitors the changes in the field of view and performs differential change detection on the continuously acquired images. If the field of view changes exceed the threshold, it means that there are still harmful birds in the field of view. At this time, the laser repeller is guided to point to the area with drastic changes in the field of view until the field of view change of the zoom camera is less than the threshold, indicating that there are no active targets in the field of view; when all harmful birds are driven away, the laser is turned off to complete the repeller operation.

[0044] Preferably, the pixel area threshold of the setting change area described in step S34 is obtained based on the value range of the pixel area of ​​the harmful bird species appearing in the current season in the wide-angle camera in combination with the bird population and seasonal changes in the aquaculture and fruit planting areas.

[0045] Compared with the prior art, the laser bird repellent device and the method of use provided by the present invention have the following advantages:

[0046] Beneficial effects:

[0047] (1) Based on a profound analysis of the hunting behavior characteristics of harmful birds in breeding or planting areas, the present invention proposes an efficient method for detecting harmful birds and identifying their landing behaviors, which only drives away harmful birds with potential hunting behaviors, greatly improving the response speed of the system and reducing the operating frequency of the system; in order to meet the needs of a wide range of monitoring areas, the camera for close-up imaging is obtained by combining a wide-angle camera with a zoom camera, which can dynamically adjust the depth of field and meet the needs of large-scale, close-up and fine imaging;

[0048] (2) Based on the biological characteristics of bird vision, the laser beam is thickened by expanding the laser spot. The enhanced laser beam will form a virtual stick effect in the bird's field of vision, and the bird will have the illusion of being driven away by the stick and quickly flee. The virtual stick does not need to directly stimulate the bird's eyes, and its efficiency is far greater than the laser's direct stimulation of the bird's eyes. Since there is no need to aim at the bird's eyes, the operation is faster and will not harm the bird's visual system.

[0049] (3) This solution uses a wide-angle camera to perform a primary identification and trajectory prediction of targets entering the monitoring risk area. When the target’s trajectory is pointing downward toward the ground or water, it is identified as a potential target. Then, a zoom camera performs a secondary target identification, making the identification of harmful birds more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 This is a device structure block diagram of a laser bird-repelling device and a method of using the present invention;

[0052] Figure 2 The present invention provides a flow chart of a laser bird repellent device and a method for using the laser bird repellent device to detect the landing of a target using a wide-angle camera. DETAILED DESCRIPTION

[0053] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] The existing laser bird repellent method uses strong laser light to illuminate the bird's eyes, causing discomfort to the bird and achieving the purpose of repelling it. This method requires the laser to be aimed at the bird's eyes, which is very difficult to operate at a long distance, takes a long time, and may cause damage to the bird's eyes. In view of this, Figure 1 As shown, on the one hand, the present invention provides a laser bird-repelling device, comprising:

[0055] The support platform is fixed relative to the ground and is used to carry the wide-angle camera, two-axis gimbal, zoom camera, laser repeller and control unit.

[0056] The wide-angle camera is arranged on the supporting platform to continuously image the set monitoring risk area, establish a unified visual reference, and realize the landing detection of the target. The set monitoring risk area mentioned here is to select an area at a certain height above the ground or water surface within the field of view of the wide-angle camera as the monitoring risk area, such as a height area several meters above the ground or water surface.

[0057] The two-axis gimbal is arranged on the support platform and performs horizontal and vertical rotational motion relative to the support platform. The movable end of the two-axis gimbal can perform pitching motion relative to the horizontal plane and rotational motion relative to the vertical direction.

[0058] The zoom camera is arranged on the active end of the two-axis gimbal and is used to focus on the monitoring risk area where the landed harmful birds are located, so as to realize secondary identification of the target; the market for wide-angle cameras is large, and the target size in the image is small. The detected targets may be fallen objects in the air, leaves, etc., and the shape and position of these targets usually remain unchanged after landing. Therefore, the use of a zoom camera for secondary identification is conducive to eliminating irrelevant targets, accurately identifying harmful birds, and improving the accuracy of subsequent laser bird repellent.

[0059] A laser repeller is arranged at the active end of the two-axis gimbal, and is used to emit a laser beam to the area where the zoom camera secondary identifies the target as a harmful bird, wherein the laser repeller is arranged in parallel with the optical path of the zoom camera and the pointing direction is synchronized;

[0060] The control unit is integrated on the supporting platform and is respectively connected to the wide-angle camera, the two-axis gimbal, the zoom camera and the laser drive away for communication. The control unit is used to receive the landing detection result of the target of the wide-angle camera and the secondary recognition result of the target of the zoom camera. The control unit also drives the two-axis gimbal or the laser drive away to operate.

[0061] Among them, the wide-angle camera realizes the landing detection of the target, which specifically includes the following contents:

[0062] 1) The wide-angle camera preprocesses the acquired image; specifically, the overall brightness and variance of the image are counted, and the image clarity is calculated using the gradient difference method. If the image clarity is less than the set clarity threshold, the gain of the wide-angle camera is adjusted, and the image after gain adjustment is subjected to histogram equalization processing to make the grayscale probability distribution of the image after histogram equalization processing uniform, thereby obtaining the preprocessed image; if the image clarity is not less than the set clarity threshold, the image is not preprocessed and directly enters the subsequent inter-frame difference processing operation.

[0063] 2) Inter-frame difference processing: perform difference processing on two consecutive pre-processed images respectively, calculate the grayscale values ​​of the pixels at corresponding positions in the two frames of images and subtract them, calculate the absolute value of the result of the grayscale value subtraction, and use the absolute value of the grayscale value subtraction result to generate a differential grayscale image.

[0064] 3) Change pixel extraction: perform change detection on the differential grayscale image and extract the change area between the two frames of images;

[0065] 4) Noise reduction processing: operate the change area through morphological algorithm, count the size of the change area, set the pixel area threshold of the change area, and set the pixel area threshold of the change area by the number of pixels occupied by the farthest typical harmful bird in the field of view; filter out the change area that is not within the pixel area threshold range;

[0066] 5) Target extraction, determine whether the remaining changed area after noise reduction processing is located in the set monitoring risk area: if the changed area is located in the set monitoring risk area, target extraction is performed on the changed area, and the center of gravity of the changed area is calculated using the weighted centroid method. The center of gravity represents the target corresponding to the changed area, and the attribute information of the target is stored in the target queue; if the changed area is not in the set monitoring risk area, target extraction is not performed; the target queue stores one or more targets that enter the set monitoring risk area, and can be saved separately in the order of the smallest row and column of the edge position of each target in the wide-angle image.

[0067] If there are multiple targets, the multiple targets are sorted into a target queue according to the order of rows and columns in the image. For images obtained at different times, the number and position of targets in the image are not exactly the same, so target queues corresponding to different times can be formed.

[0068] The weighted centroid method mentioned here is to binarize the target extraction area, such as retaining the grayscale value of the pixels within the target boundary and setting the grayscale value of the pixels outside the boundary to 0, and then obtain the target contour; use the grayscale of each pixel within the target contour and the grayscale centroid method to obtain the target's center of gravity

[0069] f(u, v) is the grayscale value at the image coordinate (u, v).

[0070] 6) Target search: identify and record the corresponding relationship of each target in continuous images, and build a trajectory queue based on each target.

[0071] The specific content is: using three-dimensional target search based on Manhattan distance, on the basis of statistically analyzing the two-dimensional Manhattan distance between the current target and another target in the image at a certain moment, analyzing the relationship between the current target and the target in the image at the previous moment, and finding the continuous correspondence of the current target in the time dimension; let the current target be k, the other target be j, is the two-dimensional Manhattan distance between the current target k and the target j at the i-th moment; is the two-dimensional Manhattan distance between the current target k at the i-th moment and the target j at the i-1th moment; min(·) is the minimum value of the two-dimensional Manhattan distance between the current target k and another target in the i-th moment image; search the target queue at the i-1th moment, when the current target k at the i-th moment and the target j at the i-1th moment satisfy the formula , the information of the current target k and target j is saved in the trajectory queue, and the information of the current target k and target j in the continuous images is recorded.

[0072] 7) Tracking: Track each target’s track queue separately to obtain the track data of each target: if the target’s track is outside the set monitoring risk area, delete the target from the target queue; if the target’s track is still within the set monitoring risk area and moves toward the ground, output the target’s track data moving toward the ground to the landing track queue;

[0073] In actual operation, the trajectory data can be converted into three-dimensional coordinates in the world coordinate system, and the trajectory curve of the corresponding target can be fitted according to the three-dimensional coordinates. If the trajectory curve is directed toward the ground of the set monitoring risk area, the current target will be retained in the target queue. If it is judged that the trajectory curve is directed outside the set monitoring risk area, it will continue to track for several sampling cycles. After confirming that the target has moved outside the set monitoring risk area, the target will be removed from the target queue.

[0074] 8) Landing detection: perform landing detection on the targets in the landing trajectory queue, calculate the changes in the adjacent position information of the trajectory to determine whether the target has landed. If it is determined that the target has not landed, further trajectory tracking is performed; if it is determined that the target has landed, the bird repelling mechanism is triggered.

[0075] Specifically, within a given time range, if the change in the height direction of the target's center of gravity within the set monitoring risk area does not exceed twice the distance of the longest path between any two points on the contour boundary corresponding to the initial dwell time, the target is judged to have landed.

[0076] 9) Information transmission: After the bird-repelling mechanism is triggered, the wide-angle camera sends the position of the landed target in a unified visual reference to the control unit, and the control unit drives the zoom camera and the laser repeller to further move and perform the repeller action.

[0077] When the control unit receives the position of the landed target in a unified visual reference, it obtains the deviation between the current position of the target and the current pointing direction of the zoom camera based on the calibration relationship between the zoom camera and the wide-angle camera, and adjusts the servo motor of the two-axis gimbal according to the deviation so that the zoom camera and the laser bird repellent point to the current position of the target. The zoom camera confirms the harmful bird for the landed target.

[0078] Among them, the zoom camera confirms the harmful bird for the landed target, that is, the zoom camera obtains continuous images of the landed target, performs differential change detection on the continuous images again, and normalizes the results of the differential change. When the change range of the landed target exceeds the set threshold, the zoom camera determines that there is an active target in the field of view as a harmful bird, wherein the set threshold of the change range of the landed target is set according to the number of pixels occupied by the farthest typical harmful bird in the field of view; the zoom camera sends a drive-away instruction to the laser bird repellent; after the laser bird repellent is started, the laser spot is enlarged and shot in the current pointing direction of the zoom camera, and a virtual rectangle surrounding the target center of gravity is constructed with the target center of gravity as the center, and several trajectories passing through the center of gravity are fitted in the rectangle, and the endpoints of each trajectory are located on two different sides of the virtual rectangle, and the output light spot of the laser bird repellent moves back and forth along the endpoints of different trajectories. The laser beam with an enlarged light spot plays a similar effect to a virtual stick in the vision of birds. After the birds perceive the light spot, they will be frightened and driven away.

[0079] In addition, the present invention provides a method for using a laser bird repellent device, comprising the following steps:

[0080] S1: configure the above-mentioned laser bird-repelling device; arrange a wide-angle camera and a two-axis gimbal at different positions of the supporting platform, respectively; the built-in servo motor of the two-axis gimbal realizes the pitch and rotation movement of the active end of the two-axis gimbal; the active end of the two-axis gimbal is configured with a zoom camera and a laser bird-repelling device; the laser bird-repelling device is arranged in parallel with the optical path of the zoom camera, and the pointing direction is synchronized; a unified visual reference is established based on the wide-angle camera, that is, the corresponding relationship between the camera coordinate system of the zoom camera and the camera coordinate system of the wide-angle camera is calibrated, so that the wide-angle camera obtains an image or the image of the zoom camera or is converted to a corresponding position in the world coordinate system; configure a control unit on the supporting platform;

[0081] S2: Within the field of view of the wide-angle camera, an area at a certain height above the ground or water surface is selected as the monitoring risk area. The wide-angle camera and zoom camera ignore objects outside the monitoring risk area.

[0082] S3: When a target enters the monitoring risk area, the wide-angle camera detects the target's landing, including the following:

[0083] S31: The wide-angle camera pre-processes the acquired image;

[0084] S32: inter-frame difference processing, performing difference processing on two consecutive pre-processed frames respectively, calculating the grayscale values ​​of corresponding pixel points in the two frames of images for subtraction, calculating the absolute value of the grayscale value subtraction result, and generating a differential grayscale image using the absolute value of the grayscale value subtraction result;

[0085] S33: extracting changed pixels, performing change detection on the differential grayscale image, and extracting the changed area between the two frames of images;

[0086] S34: noise reduction processing, operating the changed area through morphological algorithm, counting the size of the changed area, setting the pixel area threshold of the changed area, the pixel area threshold of the changed area is set by the number of pixels occupied by the typical harmful bird farthest in the field of view; filtering out the changed area that is not within the pixel area threshold range. The pixel area threshold of the changed area mentioned here is obtained according to the value range of the pixel area of ​​the harmful bird species appearing in the current season in the wide-angle camera, in combination with the bird population and seasonal changes in aquaculture and fruit planting areas. Even in the same area, the bird species appearing in different seasons are not exactly the same.

[0087] S35: Target extraction, judging whether the remaining change area after the noise reduction process is within the set monitoring risk area: if the change area is within the set monitoring risk area, target extraction is performed on the change area, the center of gravity of the change area is calculated using the weighted centroid method, the target corresponding to the change area is represented by the center of gravity, and the attribute information of the target is stored in the target queue; if the change area is not within the set monitoring risk area, target extraction is not performed;

[0088] S36: target search, identifying and recording the corresponding relationship of each target in the continuous images, and constructing a trajectory queue based on each target;

[0089] S37: Track tracking: track each target's track queue separately to obtain the track data of each target: if the track of the target track is outside the set monitoring risk area, delete the target from the target queue; if the track of the target track is still within the set monitoring risk area and moves toward the ground, output the target track data moving toward the ground to the landing track queue;

[0090] S38: Landing detection, performing landing detection on the target in the landing trajectory queue, calculating the change of the adjacent position information of the trajectory to determine whether the target has landed, if it is determined that the target has not landed, further tracking the trajectory; if it is determined that the target has landed, triggering the bird repelling mechanism; and

[0091] S39: information sending, after the bird-repelling mechanism is triggered, the zoom camera sends the position of the landed target in the unified visual reference to the control unit;

[0092] S4: After receiving the position of the landed target in the unified visual reference, the control unit drives to change the current position of the zoom camera so that the zoom camera and the laser bird repellent point to the target position;

[0093] S5; the zoom camera performs a secondary confirmation on the landed target. When the change range of the landed target exceeds the set threshold, the zoom camera sends the content that the target is a harmful bird to the control unit. At this time, the control unit starts the laser bird repellent. The laser bird repellent has a built-in laser generator and a beam expander. The laser generator is used to generate a laser beam. The beam expander thickens the laser beam. The thickened laser beam is directed to the current pointing direction of the zoom camera, and a number of trajectories passing through the center of the target are constructed. The thickened laser beam sweeps along the trajectory and appears around the harmful bird, so that the bird has a visual illusion that the thickened laser beam is a physical stick. In order to avoid being hit by the stick, the bird will quickly flee, thereby achieving timely expulsion of the hunting harmful bird;

[0094] S6: The zoom camera continuously monitors the changes in the field of view and performs differential change detection on the continuously acquired images. If the field of view changes exceed the threshold, it means that there are still harmful birds in the field of view. At this time, the laser repeller is guided to point to the area with drastic changes in the field of view until the field of view change of the zoom camera is less than the threshold, indicating that there are no active targets in the field of view; when all harmful birds are driven away, the laser is turned off to complete the repeller operation.

[0095] 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, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A laser bird-repelling device, characterized in that: include: A support platform is fixedly arranged relative to the ground; A wide-angle camera, arranged on the supporting platform, is used to continuously image the set monitoring risk area, establish a unified visual reference, and realize the landing detection of the target; A two-axis gimbal, arranged on the support platform, and capable of performing horizontal and vertical rotational motion relative to the support platform; A zoom camera, arranged on the active end of the two-axis gimbal, is used to focus on the monitoring risk area where the landed harmful bird is located, so as to realize secondary identification of the target; A laser repeller is arranged at the active end of the two-axis gimbal, and is used to emit a laser beam to the area where the zoom camera secondary identifies the target as a harmful bird, wherein the laser repeller is arranged in parallel with the optical path of the zoom camera and the pointing direction is synchronized; The control unit is integrated on the supporting platform and is respectively connected to the wide-angle camera, the two-axis gimbal, the zoom camera and the laser drive away for communication. The control unit is used to receive the landing detection result of the target of the wide-angle camera and the secondary recognition result of the target of the zoom camera. The control unit also drives the two-axis gimbal or the laser drive away to operate.

2. A laser bird repellent device according to claim 1, characterized in that: The set monitoring risk area is an area at a certain height above the ground or water surface selected within the field of view of the wide-angle camera as the monitoring risk area.

3. A laser bird-repellent device according to claim 2, characterized in that: The wide-angle camera realizes the landing detection of the target, including the following contents: The wide-angle camera pre-processes the acquired images; Inter-frame difference processing: perform difference processing on two consecutive pre-processed frames, calculate the grayscale values ​​of the corresponding pixels in the two frames, subtract them, calculate the absolute value of the grayscale value subtraction result, and use the absolute value of the grayscale value subtraction result to generate a differential grayscale image; Change pixel extraction: perform change detection on the differential grayscale image and extract the change area between the two frames of images; Noise reduction processing: operate the change area through morphological algorithm, count the size of the change area, set the pixel area threshold of the change area, and set the pixel area threshold of the change area according to the number of pixels occupied by the farthest typical harmful bird in the field of view; filter out the change area that is not within the pixel area threshold range; Target extraction: determine whether the remaining changed area after noise reduction processing is within the set monitoring risk area. If the changed area is within the set monitoring risk area, target extraction is performed on the changed area. The center of gravity of the changed area is calculated using the weighted centroid method. The center of gravity represents the target corresponding to the changed area, and the target attribute information is stored in the target queue. If the change area is not within the set monitoring risk area, no target extraction will be performed; Target search, identifying and recording the corresponding relationship of each target in continuous images, and building a trajectory queue based on each target; Trajectory tracking: Track each target’s trajectory queue separately to obtain the trajectory data of each target: if the target’s trajectory is outside the set monitoring risk area, delete the target from the target queue; If the target track is still within the set monitoring risk area and moving toward the ground, the target track data moving toward the ground is output to the landing track queue; Landing detection: perform landing detection on the targets in the landing trajectory queue, calculate the changes in the adjacent position information of the trajectory to determine whether the target has landed, and further track the trajectory if it is determined that the target has not landed; If the target is judged to have landed, the bird-repelling mechanism will be triggered; as well as After the information is sent and the bird-repelling mechanism is triggered, the wide-angle camera sends the position of the landed target in a unified visual reference to the control unit, and the control unit drives the zoom camera and the laser repeller to further move and perform the repeller action.

4. A laser bird-repellent device according to claim 3, characterized in that: The wide-angle camera preprocesses the acquired image, specifically, statistics the overall brightness and variance of the image, and calculates the clarity of the image using a gradient difference method. If the clarity of the image is less than a set clarity threshold, the gain of the wide-angle camera is adjusted, and the image after gain adjustment is subjected to histogram equalization processing to make the grayscale probability distribution of the image after histogram equalization processing uniform, thereby obtaining a preprocessed image; if the clarity of the image is not less than the set clarity threshold, the image is not preprocessed.

5. The laser bird-repelling device according to claim 4, characterized in that: The target search identifies and records the corresponding relationship of each target in the continuous images, and constructs a trajectory queue based on each target. Specifically, a three-dimensional target search based on Manhattan distance is adopted. On the basis of statistically analyzing the two-dimensional Manhattan distance between the current target and another target in the image at a certain moment, the relationship between the current target and the target in the image at the previous moment is analyzed to find the continuous corresponding relationship of the current target in the time dimension; let the current target be k, and the other target be j, is the two-dimensional Manhattan distance between the current target k and the target j at the i-th moment; is the two-dimensional Manhattan distance between the current target k at the i-th moment and the target j at the i-1-th moment; min(·) is the minimum value of the two-dimensional Manhattan distance between the current target k and another target in the image at the i-th moment; Search the target queue at the i-1th moment. When the current target k at the i-th moment and the target j at the i-1th moment satisfy the formula , the information of the current target k and target j is saved in the trajectory queue, and the information of the current target k and target j in the continuous images is recorded.

6. The laser bird-repelling device according to claim 5, characterized in that: The landing detection performs landing detection on the targets in the landing trajectory queue, calculates the changes in adjacent position information of the trajectory to determine whether the target has landed. Specifically, within a given time range, if the position change of the center of gravity of the target in the height direction within the set monitoring risk area does not exceed twice the distance of the longest path between any two points on the contour boundary corresponding to the initial stop time, the target is determined to have landed.

7. The laser bird-repelling device according to claim 3, characterized in that: After the control unit receives the position of the landed target in the unified visual reference, it obtains the deviation between the current position of the target and the current pointing direction of the zoom camera based on the calibration relationship between the zoom camera and the wide-angle camera, and adjusts the servo motor of the two-axis gimbal according to the deviation, so that the zoom camera and the laser bird repellent point to the current position of the target, and the zoom camera confirms the harmful bird for the landed target.

8. The laser bird-repelling device according to claim 7, characterized in that: The zoom camera confirms the harmful bird for the landed target, that is, the zoom camera obtains continuous images of the landed target, performs differential change detection on the continuous images again, and normalizes the results of the differential change. When the change range of the landed target exceeds the set threshold, the zoom camera determines that there is an active target in the field of view as a harmful bird, wherein the set threshold of the change range of the landed target is set according to the number of pixels occupied by the farthest typical harmful bird in the field of view; the zoom camera sends a repelling instruction to the laser bird repellent; after the laser bird repellent is started, the laser spot is enlarged and shot in the current pointing direction of the zoom camera, and a virtual rectangle surrounding the target center of gravity is constructed with the target center of gravity as the center, and a number of trajectories passing through the center of gravity are fitted in the rectangle, and the endpoints of each trajectory are located on two different sides of the virtual rectangle, and the output spot of the laser bird repellent moves back and forth along the endpoints of different trajectories.

9. A method for using a laser bird repellent device, characterized in that: The steps include: S1: configure the laser bird-repelling device as described in any one of claims 1 to 8; arrange a wide-angle camera and a two-axis gimbal at different positions of the supporting platform, respectively; the built-in servo motor of the two-axis gimbal realizes the pitch and rotation movement of the active end of the two-axis gimbal; the active end of the two-axis gimbal is configured with a zoom camera and a laser repeller; the optical path of the laser repeller is parallel to that of the zoom camera, and the pointing direction is synchronized; a unified visual reference is established based on the wide-angle camera, that is, the corresponding relationship between the camera coordinate system of the zoom camera and the camera coordinate system of the wide-angle camera is calibrated, so that the wide-angle camera obtains an image or the image of the zoom camera or is converted to a corresponding position in the world coordinate system; configure a control unit on the supporting platform; S2: Within the field of view of the wide-angle camera, an area at a certain height above the ground or water surface is selected as the monitoring risk area. The wide-angle camera and zoom camera ignore objects outside the monitoring risk area. S3: When a target enters the monitoring risk area, the wide-angle camera detects the target's landing, including the following: S31: The wide-angle camera pre-processes the acquired image; S32: inter-frame difference processing, performing difference processing on two consecutive pre-processed frames respectively, calculating the grayscale values ​​of corresponding pixel points in the two frames of images for subtraction, calculating the absolute value of the grayscale value subtraction result, and generating a differential grayscale image using the absolute value of the grayscale value subtraction result; S33: extracting changed pixels, performing change detection on the differential grayscale image, and extracting the changed area between the two frames of images; S34: noise reduction processing, operating the changed area through a morphological algorithm, counting the size of the changed area, setting a pixel area threshold of the changed area, and filtering out the changed area that is not within the pixel area threshold range; S35: Target extraction, judging whether the remaining changed area after the noise reduction process is located in the set monitoring risk area: if the changed area is located in the set monitoring risk area, then target extraction is performed on the changed area, the center of gravity of the changed area is calculated using the weighted centroid method, the center of gravity is used to represent the target corresponding to the changed area, and the attribute information of the target is stored in the target queue; If the change area is not within the set monitoring risk area, no target extraction will be performed; S36: target search, identifying and recording the corresponding relationship of each target in the continuous images, and constructing a trajectory queue based on each target; S37: Track tracking: track each target in the track queue separately to obtain the track data of each target: if the track of the target is outside the set monitoring risk area, delete the target in the target queue; If the target track is still within the set monitoring risk area and moving toward the ground, the target track data moving toward the ground is output to the landing track queue; S38: Landing detection, performing landing detection on the target in the landing trajectory queue, calculating the change of the adjacent position information of the trajectory to determine whether the target has landed, if it is determined that the target has not landed, further tracking the trajectory; if it is determined that the target has landed, triggering the bird repelling mechanism; and S39: information sending, after the bird-repelling mechanism is triggered, the zoom camera sends the position of the landed target in the unified visual reference to the control unit; S4: After receiving the position of the landed target in the unified visual reference, the control unit drives to change the current position of the zoom camera so that the zoom camera and the laser bird repellent point to the target position; S5; the zoom camera performs a secondary confirmation on the landed target. When the change range of the landed target exceeds the set threshold, the zoom camera sends the content that the target is a harmful bird to the control unit. At this time, the control unit starts the laser bird repellent. The laser bird repellent has a built-in laser generator and a beam expander. The laser generator is used to generate a laser beam. The beam expander thickens the laser beam. The thickened laser beam is directed to the current pointing direction of the zoom camera, and a number of trajectories passing through the center of the target are constructed. The thickened laser beam sweeps along the trajectory and appears around the harmful bird, so that the bird has a visual illusion that the thickened laser beam is a physical stick. In order to avoid being hit by the stick, the bird will quickly flee, thereby achieving timely expulsion of the hunting harmful bird; S6: The zoom camera continuously monitors the changes in the field of view and performs differential change detection on the continuously acquired images. If the field of view changes exceed the threshold, it means that there are still harmful birds in the field of view. At this time, the laser repeller is guided to point to the area with drastic changes in the field of view until the field of view change of the zoom camera is less than the threshold, indicating that there are no active targets in the field of view; when all harmful birds are driven away, the laser is turned off to complete the repeller operation.

10. The method for using the laser bird repellent device according to claim 9, characterized in that: The pixel area threshold of the setting change area described in step S34 is obtained based on the value range of the pixel area of ​​the harmful bird species appearing in the current season in the wide-angle camera in combination with the bird population and seasonal changes in the aquaculture and fruit planting areas.

Images