An automated waterbird biodiversity monitoring and collection device

Abstract

The application relates to an automatic water bird biodiversity monitoring and collecting device; the device comprises an adaptive adjusting frame, a bionic attraction system and a perception layer device group, the adaptive adjusting frame is installed in a diving area and can be adaptively adjusted according to air volume and air direction; the bionic attraction system is installed on the adaptive adjusting frame, is internally integrated with acoustics, optics and pheromone induction, can induce water birds to approach and collect required information after the water birds approach according to actual conditions; the perception layer device group is installed on the adaptive adjusting frame, is used for monitoring water birds, collecting images and sounds of the water birds and monitoring water areas where the water birds are located and collecting water area information that influences the water birds. The adaptive adjusting frame is used for realizing environmental adaptability adjustment, multi-modal perception and edge computing technology are combined, the problems of low data collection precision and difficult multi-source information integration of a traditional monitoring method in a complex environment are effectively solved, and the monitoring precision can be improved, environmental adaptability is enhanced and multi-source data integration is realized.

Description

Technical Field

[0001] This application relates to the field of wetland ecological monitoring technology, and in particular to an automated device for monitoring and collecting biodiversity data of waterbirds. Background Technology

[0002] Monitoring waterbird biodiversity has become a core task of ecological conservation. Traditional monitoring methods mainly rely on manual patrols and fixed infrared sensor camera networks, which have revealed significant technical shortcomings in practical applications. In terms of perception, existing infrared camera technology struggles to cope with highly reflective water surfaces and nighttime imaging conditions, resulting in low effective recognition rates; acoustic monitoring equipment is susceptible to environmental noise interference and has insufficient ability to capture the calls of small waterbirds. Regarding multi-source data integration, existing systems operate with each sensor unit operating independently, leading to a disconnect between hydrological monitoring data and bird behavior information, making it difficult to promptly identify correlations in ecological events. In terms of data processing, traditional solutions require the complete transmission of raw monitoring data back to the cloud, imposing a significant communication cost burden in remote wetland areas.

[0003] Recent attempts to improve related technologies still have significant limitations: while multispectral imaging technology has improved adaptability to backlit scenes, it has failed to effectively solve the problem of polarized light interference from the water surface; voiceprint recognition models directly apply urban environmental training data, resulting in unsatisfactory accuracy in recognizing the unique low-frequency sounds of wetlands; and fixed-threshold early warning mechanisms lack adaptability to environmental variables, leading to a persistently high false alarm rate under complex weather conditions. These technological bottlenecks severely restrict the timeliness and accuracy of wetland ecological protection.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] The purpose of this application is to provide an automated waterbird biodiversity monitoring and data collection device and method, which has the advantages of improving monitoring accuracy, enhancing environmental adaptability, realizing multi-source data integration and reducing communication costs.

[0006] This application provides an automated waterbird biodiversity monitoring and data collection device, the technical solution of which is as follows: It includes: an adaptive adjustment frame, installed in the diving area, capable of adaptive adjustment according to wind volume and direction; a biomimetic attraction system, installed on the adaptive adjustment frame, integrating acoustic, optical, and pheromone induction, capable of inducing waterbirds to approach and collecting the required information according to actual conditions; a sensing layer device group, installed on the adaptive adjustment frame, used to monitor waterbirds and collect their images and sounds, while simultaneously monitoring the surrounding water area and collecting information on water areas affecting the waterbirds; an edge computing hub, connected to the sensing layer device group, used to process and transmit the collected image and sound information; and a remote monitoring platform, connected to the edge computing hub, used to receive data from the edge computing hub and perform three-dimensional reconstruction of species distribution.

[0007] Furthermore, this application proposes that the adaptive adjustment frame includes: a waterproof enclosure, with an edge computing hub installed inside; positioning cylinders installed on both sides of the waterproof enclosure, each equipped with anchors for insertion into the soil to anchor the entire device; dynamic adjustment cylinders, with at least three sets arranged in an isosceles triangle on the top of the waterproof enclosure; a rotating base installed on top of the dynamic adjustment cylinders; a gyroscope installed at the center of the bottom of the rotating base; and an airflow and wind speed sensor installed at the bottom of the rotating base to monitor airflow and direction. The sensor processes the collected information, matches it with the information collected by the gyroscope, and controls the dynamic adjustment cylinders and the rotating base to adjust in real time, enabling the sensing layer device group to perform real-time adjustments during image acquisition.

[0008] Furthermore, this application also proposes that the biomimetic attraction system includes a waterproof housing and a flexible solar panel covering the waterproof housing.

[0009] Furthermore, this application also proposes that the bionic attraction system includes an optical induction module, which includes a green LED array, an infrared LED array, and a UV laser. The green LED array, infrared LED array, and UV laser are all mounted on a waterproof housing, and the green LED array, infrared LED array, and UV laser are all connected to a constant current drive circuit.

[0010] Furthermore, this application also proposes that the biomimetic attraction system further includes an acoustic induction module, which comprises: a microcontroller as a central hub; a voiceprint classification indexer connected to the microcontroller for collecting sound information and performing a retrieval; a solid-state storage chip connected to the microcontroller for extracting the corresponding audio file from the solid-state storage chip after the voiceprint classification indexer completes the retrieval; a DSP processor connected to a digital noise reduction chip for real-time elimination of environmental noise and outputting an analog signal to a dynamic gain controller; and a speaker connected to the dynamic gain controller for final audio output.

[0011] Furthermore, this application also proposes that the pheromone induction module includes: a storage tank for storing pheromone solution; a micro-flow metering pump, the inlet of which is connected to the storage tank via a pipe and a distribution valve, and the outlet of which is connected to an atomizing nozzle via a pipe; and a fan, the storage tank being installed on the atomizing nozzle for blowing out the atomized pheromone.

[0012] Furthermore, this application also proposes that the sensing layer device group includes: an infrared thermal imager, mounted on a rotating base, for sensing the infrared radiation of waterbirds; a polarized high-definition camera, mounted on the rotating base, for capturing images of waterbirds after the infrared thermal imager captures their infrared radiation information; an acoustic acquisition microphone, connected to a microcontroller, for inputting the acquired sound information into a voiceprint classification indexer for retrieval; and a hydrological sensor group, mounted on the bottom of an adaptive adjustment frame, for monitoring the surrounding water area.

[0013] Furthermore, this application also proposes that the hydrological sensor group includes: a water level gauge, which is installed at the bottom of the adaptive adjustment frame to measure the water depth in real time to determine the extent of wetland inundation and to trigger early warning monitoring of waterbird foraging behavior when the water level continues to drop; and a dissolved oxygen sensor, which is used to detect the dissolved oxygen concentration in the water area, to warn of the risk of fish mortality, and to trigger early warning monitoring of waterbird foraging behavior.

[0014] Furthermore, this application also proposes that the edge computing hub includes: a biometric identification processor, which is connected to an acoustic acquisition microphone and a polarized high-definition camera, for data processing of the sound and image information acquired by the acoustic acquisition microphone and the polarized high-definition camera; and a wireless data transmission module, which is connected to a remote monitoring platform and equipped with 4G / 5G / LoRa transmission protocols.

[0015] Furthermore, this application also proposes that the remote monitoring platform includes: a 3D reconstruction engine that integrates GPS location data to generate a waterbird distribution heat map; and an early warning and dispatch module that connects to the hydrological sensor group and automatically triggers a drone patrol command when a rare species is identified.

[0016] As can be seen from the above, the automated waterbird biodiversity monitoring and data collection device and method provided in this application achieves environmental adaptability adjustment through adaptive frame adjustment. Combined with multimodal sensing and edge computing technology, it effectively solves the problems of low data collection accuracy and difficulty in integrating multi-source information in complex environments by traditional monitoring methods. It has the advantages of improving monitoring accuracy, enhancing environmental adaptability, realizing multi-source data integration and reducing communication costs. Attached Figure Description

[0017] The above and other objects, features and advantages of this application will become more apparent from the more detailed description of exemplary embodiments thereof in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments thereof.

[0018] Figure 1 This is a schematic diagram of the overall structure shown in the embodiments of this application;

[0019] Figure 2 This is a schematic diagram illustrating the locations of the various institutions as shown in the embodiments of this application;

[0020] Figure 3 This is a schematic diagram of the internal structure of the biomimetic attraction system shown in the embodiments of this application;

[0021] Figure 4 This is a schematic diagram of the sensing layer device group and edge computing hub shown in the embodiments of this application;

[0022] Figure 5 This is a schematic diagram of the acoustic induction module shown in an embodiment of this application;

[0023] Figure 6 This is a schematic diagram of a biological species identification processor shown in an embodiment of this application;

[0024] Figure 7 This is a schematic diagram of the connection of the hydrological sensor group shown in an embodiment of this application;

[0025] Figure 8 This is a schematic diagram of the hydrological sensor group structure shown in an embodiment of this application;

[0026] Figure 9 This is a schematic diagram of a remote monitoring platform shown in an embodiment of this application.

[0027] Figure label:

[0028] 1-Adaptive adjustment frame, 11-Dynamic adjustment cylinder, 12-Positioning cylinder, 13-Anchor nail, 14-Air volume and wind speed sensor, 15-Gyroscope, 16-Waterproof chassis, 17-Rotating base;

[0029] 2-Remote monitoring platform;

[0030] 3-Bionic attraction system, 31-Waterproof housing, 32-Flexible solar panel, 33-Optical induction module, 331-Green LED array, 332-Infrared LED array, 333-UV ultraviolet laser, 34-Acoustic induction module, 341-Speaker, 342-Microcontroller, 343-Solid-state storage chip, 344-Voiceprint classification indexer, 345-DSP processor, 346-Digital noise reduction chip, 347-Dynamic gain controller, 35-Pheromone induction module, 351-Microflow metering pump, 352-Liquid storage tank, 353-Dispensing valve, 354-Atomizing nozzle, 355-Fan;

[0031] 4-Sensing layer equipment group, 41-Infrared thermal imager, 42-Polarized high-definition camera, 43-Acoustic acquisition microphone, 44-Hydrological sensor group, 441-Water level gauge, 442-Dissolved oxygen sensor;

[0032] 5-Edge computing hub, 51-Biological species identification processor, 52-Wireless data transmission module. Detailed Implementation

[0033] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following drawings indicate similar items; therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0034] In existing technologies, waterbird biodiversity monitoring has long relied on manual patrols and fixed sensor networks. Traditional infrared cameras face interference from water surface reflections in wetland environments, resulting in the loss of image features; acoustic monitoring equipment is affected by environmental noise, making it difficult to accurately identify target species; and hydrological sensors operate independently, failing to form an effective correlation with biological behavior data. Monitoring systems generally suffer from problems such as limited sensing dimensions, fragmented multi-source data, and wasted computing resources, making it difficult to meet the real-time requirements of wetland ecological protection.

[0035] To address the aforementioned challenges, the research team conducted a systematic study of the complexities of wetland environments. First, they analyzed the correlation between waterbird behavior and environmental parameters, discovering that traditional single sensors could not capture multidimensional ecological information. By studying the phototaxis, acoustic response characteristics, and chemical sensing mechanisms of waterbirds, a multimodal induction and data fusion scheme was proposed. To overcome data transmission bottlenecks, an edge computing architecture was introduced to optimize the data processing flow. Ultimately, a holistic solution was developed, encompassing environmental adaptive regulation, collaborative acquisition of multi-source data, and localized intelligent processing.

[0036] Therefore, this application proposes an automated waterbird biodiversity monitoring and collection device, including an adaptive adjustable frame 1 installed in a diving area, the frame being equipped with an environmental perception module to achieve dynamic stability; a biomimetic attraction system 3 integrating acoustic, optical, and pheromone-inducing functions to attract target species; a group of sensing layer devices 4 with image, sound, and water information collection capabilities; an edge computing hub 5 for processing and transmitting data; and a remote monitoring platform 2 for performing three-dimensional reconstruction.

[0037] Among them, the adaptive adjustment frame 1 refers to a support structure with environmental responsiveness, specifically achieved through a combination of wind volume and speed sensors 14 and dynamic adjustment cylinders 11. It adjusts the equipment posture by monitoring meteorological parameters in real time to ensure data acquisition stability. The biomimetic attraction system 3 refers to a composite induction device that simulates natural ecological signals, specifically achieved through a combination of a multispectral light source array, a voiceprint database, and a mist pheromone release device, implementing precise attraction based on the biological characteristics of different species. The sensing layer device group 4 refers to a multimodal data acquisition unit, specifically achieved through the collaborative work of an infrared thermal imager 41, a polarized light camera, and acoustic sensors, simultaneously acquiring biological behavior and aquatic environmental data. The edge computing hub 5 refers to a localized data processing unit, specifically achieved through a combination of biometric recognition algorithms and wireless transmission modules, completing data preprocessing at the device end. The remote monitoring platform 2 refers to a spatial visualization analysis system, specifically achieved through the fusion of a 3D modeling engine and hydrological correlation algorithms, dynamically presenting species distribution patterns.

[0038] Specifically, when the device is deployed in a wetland environment, the adaptive adjustment rack 1 adjusts the height and angle of the equipment according to real-time wind speed to eliminate the influence of water surface reflection on imaging. The biomimetic attraction system 3 attracts target waterbirds into the monitoring area by playing specific frequency sound waves, emitting ultraviolet spectra, and releasing pheromones. The sensing layer device group 4 simultaneously initiates infrared thermal imaging tracking, polarized high-definition shooting, and water parameter acquisition to obtain multi-dimensional ecological data. The edge computing hub 5 performs noise reduction and feature extraction on the raw data, transmitting only valid information to the remote platform. The remote monitoring platform 2 integrates spatiotemporal data to generate a three-dimensional heat map, providing visualization support for ecological protection decision-making.

[0039] Compared to existing technologies, traditional monitoring equipment uses fixed sensors operating independently, which cannot eliminate the impact of environmental interference on data quality and lacks multi-source data fusion capabilities. This solution ensures data acquisition stability through an environmental adaptive adjustment mechanism, improves the target species capture rate through multimodal induction technology, and optimizes data transmission efficiency through an edge computing architecture, forming a complete chain solution from data acquisition to decision support.

[0040] Through the above technical solutions, this application effectively solves the problem of optical imaging distortion in wetland environments, improves the accuracy of species identification in complex acoustic environments, and enables dynamic correlation analysis between biological behavior and water parameters. Localized data processing reduces network transmission load, making the monitoring system operable in remote wetlands. The spatiotemporal fusion of multi-dimensional data provides a more complete ecological information foundation for biodiversity research, supporting the precise implementation of conservation measures.

[0041] This application further proposes an edge computing hub 5 installed inside a waterproof enclosure 16; positioning cylinders 12 installed on both sides of the waterproof enclosure 16, with anchor nails 13 for insertion into the soil to anchor the entire device; at least three sets of dynamic adjustment cylinders 11 arranged in an isosceles triangle on the top of the waterproof enclosure 16; a rotating base 17 installed on top of the dynamic adjustment cylinders 11; a gyroscope 15 installed at the center of the bottom of the rotating base 17; and an airflow and wind speed sensor 14 installed at the bottom of the rotating base 17 for monitoring airflow and wind direction. After processing the collected information, the sensor matches the information collected by the gyroscope 15 to control the dynamic adjustment cylinders 11 and the rotating base 17 to make real-time adjustments, so that the sensing layer device group 4 can perform real-time adjustments during image acquisition.

[0042] Among them, the waterproof enclosure 16 refers to a protective shell with a sealed structure, which can be made of aluminum alloy with silicone sealing rings, used to protect the internal electronic equipment from moisture corrosion. The positioning cylinder 12 refers to a pneumatic actuator with a telescopic rod, which can be implemented using a double-acting cylinder with stainless steel anchors 13, using hydraulically driven anchors 13 to insert into the soil to fix the device. The dynamic adjustment cylinder 11 refers to a linear actuator that can precisely control the stroke, which can be implemented using a servo electric cylinder with a ball screw, forming a stable three-point support structure with an isosceles triangle layout. The rotating base 17 refers to a load-bearing platform that can rotate around a vertical axis, which can be implemented using crossed roller bearings with a stepper motor, used to support the sensing device and adjust the horizontal angle. The gyroscope 15 refers to an inertial sensor that measures angular velocity, which can be implemented using a MEMS three-axis gyroscope 15, used to monitor the device's tilt angle in real time. The airflow and wind speed sensor 14 refers to a device that measures airflow parameters, which can be implemented using a hot-wire anemometer with a wind vane, used to collect environmental wind force data.

[0043] Specifically, anchor bolts 13 are hydraulically driven by positioning cylinders 12 to insert into the soft soil layer of the wetland, forming a multi-point anchoring foundation. Three sets of dynamic adjustment cylinders 11 form an isosceles triangular support surface on the top of the waterproof housing 16, and the tilt angle of the platform is changed by independently adjusting the extension and retraction of each cylinder. The rotating base 17 rotates around the vertical axis under the drive of a stepper motor, and forms horizontal adjustment capability in conjunction with the real-time attitude data fed back by the gyroscope 15. The wind speed and volume sensor 14 continuously monitors the environmental wind parameters, and combined with the tilt angle data collected by the gyroscope 15, calculates the extension and retraction compensation value of each cylinder through a closed-loop control algorithm, drives the dynamic adjustment cylinders 11 to adjust the height, and controls the rotation of the rotating base 17 to counteract the impact of wind on the stability of the equipment. This forms a deep coupling between the mechanical structure and the control system, maintaining a stable observation angle of the sensing equipment under complex weather conditions.

[0044] Compared to existing technologies, traditional wetland monitoring devices mostly employ fixed support structures, which cannot cope with equipment swaying caused by wind changes. For example, the existing technology with publication number CN117173631 A uses a rigid tripod to fix the infrared camera, and the image blur rate rises to over 60% when the wind speed exceeds level 5. In contrast, this solution, through the synergistic action of the dynamically adjusting cylinder 11 and the rotating base 17, can adjust the device's posture in real time when the wind speed fluctuates, enabling the polarized light camera to maintain a stable focus even under level 6 wind conditions.

[0045] Through the above technical solution, this application achieves adaptive attitude adjustment of the monitoring device in wetland environments, effectively overcoming the problem of equipment swaying caused by wind and water flow impact. The combination of the anchoring structure and the dynamic adjustment mechanism ensures that the device remains stable and fixed in soft soil, and the closed-loop control system ensures that the sensing device is always at the optimal observation angle, thereby improving the data acquisition quality under complex meteorological conditions.

[0046] This application further proposes a biomimetic attraction system 3 including a waterproof housing 31 and a flexible solar panel 32 covering the waterproof housing 31, as well as an optical induction module 33. The optical induction module 33 includes a green LED array 331, an infrared LED array 332, and a UV laser 333. The green LED array 331, the infrared LED array 332, and the UV laser 333 are all mounted on the waterproof housing 31, and the green LED array 331, the infrared LED array 332, and the UV laser 333 are all connected to a constant current drive circuit.

[0047] Among them, the waterproof casing 31 refers to a sealed outer shell, which can be injection molded from polycarbonate material to prevent moisture from corroding the internal optical components. The flexible solar panel 32 refers to a bendable photovoltaic module, which can be implemented using amorphous silicon thin-film batteries and is attached to the casing surface to provide continuous power to the system. The green LED array 331 refers to a matrix composed of multiple green light-emitting diodes, which can be implemented using LED chips with wavelengths of 520-560nm, covering the wavelength range sensitive to the retina of waterbirds. The infrared LED array 332 refers to a group of diodes emitting invisible infrared light, which can be implemented using LED modules with a wavelength of 850nm for interference-free nighttime illumination. The UV laser 333 refers to a device that generates ultraviolet light beams, which can be implemented using a semiconductor laser with a wavelength of 405nm to stimulate the ultraviolet visual response of specific migratory species. The constant current drive circuit refers to a power supply module that stabilizes the output current, which can be implemented using an LM317 voltage regulator chip in conjunction with a PWM dimming circuit to ensure stable brightness and flicker-free operation of each light source.

[0048] Specifically, the waterproof casing 31 protects the internal optical components from the high humidity of the wetland environment through a sealed structure, while the flexible solar panel 32 is integrated into the casing surface to continuously convert light energy into electrical energy. The green LED array 331 emits light in the 500-570nm wavelength range during daytime operation, matching the absorption peak of the photosensitive pigment in the retina of waterbirds. The infrared LED array 332 activates at night, emitting invisible infrared light to avoid disturbing waterbirds and working in conjunction with thermal imaging equipment for monitoring. The UV laser 333 emits specific wavelengths of ultraviolet light to create visual stimulation for species with ultraviolet visual perception capabilities. These three light sources are independently controlled by a constant current drive circuit and combined according to ambient lighting conditions and the activity patterns of the target species to achieve multispectral synergistic induction.

[0049] Compared to existing technologies, traditional optical guidance devices mostly use a single white light or monochromatic LED light source, which cannot cover the diverse visual perception range of waterbirds. For example, while conventional infrared lighting is suitable for nighttime use, the lack of an ultraviolet band makes it unsuitable for attracting migratory species that rely on ultraviolet light for navigation. This solution, however, uses a combination of green light, infrared, and ultraviolet bands to simultaneously meet the visual sensitivity requirements of different time periods and species, overcoming the spectral limitations of a single light source.

[0050] Through the above technical solution, this application realizes a multi-band composite optical induction mechanism, effectively covering the visually sensitive range of waterbirds. The green light band precisely matches the photosensitive characteristics of the retina, the infrared band provides interference-free nighttime illumination, and the ultraviolet band stimulates the navigation instincts of migratory species. The three-band light source is stably output through a constant current circuit, avoiding flicker interference and significantly improving the attraction efficiency for diverse waterbird populations under different environmental conditions.

[0051] This application further proposes that the biomimetic attraction system 3 also includes an acoustic induction module 34. The acoustic induction module 34 includes a microcontroller 342 as a central hub, a voiceprint classification indexer 344 connected to the microcontroller 342 for collecting sound information and performing retrieval, a solid-state storage chip 343 connected to the microcontroller 342 for extracting the corresponding audio file after the voiceprint classification indexer 344 completes the retrieval, a DSP processor 345 connected to a digital noise reduction chip 346 for real-time elimination of environmental noise and output of analog signals to a dynamic gain controller 347, and a speaker 341 connected to the dynamic gain controller 347 for final audio output.

[0052] Among them, the microcontroller 342 refers to the embedded control unit, which can be implemented using the STM32H743 series chip, and is used for timing control of the voiceprint retrieval and audio output process. The voiceprint classification indexer 344 refers to the feature matching module based on the KNN algorithm, which can be implemented using a Raspberry Pi CM4 computing module, and is used to sort real-time voiceprints and pre-stored samples based on similarity. The solid-state storage chip 343 refers to the non-volatile storage medium, which can be implemented using a KIOXIA BG5 series SSD, and is used to store a standard voiceprint database for different waterbird species. The digital noise reduction chip 346 refers to the active noise cancellation integrated circuit, which can be implemented using the ADAU1777 chip from Analog Devices, and cancels environmental noise by generating inverted sound waves. The dynamic gain controller 347 refers to the adaptive volume adjustment circuit, which can be implemented using the LMH6401 programmable gain amplifier from Texas Instruments, and adjusts the output intensity according to the ambient sound pressure level.

[0053] Specifically, after the acoustic acquisition device captures ambient sound waves, the raw audio signal is transmitted to a digital noise reduction chip 346 for preliminary filtering. The noise-reduced signal is split into two paths: one path is input to a DSP processor 345 for voiceprint feature extraction, and the other path is temporarily stored in a buffer. The voiceprint feature vector extracted by the DSP processor 345 is sent to a voiceprint classification indexer 344 for similarity matching with a pre-stored bird voiceprint database in a solid-state storage chip 343. When the matching degree exceeds a set threshold, the indexer sends a command to the microcontroller 342 to trigger the retrieval of the standard call audio file of the corresponding species. The retrieved audio file undergoes secondary noise reduction processing by the DSP processor 345 to eliminate background noise introduced during storage. The processed signal is input to a dynamic gain controller 347, which automatically adjusts the output power according to the real-time monitored ambient noise intensity, and is finally played directionally through a waterproof speaker 341.

[0054] Compared to existing technologies, traditional acoustic monitoring equipment uses only a single noise reduction module and lacks an active voiceprint matching mechanism. For example, patent CN117727333 B only uses a general-purpose filter, which cannot effectively distinguish between the frequency bands of waterbird calls and wind noise. This solution constructs a two-stage processing architecture of digital noise reduction and analog gain, which preserves the high-frequency components of the target voiceprint while suppressing low-frequency environmental interference. Combined with the fast retrieval function of the pre-stored voiceprint database, it achieves accurate reproduction of the acoustic characteristics of specific species.

[0055] Through the above technical solutions, this application effectively solves the technical problem of low target voiceprint recognition rate in complex sound field environments of wetlands. The digital noise reduction chip 346 directionally eliminates low-frequency wind noise below 200Hz, the voiceprint classification indexer 344 achieves millisecond-level feature matching, and the dynamic gain control ensures that the induced sound wave maintains a sound pressure level of more than 85dB within an effective propagation distance of 30 meters. The overall system improves the voiceprint recognition accuracy of small and medium-sized water birds to a practical level.

[0056] This application further proposes a pheromone induction module 35, including a storage tank 352, a micro-flow metering pump 351, a distribution valve 353, an atomizing nozzle 354, and a blower 355. The storage tank 352 stores the pheromone solution. The inlet of the micro-flow metering pump 351 is connected to the storage tank 352 through a pipe and the distribution valve 353, and the outlet is connected to the atomizing nozzle 354 through a pipe. The blower 355 is installed on the atomizing nozzle 354 to blow out the atomized pheromone.

[0057] The storage tank 352 is a sealed container for storing liquid pheromones, which can be made with a polyethylene inner liner and a stainless steel outer shell. Its double-layer structure blocks ultraviolet light and maintains the chemical stability of the solution. The micro-flow metering pump 351 is a device for precisely controlling the liquid delivery flow rate, which can be implemented using a stepper motor to drive a ceramic piston. The pumping volume per cycle is controlled within the range of 0.1-5 ml by adjusting the stepper angular displacement. The distribution valve 353 is a multi-channel fluid switching device, which can be implemented using a three-way solenoid valve to select the pheromone type from different storage tanks 352. The atomizing nozzle 354 is an atomizing device that converts liquid into micron-sized aerosols, which can be implemented using a piezoelectric ceramic high-frequency vibrating plate, breaking the liquid into particles with a diameter of 10-50 microns. The fan 355 is a power device that generates directional airflow, which can be implemented using a brushless DC motor to drive a centrifugal impeller, with an adjustable wind speed range of 0.5-5 m / s.

[0058] Specifically, the pheromone solution stored in the storage tank 352 is switched to the storage unit corresponding to the target species via the distribution valve 353. The micro-flow metering pump 351 extracts a quantitative amount of solution according to a preset program and delivers it to the atomizing nozzle 354. After the atomizing nozzle 354 converts the liquid pheromone into aerosol particles, the airflow generated by the fan 355 directionally diffuses the atomized particles to the target area. The pheromone release rate is controlled by adjusting the pulse frequency of the micro-flow metering pump 351, and the diffusion range is controlled within a 5-20 meter range by adjusting the wind speed of the fan 355. For different waterbird species, for example, trans-2-hexenal is used to simulate the odor of aquatic plants for geese and ducks, and geosmin is used to simulate the metabolites of bottom sediment microorganisms for plovers and sandpipers, achieving species-specific induction.

[0059] Compared to existing technologies, traditional waterbird monitoring devices release pheromones using open evaporation discs or manual spraying, which suffers from drawbacks such as uncontrollable release dosage, random diffusion range, and inability to switch pheromone types. This solution achieves on-demand switching of pheromone types and precise adjustment of release volume through the coordinated control of a micro-flow metering pump 351 and a distribution valve 353; the combined application of atomizing nozzles 354 and a fan 355 increases pheromone diffusion efficiency to more than three times that of traditional methods; and the double-layered storage tank 352 ensures that the pheromone remains effective for at least six months in the field.

[0060] Through the above technical solution, this application solves the problem that traditional devices cannot accurately regulate pheromone release according to environmental conditions and target species, achieving targeted induction of specific waterbird species and significantly improving the target capture rate of monitoring equipment in complex wetland environments. Quantitative release control avoids ecological disturbances caused by excessive pheromone release, while adapting to operational needs under different seasons and climatic conditions.

[0061] This application further proposes a sensing layer device group 4 including an infrared thermal imager 41, a polarized high-definition camera 42, an acoustic acquisition microphone 43, and a hydrological sensor group 44. The infrared thermal imager 41 is mounted on a rotating base 17, the polarized high-definition camera 42 is mounted on a rotating base 17, the acoustic acquisition microphone 43 is connected to a microcontroller 342, and the hydrological sensor group 44 is mounted on the bottom of an adaptive adjustment frame 1.

[0062] Among them, the infrared thermal imager 41 refers to a device that identifies targets by detecting the infrared radiation emitted by living organisms. Specifically, it can be implemented using an uncooled microbolometer array, with a thermal sensitivity of up to 0.05℃. This device can operate continuously in complete darkness, effectively overcoming the failure problem of traditional optical equipment at night or in smoggy environments. The polarized high-definition camera 42 refers to an optical imaging device that uses polarization filtering technology to suppress interference from water surface specular reflections. Specifically, it can be implemented using a four-way polarization sensor combined with Malus's law algorithm. This device significantly improves the ability to capture feather texture features by separating reflected light components with different polarization directions. The acoustic acquisition microphone 43 refers to a sound acquisition device with wideband response characteristics. Specifically, it can be implemented using a MEMS microphone array combined with a beamforming algorithm. This device suppresses environmental noise through spatial filtering, ensuring effective capture of low-frequency sounds. The hydrological sensor group 44 refers to a composite sensor integrating multiple water environment parameter detection functions. Specifically, it can be implemented using a combination of a pressure level gauge 441 and a fluorescence dissolved oxygen sensor 442. This device acquires real-time data on water level fluctuations and dissolved oxygen concentration changes by directly contacting the water surface.

[0063] Specifically, the infrared thermal imager 41 first detects potential targets through thermal radiation characteristics. When a heat source matching the body temperature characteristics of birds is detected, the polarized high-definition camera 42 is triggered to start shooting. The polarized imaging module filters out the interference of polarized reflected light from the water surface and focuses on the diffuse reflected light component of the waterbird's body surface, thereby acquiring key biological characteristics such as feather texture and beak morphology. The acoustic acquisition microphone 43 simultaneously acquires environmental sound wave signals, extracts call feature parameters after noise reduction processing, and compares them in real time with a pre-stored voiceprint database. The hydrological sensor group 44 continuously monitors water level changes and dissolved oxygen concentration. When the rate of water level decline exceeds a threshold, it automatically associates the current area's waterbird activity data, providing environmental variable support for behavior prediction.

[0064] Compared to existing technologies, traditional monitoring systems often employ a single infrared camera or ordinary optical camera, which struggles to effectively distinguish waterbirds from background interference in highly reflective water environments. For example, the infrared camera solution disclosed in CN117173631 A has a water surface area recognition rate of less than 40%. This solution, however, utilizes the combined operation of infrared thermal imaging and polarized light imaging, ensuring both initial target screening capabilities in dark environments and resolving imaging distortion issues on highly reflective surfaces. Furthermore, existing hydrological monitoring equipment often operates independently, failing to correlate biological behavior data in real time; for instance, key correlation signals triggering stork flocking behavior during sudden water level drops are frequently overlooked. This solution, through spatiotemporally synchronized multi-source data acquisition, achieves dynamic correlation analysis between hydrological parameters and waterbird activity.

[0065] Through the above technical solutions, this application effectively improves the accuracy and reliability of waterbird identification in complex aquatic environments, solving the problem of low water surface area recognition rate caused by the single dimension of traditional visual perception. Simultaneously, through the collaborative work of the hydrological sensor group 44 and the biological monitoring equipment, real-time data fusion of environmental parameters and biological behavior is achieved, significantly shortening the response delay time for ecological events. For example, in scenarios where wetland water levels are rapidly declining, the system can simultaneously capture data on stork aggregation behavior and hydrological changes, providing a basis for timely initiation of protection measures.

[0066] This application further proposes a hydrological sensor group 44 including a water level gauge 441 and a dissolved oxygen sensor 442. The water level gauge 441 is installed at the bottom of the adaptive adjustment frame 1 and is used to measure the water depth in real time to determine the extent of wetland flooding and to trigger early warning monitoring of waterbird foraging behavior when the water level continues to drop. The dissolved oxygen sensor 442 is used to detect the dissolved oxygen concentration in the water area and to trigger early warning monitoring of waterbird foraging behavior when the risk of fish mortality is detected.

[0067] Among them, the water level gauge 441 refers to a device that measures changes in water depth through the principle of pressure sensing. Specifically, it can be implemented using an immersion-type liquid level transmitter, and its bottom mounting method can accurately reflect the dynamic changes of the boundary of the wetland inundation zone. The dissolved oxygen sensor 442 refers to a device that detects the concentration of dissolved oxygen in water based on the principle of fluorescence quenching. Specifically, it can be implemented using an optical dissolved oxygen probe, and its detection data is used to assess the risk of deterioration of the fish's living environment.

[0068] Specifically, the water level gauge 441 constructs a model of the inundation zone's extent by continuously monitoring water depth changes, and activates an early warning mechanism when a linear downward trend in water level is detected. The dissolved oxygen sensor 442 simultaneously monitors the dissolved oxygen status of the water area, generating a risk signal when the concentration falls below the fish survival threshold. The data from both sensors are correlated and analyzed in the edge computing hub 5. When a drop in water level leads to the exposure of shallow areas or insufficient dissolved oxygen causes fish mortality, the system automatically triggers an early warning of waterbird foraging behavior, forming a real-time mapping between hydrological parameters and ecological behavior.

[0069] Compared to existing technologies, traditional systems rely solely on independent water level sensors and fixed thresholds to trigger early warnings, failing to identify gradual ecological changes. This solution, through the collaborative operation of dual sensors, establishes a dynamic correlation model between water level changes and dissolved oxygen concentration. This model can capture the causal relationship between the reduction in fish resources and the shift of waterbird foraging areas during a continuous decline in water level, overcoming the misjudgment problem caused by single-parameter monitoring.

[0070] Through the above technical solution, this application realizes multi-dimensional correlation monitoring of hydrological parameters and waterbird behavior. The water level gauge 441 dynamically captures the changing trend of the flooded area, and the dissolved oxygen sensor 442 assesses the survival status of fish in real time. The collaborative analysis of the two can accurately predict changes in waterbird foraging behavior patterns, solving the technical defect of the disconnect between hydrological data and ecological event response in traditional systems.

[0071] This application further proposes an edge computing hub 5 including a biometric identification processor 51 and a wireless data transmission module 52. The biometric identification processor 51 is connected to an acoustic acquisition microphone 43 and a polarized high-definition camera 42 for data processing of the acquired sound and image information. The wireless data transmission module 52 is connected to a remote monitoring platform 2 and is equipped with 4G / 5G / LoRa transmission protocols.

[0072] Among them, the biometric identification processor 51 refers to the computing unit deployed locally on the monitoring device, which can be implemented using an embedded AI chip. It is used to directly execute voiceprint feature extraction and image classification algorithms on the device side, avoiding the need to transmit the original audio and video data back to the cloud. The wireless data transmission module 52 refers to a transmission device that supports multiple communication protocols. It can be implemented using a multi-mode communication module, which reduces network dependence by automatically selecting the optimal transmission path, such as switching to LoRa long-distance low-power transmission when the 4G signal is weak.

[0073] Specifically, the audio stream captured by the acoustic acquisition microphone 43 and the video stream captured by the polarized high-definition camera 42 are input in real time into the biological species identification processor 51, and the voiceprint feature vector and visual recognition results are matched and verified locally. When the species matching degree of the acoustic and visual data reaches a preset threshold, only the compressed structured data is transmitted to the remote monitoring platform 2. The wireless data transmission module 52 dynamically selects the transmission protocol according to the on-site network quality. In areas with public network coverage, 5G is used to transmit real-time data first, and in areas without public network coverage, key monitoring results are sent periodically via the LoRa protocol.

[0074] Compared to existing technologies, traditional monitoring systems rely on cloud servers to process raw audio and video data, requiring continuous transmission of high-bitrate data streams, leading to a surge in communication costs. This solution, by deploying a bio-species identification processor 51 at the edge, transforms raw audio waveforms and video frames into structured feature data, reducing the daily data transmission volume per monitoring point by approximately 90%. Simultaneously, the multi-protocol wireless data transmission module 52 overcomes the limitations of a single communication method, reducing communication energy consumption by more than 75% in remote wetland scenarios compared to solutions that rely on fixed 4G transmission.

[0075] Through the above technical solution, this application effectively resolves the contradiction between high bandwidth requirements and limited communication conditions in wetland monitoring scenarios, enabling real-time collaborative processing of acoustic and visual data. Localized intelligent processing avoids the remote transmission of massive amounts of audio and video data, significantly reducing deployment costs; the multi-protocol transmission mechanism ensures data accessibility in different network environments, enabling the monitoring system to operate stably even in areas without public network coverage.

[0076] This application further proposes a remote monitoring platform 2 including a 3D reconstruction engine that integrates GPS location data to generate a waterbird distribution heat map; and an early warning and dispatch module that connects to the hydrological sensor group 44 and automatically triggers drone patrol commands when rare species are identified.

[0077] The 3D reconstruction engine is a computational system that spatially correlates discrete GPS coordinates with species observation data. Specifically, it can be implemented using ArcGIS Pro's 3D spatial interpolation algorithm, transforming discrete observation points into a continuously distributed surface through Kriging interpolation. The early warning and dispatch module is a control unit that establishes linkage between sensor data and protection equipment. Specifically, it can use the MQTT protocol to establish a communication link between hydrological sensors and UAV control terminals, automatically sending patrol commands when the species identification confidence level exceeds a preset threshold.

[0078] Specifically, the 3D reconstruction engine receives GPS positioning data and species identification results from multiple monitoring points, and generates a distribution model with thermal gradient characteristics through spatial interpolation algorithms. This model maps the frequency of waterbird activity to color intensity values, and overlays wetland topographic data in a 3D geographic coordinate system to form a visualized heat map. The early warning and dispatch module receives water level change data and species identification results collected by hydrological sensors in real time. When a national first-class protected bird is detected and the water level drops at a rate exceeding 5 centimeters per hour, it automatically generates a patrol task instruction containing coordinate parameters and transmits it to the nearest drone base station via a low-altitude communication network.

[0079] In some specific implementations, the 3D reconstruction engine can be configured to perform data fusion updates every 30 minutes, and the heatmap rendering uses WebGL technology to achieve real-time visualization on the browser side. The early warning and scheduling module can be set with multiple trigger conditions. For example, when the dissolved oxygen concentration is below 4 mg / L and black-faced spoonbill activity is detected at the same time, a drone equipped with water quality sampling equipment is automatically dispatched to perform emergency monitoring.

[0080] Compared to existing technologies, traditional monitoring systems use independent databases to store geographic information and biological data, requiring manual comparison of electronic maps and observation records, resulting in event response delays exceeding 6 hours. This solution utilizes a spatial data fusion engine to achieve real-time correlation between geographic information and ecological data, reducing data processing time to less than 20 seconds. Existing early warning mechanisms rely on fixed threshold judgments, while this solution introduces multi-parameter correlation analysis, such as simultaneously considering species protection levels and hydrological change rates, improving early warning accuracy to over 92%.

[0081] Through the aforementioned technical solutions, this application achieves intelligent linkage between hydrological monitoring data and species conservation actions. When an abnormal drop in water level is detected in an area where rare waterbirds are active, an emergency drone patrol can be initiated within 3 minutes, improving the response speed by 40 times compared to traditional manual dispatching methods. The dynamic generation of three-dimensional heat maps allows reserve managers to intuitively grasp the spatial distribution characteristics of waterbird populations, providing data support for habitat restoration projects.

[0082] This application further proposes a data processing flow for the acoustic acquisition microphone 43 and the biological species identification processor 51, including: acquiring the sound spectrogram of the protected area and segmenting it into time-frequency units; constructing an undirected graph of time-frequency units and calculating the node proximity similarity index; extracting the cluster contours through a fast convex hull algorithm; and outputting a morphological feature vector containing the center frequency and the discrete coefficient of contour curvature.

[0083] The time-frequency unit refers to the basic processing unit after discretizing the continuous acoustic signal in the time and frequency dimensions. Specifically, it can be implemented using a short-time Fourier transform, with each 30 millisecond as a time window and each 100 Hz as a frequency band, thus transforming the acoustic signal into a two-dimensional time-frequency matrix. The undirected graph refers to a graph data structure constructed using time-frequency units as nodes and the similarity between nodes as edge weights. Specifically, cosine similarity can be used to calculate the similarity of the spectral energy distribution of adjacent nodes, distinguishing effective voiceprints from environmental noise by establishing topological relationships between nodes. The proximity similarity index is an indicator that quantifies the degree of correlation between two time-frequency units in the time and frequency domains. Specifically, a dynamic time warping algorithm can be used to calculate the matching degree of energy change trajectories of adjacent nodes, used to identify continuous call signals. The fast convex hull algorithm is a calculation method for extracting the boundary of the largest convex polygon from a set of discrete data points. Specifically, it can be implemented using the Andrew monotonic chain algorithm, extracting the geometric contours of voiceprint clusters through polar angle sorting and stack operations. The center frequency refers to the main energy concentration band of the voiceprint signal in the spectrum, specifically determined by the centroid calculation method of the power spectral density curve. The contour curvature dispersion coefficient is a statistical measure that describes the complexity of the shape of the boundary of voiceprint clustering. Specifically, it can be obtained by calculating the ratio of the standard deviation to the mean of the curvature at each point on the contour line.

[0084] Specifically, after the sound signal is captured by the microphone, it is first converted into a spectrogram and then segmented into a discrete time-frequency unit matrix. An undirected graph model is then constructed, with each time-frequency unit serving as a graph node, and the edge weights between nodes calculated using the nearest neighbor similarity index. By analyzing the graph's topological structure, the distribution patterns of environmental noise and target voiceprints are effectively distinguished. A fast convex hull algorithm is applied to extract the geometric boundaries of voiceprint clusters, overcoming the contour blurring problem caused by traditional clustering methods in reverberant environments. The final output morphological feature vector integrates two types of parameters: center frequency and contour curvature discrete coefficients, providing multi-dimensional discrimination criteria for subsequent species identification.

[0085] Compared to existing technologies, traditional voiceprint recognition methods typically use fixed thresholds to segment spectrograms, which are susceptible to low-frequency noise interference in wetland reverberation environments. Our proposed solution, however, constructs an undirected graph of time-frequency units and leverages the topological analysis capabilities of graph structures to effectively distinguish the spatiotemporal distribution differences between the target voiceprint and ambient noise. Compared to the conventional K-means clustering algorithm, the fast convex hull algorithm exhibits higher geometric accuracy in extracting voiceprint boundaries, particularly in handling acoustic scenes with complex reverberation, where it maintains the stability of contour features.

[0086] Through the above technical solution, this application solves the problem of inaccurate low-frequency acoustic feature extraction in wetland environments, significantly improving the reliability of acoustic monitoring in reverberant environments. By using undirected graph modeling of time-frequency units and convex hull contour extraction, the interference of environmental noise on acoustic features is effectively suppressed, enhancing the discriminative power of feature vectors and providing a high-quality acoustic data foundation for subsequent biological species identification.

[0087] The waterbird image information is processed using a visual recognition algorithm based on an improved YOLOv7 network. The following steps are executed in the biological species identification processor 51: Retinex enhancement and super-resolution reconstruction are performed on the images from the polarized high-definition camera 42; feather texture features are focused using an attention mechanism; and species identification results and confidence scores are output. During image preprocessing: polarization component fusion: the light intensity difference between 0°, 45°, and 90° polarized images is calculated to generate a synthetic image that suppresses reflection; multi-scale Retinex enhancement: illumination correction is performed using three Gaussian kernels of different scales (scale parameters 15 / 80 / 250 pixels); network structure improvement: a polarization attention module (PAM) is added to the third layer of the backbone network: outputting a feature map with 3 channels, corresponding to the polarization component weights; loss function optimization: the loss weight for the feather region is increased to 3 times that of other regions; accuracy verification mechanism: the output confidence threshold is set to 0.92; similar species differentiation: determination is aided by the cosine similarity of the 128-dimensional LBP texture vector.

[0088] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. An automated waterbird biodiversity monitoring and data collection device, characterized in that: include: An adaptive adjustment rack, which is installed in the submersible area and can adaptively adjust according to air volume and direction; A biomimetic attraction system is installed on the adaptive adjustment frame and integrates acoustic, optical, and pheromone induction technologies. It can attract waterbirds to the location and collect the required information according to the actual situation. A group of sensing layer devices is installed on the adaptive adjustment frame to monitor waterbirds and collect their images and sounds, while also monitoring the water area in which they are located and collecting information on the water areas that affect the waterbirds. An edge computing hub, which is connected to the sensing layer device group, is used to process and transmit the acquired image and sound information; A remote monitoring platform, which is connected to the edge computing center, is used to receive data from the edge computing center and perform three-dimensional reconstruction of species distribution; The adaptive adjustment rack includes: A waterproof enclosure, wherein the edge computing hub is installed inside the waterproof enclosure; Positioning cylinders are installed on both sides of the waterproof housing, and anchors are installed on the positioning cylinders for inserting into the soil to anchor the entire equipment. The dynamic adjustment cylinder is provided in at least three sets, and the three sets of dynamic adjustment cylinders are installed in an isosceles triangle on the top of the waterproof housing; A rotating base, which is mounted on top of the dynamically adjusting cylinder; A gyroscope is mounted at the center of the bottom of the rotating base; An air volume and wind speed sensor is installed at the bottom of the rotating base to monitor air volume and wind direction. After processing the collected information, it is matched with the information collected by the gyroscope to control the dynamic adjustment cylinder and the rotating base to make real-time adjustments, so that the sensing layer device group can perform real-time adjustments during the image acquisition process. The biomimetic attraction system includes a waterproof housing and a flexible solar panel covering the waterproof housing; The biomimetic attraction system includes an optical induction module, which includes a green LED array, an infrared LED array, and a UV laser. The green LED array, the infrared LED array, and the UV laser are all mounted on the waterproof housing, and the green LED array, the infrared LED array, and the UV laser are all connected to a constant current drive circuit. The biomimetic attraction system further includes an acoustic induction module, which comprises: The microcontroller serves as the central hub; A voiceprint classification indexer, connected to the microcontroller, is used to collect voice information and then perform a retrieval. A solid-state storage chip, connected to the microcontroller, is used to extract the corresponding audio file from the solid-state storage chip after the voiceprint classification indexer completes the retrieval; A DSP processor, which is connected to a digital noise reduction chip, eliminates environmental noise in real time and outputs an analog signal to a dynamic gain controller. A speaker, connected to the dynamic gain controller, is used for final audio output; The pheromone induction module includes: A storage tank for storing a pheromone solution; A micro-flow metering pump, wherein the inlet of the micro-flow metering pump is connected to the storage tank through a pipe and a distribution valve, and the outlet is connected to an atomizing nozzle through a pipe; A blower is provided, and the liquid storage tank is installed on the atomizing nozzle to blow out the atomized pheromones.

2. The automated waterbird biodiversity monitoring and data collection device according to claim 1, characterized in that: The sensing layer device group includes: An infrared thermal imager, mounted on the rotating base, is used to sense infrared radiation from waterbirds; A polarized high-definition camera is mounted on the rotating base and is used by the infrared thermal imager to capture infrared radiation information of waterbirds and then photograph and collect images of the waterbirds. An acoustic acquisition microphone is connected to the microcontroller and is used to input the acquired sound information into the voiceprint classification indexer for retrieval. A hydrological sensor array, which is installed at the bottom of the adaptive adjustment frame, is used to monitor the water area.

3. The automated waterbird biodiversity monitoring and data collection device according to claim 2, characterized in that: The hydrological sensor array includes: A water level gauge is installed at the bottom of the adaptive adjustment frame to measure water depth in real time to determine the extent of wetland flooding, and to trigger early warning monitoring of waterbird foraging behavior when the water level continues to drop. Dissolved oxygen sensors are used to detect the dissolved oxygen concentration in the water, provide early warnings of fish mortality risks, and trigger early warning monitoring of waterbird foraging behavior.

4. The automated waterbird biodiversity monitoring and data collection device according to claim 3, characterized in that: The edge calculation center includes: A biological species identification processor, which is connected to the acoustic acquisition microphone and the polarized high-definition camera, is used to process the sound information and image information acquired by the acoustic acquisition microphone and the polarized high-definition camera; A wireless data transmission module is connected to the remote monitoring platform and is equipped with 4G / 5G / LoRa transmission protocols.

5. The automated waterbird biodiversity monitoring and data collection device according to claim 4, characterized in that: The remote monitoring platform includes: A 3D reconstruction engine that integrates GPS location data to generate a heat map of waterbird distribution; The early warning and dispatch module is connected to the hydrological sensor group and automatically triggers the drone patrol command when a rare species is detected.

Images