An oil spill detection system and method based on a drone

Abstract

The application belongs to the field of oil stain detection, and particularly relates to an oil stain detection system and method based on a UAV, which comprises a high-frequency modulation laser emission module, a measurement spectrum detection module, a reference spectrum detection module, an airborne communication and control module, and a dual-mode fusion analysis and oil stain determination module. The high-frequency modulation laser emission module is used for emitting high-frequency modulation laser with a repetition frequency of 1 kHz to 10 kHz to a target area. The measurement spectrum detection module is used for collecting measurement spectrum data of the target area in the on-off cycle of the high-frequency modulation laser. The reference spectrum detection module is used for synchronously collecting reference spectrum data. The airborne communication and control module is used for injecting a time stamp into to-be-sent data and sending the to-be-sent data containing the time stamp to a cloud platform, so that the cloud platform aligns the to-be-sent data according to the time stamp. The dual-mode fusion analysis and oil stain determination module is used for background deduction processing, extraction of a fluorescent component, calculation of reflectivity and a comprehensive oil stain index, and generation of an oil stain determination result according to the comparison result of the comprehensive oil stain index and a preset threshold. The application realizes real-time detection under daytime light conditions.

Description

Technical Field

[0001] This application belongs to the field of oil pollution detection technology, specifically relating to an oil pollution detection system and method based on an unmanned aerial vehicle (UAV). Background Technology

[0002] Currently, methods for monitoring oil spills on water surfaces can be mainly divided into two categories: fixed-point monitoring and remote sensing monitoring. Fixed-point monitoring methods include video monitoring and fluorescence-induced monitoring. However, these methods have a small monitoring range, cannot handle large areas of water, are greatly affected by environmental factors, and have a high false alarm rate. Remote sensing monitoring is mainly achieved through satellites, manned aircraft, or unmanned aerial vehicles (UAVs) equipped with remote sensing sensors. Among these, UAV-based platforms, with their characteristics of speed, efficiency, and real-time performance, have become an important means of emergency oil spill remote sensing monitoring.

[0003] Laser-induced fluorescence (LIF) is a rapidly developing method for oil spill detection in recent years. This technology uses a laser to excite fluorescent substances (mainly polycyclic aromatic hydrocarbons) in oil spills, causing them to emit characteristic fluorescence. Qualitative and quantitative analysis of the oil spill can be achieved by detecting the fluorescence spectrum, and it is widely recognized as one of the most effective means of detecting marine oil spills.

[0004] Hyperspectral remote sensing technology is also widely used in the field of oil spill monitoring on water. Existing UAV-borne oil spill detection schemes generally employ imaging hyperspectral technology, acquiring image data of oil spill areas through pushbroom or frame imaging. However, this imaging scheme has an inherent drawback of complex data processing: the acquired scattered image patches must undergo processing steps such as stitching, geometric correction, radiometric correction, and fusion to generate a complete and usable georeferenced spectral data cube, resulting in a significant delay from data acquisition to obtaining the final results.

[0005] Furthermore, existing laser-induced fluorescence oil stain detection systems face a key technical bottleneck in practical applications—strong background light interference under daytime sunlight. Sunlight contains a wide range of ultraviolet to near-infrared spectral components, and this background light will: generate a large amount of background noise signals, drowning out the weak fluorescence signals generated by oil stains, resulting in a significant decrease in the signal-to-noise ratio; under high-intensity sunlight, the detector is prone to saturation, making it impossible to accurately measure the fluorescence signal of oil stains; there is an overlap between the spectral distribution of sunlight and the fluorescence spectrum of oil stains, making it difficult to accurately distinguish between the fluorescence signal of oil stains and the background signal of sunlight under daytime conditions; strong ambient light will limit the effective penetration depth of the laser, affecting the ability to detect oil stains below the water surface.

[0006] Although existing studies have attempted to reduce ambient light interference using filters, dichroic mirrors, and modulation / demodulation techniques, the detection performance of current solutions under strong daylight conditions remains unsatisfactory. While some laser fluorescence detection systems claim to be capable of daytime operation, they typically require limiting the monitoring altitude to a relatively low range (e.g., 0.5–8.0 meters), making it difficult to meet the practical needs of large-scale, high-efficiency inspections by drones.

[0007] In summary, the existing technologies have the following main shortcomings: existing laser-induced fluorescence detection technology is severely affected by background light interference in daylight conditions, resulting in a significant reduction in detection sensitivity and accuracy; existing UAV imaging hyperspectral oil spill detection systems have complex data processing procedures and poor timeliness, making it difficult to achieve near real-time emergency monitoring response; existing solutions are mostly single detection modes (simple fluorescence detection or simple hyperspectral imaging), and the reliability and accuracy of detection need to be improved; in existing dual-channel solutions (such as the combination of LIF and visual camera), the two detection methods are relatively independent, lacking a deep-level fusion analysis mechanism, and failing to achieve true synergistic effects.

[0008] To address the aforementioned issues, it is necessary to develop an unmanned aerial vehicle (UAV)-borne oil spill detection system that can accurately detect oil spills even under strong daylight conditions and possess near real-time monitoring capabilities. Summary of the Invention

[0009] To accurately detect oil spills in real time even under strong daylight conditions, in a first aspect, embodiments of this application provide an oil spill detection system based on an unmanned aerial vehicle (UAV), comprising: A high-frequency modulated laser emitting module, deployed on a drone platform, is used to emit a high-frequency modulated laser with a repetition frequency of 1kHz to 10kHz into a target area during flight following the drone platform, in order to excite the oil stains to produce fluorescence. The measurement spectrum detection module is used to collect measurement spectrum data of the target area during the switching cycle of the high-frequency modulated laser, where the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal. The reference spectral detection module is installed vertically upwards, facing the sky, and is used to synchronously acquire reference spectral data, including atmospheric path radiation and downward solar spectral irradiance over the target area. The reference spectral detection module and the measurement spectral detection module use the same spectral sensing unit and perform synchronous data acquisition based on a unified time base. The airborne communication and control module is used to receive measurement spectral data from the measurement spectral detection module, reference spectral data from the reference spectral detection module, the GNSS position of the target area, the IMU attitude of the UAV platform, and the laser timing synchronization signal to obtain the data to be transmitted; inject timestamps generated by its own clock source into the data to be transmitted, and send the data to be transmitted containing the timestamps to the cloud platform so that the cloud platform aligns the data to be transmitted according to the timestamps, reconstructs all parameters collected at the same physical moment into a fused data frame with spatiotemporal consistency, and removes abnormal data frames; The dual-modal fusion analysis and oil contamination determination module runs on a cloud platform. It is used to perform background subtraction processing on the corresponding spectral data of the on-laser using the off-laser corresponding spectral data. Based on the laser timing synchronization signal, it uses digital lock-in amplification technology to extract fluorescence intensity from the background-subtracted on-laser spectral data. It calculates reflectivity based on the off-laser corresponding spectral data, reference spectral data, and solar zenith angle. It calculates the comprehensive oil contamination index using fluorescence intensity and reflectivity, and generates the oil contamination determination result based on the comparison result of the comprehensive oil contamination index and the preset threshold.

[0010] In one possible implementation, the airborne communication and control module is also used to control the pulse timing of laser emission to ensure that laser excitation and measurement spectral data acquisition are synchronized; continuously monitor and parse instructions from the cloud platform, including remote dynamic adjustment commands for the UAV flight path, laser emission parameters, spectral detection module integration time and acquisition frequency, and UAV gimbal pointing angle, and distribute the instructions to the UAV platform's flight control system and corresponding payload actuators.

[0011] In one possible implementation, The background subtraction processing of the corresponding spectral data of the on-laser using the off-laser corresponding spectral data includes: Let the spectrum when the laser is turned on be The spectrum when the laser is off is Firstly, in the characteristic wavelength range of oil stain fluorescence [ , The two values ​​are numerically summed separately to obtain the total spectral intensity under the corresponding state: , ; in, This represents the cumulative spectral value when the laser is on. This represents the cumulative spectral value when the laser is off. and These represent the lower and upper limits of the characteristic wavelength range of oil stain fluorescence, respectively. Background subtraction processing of the corresponding spectral data of the on-laser using the off-laser corresponding spectral data is expressed as follows: .

[0012] In one possible implementation, The calculation of reflectivity based on the corresponding laser spectral data, reference spectral data, and solar zenith angle includes: Real-time atmospheric correction is performed on the measured spectral data using reference spectral data to calculate surface reflectance. ; in: To measure the value of the laser spectrum in the spectral data, Atmospheric path radiation, This refers to the downward solar spectral irradiance. This is the solar zenith angle.

[0013] In one possible implementation, The calculation of the comprehensive oil pollution index using fluorescence intensity and reflectance includes: Min-Max normalization of reflectivity: Where: λ is the wavelength. R (λ) represents the spectral reflectance at wavelength λ. The normalized spectral reflectance, and These are the minimum and maximum values ​​of spectral reflectance, respectively. Quantification of normalized spectral reflectance using Pearson correlation coefficient oil stain reference spectral reflectance The degree of linear correlation in waveform morphology; Pearson correlation coefficient ; The mean is: , ; Here, n represents the band value, and n is the number of bands. Calculate Pearson similarity S ; The fluorescence intensity F and Pearson similarity S are fused for analysis, and the comprehensive oil pollution index is calculated as α×F+β×S, where α and β are fusion weight coefficients.

[0014] In one possible implementation, the system further includes a terminal output module, used to dynamically calculate and generate a geometric unit representing the spatial coverage of this acquisition based on the GNSS coordinates in each fused data frame and the instantaneous field of view radius on the associated digital geographic base map. The module dynamically binds the comprehensive oil pollution index from the dual-modal fusion analysis and oil pollution determination module as an attribute field to the geometric unit. The system then uses a graphics processor to render the discrete geometric units with oil pollution attributes, stitching and fusing them in real time into a spatially continuous thematic map of oil pollution distribution. The rendering result is then pushed to the user monitoring interface as a low-latency video stream or a dynamically updated interactive layer.

[0015] In one possible implementation, the system also includes a mission planning and monitoring interface, which runs on the user terminal and maintains bidirectional communication with the cloud platform. This interface is used to set flight altitude, speed, spectral acquisition trigger conditions, gimbal observation angle, laser emission parameters, and sensor operating parameters, as well as to issue intervention commands based on real-time monitoring information to correct flight paths, start and stop payloads, adjust acquisition parameters, and change gimbal pointing.

[0016] Secondly, embodiments of this application provide an oil spill detection method based on an unmanned aerial vehicle (UAV), comprising: A high-frequency modulated laser with a repetition frequency of 1 kHz to 10 kHz is emitted toward the target area to excite the oil stain to produce fluorescence; The target area is collected by measuring spectral data during the switching cycle of a high-frequency modulated laser, where the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal. Simultaneously acquire reference spectral data, including atmospheric path radiation and downward solar spectral irradiance over the target area; The system receives measurement spectral data from the measurement spectral detection module, reference spectral data from the reference spectral detection module, the GNSS position of the target area, the IMU attitude of the UAV platform, and the laser timing synchronization signal to obtain the data to be transmitted. It injects a timestamp into the data to be transmitted and sends the data to be transmitted containing the timestamp to the cloud platform so that the cloud platform aligns the data to be transmitted according to the timestamp, reconstructs all parameters collected at the same physical moment into a fused data frame with spatiotemporal consistency, and removes abnormal data frames. Background subtraction is performed on the corresponding spectral data of the off laser by using the corresponding spectral data of the on laser. Based on the laser timing synchronization signal, the fluorescence intensity is extracted from the background-subtracted on laser spectral data using digital lock-in amplification technology. The reflectance is calculated based on the corresponding spectral data of the off laser, the reference spectral data, and the solar zenith angle. The comprehensive oil pollution index is calculated using the fluorescence intensity and reflectance. Based on the comparison between the comprehensive oil pollution index and the preset threshold, the oil pollution judgment result is generated.

[0017] In one possible implementation, the method further includes: Control the pulse timing of laser emission to ensure that laser excitation and measurement spectral data acquisition are synchronized; It continuously listens for and parses commands from the cloud platform, including remote dynamic adjustment commands for the UAV's flight path, laser emission parameters, spectral detection module integration time and acquisition frequency, and the UAV gimbal pointing angle, and distributes the commands to the UAV platform's flight control system and corresponding payload actuators.

[0018] Thirdly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the aforementioned oil spill detection methods based on unmanned aerial vehicles.

[0019] This application provides an oil spill detection system based on a drone, comprising: a high-frequency modulated laser emission module deployed on a drone platform, used to emit a high-frequency modulated laser with a repetition frequency of 1kHz to 10kHz towards a target area during flight following the drone platform, to excite the oil spill to produce fluorescence; a measurement spectrum detection module, used to collect measurement spectrum data of the target area during the switching cycle of the high-frequency modulated laser, wherein the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal; a reference spectrum detection module, installed vertically upwards and facing the sky, used to synchronously collect reference spectrum data, including atmospheric path radiation and downward solar spectral irradiance above the target area; the reference spectrum detection module and the measurement spectrum detection module use the same spectral sensing unit and perform synchronous data acquisition based on a unified time base; and an airborne communication and control module, used to receive measurement spectrum data from the measurement spectrum detection module and reference spectrum data from the reference spectrum detection module. The system obtains the data to be transmitted from the spectral data, the GNSS position of the target area, the IMU attitude of the UAV platform, and the laser timing synchronization signal. A timestamp generated by its own clock source is injected into the data to be transmitted, and the data containing the timestamp is sent to the cloud platform so that the cloud platform aligns the data to be transmitted according to the timestamp. All parameters collected at the same physical moment are reconstructed into a fused data frame with spatiotemporal consistency, and abnormal data frames are eliminated. The dual-modal fusion analysis and oil contamination determination module, running on the cloud platform, performs background subtraction processing on the spectral data corresponding to the off-laser, extracts fluorescence intensity from the background-subtracted on-laser spectral data using digital phase-locked amplification technology based on the laser timing synchronization signal, and calculates reflectivity based on the off-laser spectral data, reference spectral data, and solar zenith angle. A comprehensive oil contamination index is calculated using fluorescence intensity and reflectivity, and an oil contamination determination result is generated based on the comparison between the comprehensive oil contamination index and a preset threshold. This application comprehensively acquires oil pollution information from two dimensions: chemical characteristics (fluorescence spectral characteristics) and physical characteristics (reflectivity characteristics). The two methods complement each other and work synergistically. The integrated judgment result after fusion is significantly better than that of a single detection mode, and can achieve real-time detection under daylight conditions. Attached Figure Description

[0020] Figure 1 A schematic diagram of an oil spill detection system based on an unmanned aerial vehicle (UAV) is provided as an embodiment of this application. Figure 2 A schematic diagram of a method for detecting oil spills based on an unmanned aerial vehicle (UAV) provided in this application embodiment; Figure 3 This is a schematic diagram of another oil spill detection method based on an unmanned aerial vehicle (UAV) provided in an embodiment of this application. Detailed Implementation

[0021] Firstly, see [the following] Figure 1 An oil spill detection system based on an unmanned aerial vehicle (UAV) is provided in this application embodiment, comprising: The high-frequency modulated laser emitting module 101 is deployed on a drone platform and is used to emit a high-frequency modulated laser with a repetition frequency of 1kHz to 10kHz into the target area during flight following the drone platform, so as to excite the oil stain to produce fluorescence.

[0022] The UAV platform serves as the carrier and mobile platform for the detection system, providing flight capability and position stabilization. It integrates a flight control system, an attitude stabilization system, and a data transmission system, ensuring stable operation of the detection equipment during flight. The UAV platform is a multi-rotor UAV, manufactured using carbon fiber composite materials, and possesses the following characteristics: flight time of at least 30 minutes, payload of at least 3 kg, adjustable flight altitude of 10-100 meters, positioning accuracy better than ±0.5 meters, and wind resistance up to level 5.

[0023] The high-frequency modulated laser emission module is installed on the UAV platform and is used to emit a high-frequency modulated excitation laser to the target area to excite the oil stains to generate a fluorescence signal. The laser is an ultraviolet laser with a wavelength of 385nm, a pulse width of <10ns, and a repetition frequency of 1-10kHz. The laser pulse is precisely controlled by high-frequency modulation technology to excite the oil stains to generate fluorescence. The laser power is adjustable, with an adjustment range of 1-100mW to adapt to different detection distances. There is a beam shaper in front of the laser to adjust the diameter of the output spot to adapt to different detection areas.

[0024] The core technological advantage of the high-frequency modulated laser emission module lies in: 1. High-frequency modulation timing synchronization: The laser emission module and the spectral detection module are synchronized at high frequency based on a unified clock source to ensure that the fluorescence signal is collected within the precise time window after laser pulse excitation, effectively distinguishing laser-induced fluorescence from ambient background light; 2. Pulse Code Modulation: Specific pulse sequence encoding (such as pseudo-random code or sinusoidal modulation) is used to facilitate correlation detection in subsequent signal processing, further improving the signal-to-noise ratio; 3. Power adaptive adjustment: The laser power is automatically adjusted according to the real-time echo signal intensity to avoid detector saturation while ensuring effective excitation of the fluorescence signal.

[0025] The measurement spectrum detection module 102 is used to collect measurement spectrum data of the target area during the switching cycle of the high-frequency modulated laser, wherein the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal.

[0026] This module is used to simultaneously receive and detect two optical signals: the characteristic fluorescence signal generated by oil stains under laser excitation, and the reflected light signal from the target area to the laser or ambient light. Through synchronous acquisition and processing of these signals, high signal-to-noise ratio measurement spectral data is obtained. The integrated spectrometer operates in the 200nm-1000nm range, fully covering the ultraviolet (UV), visible (Vis), and part of the near-infrared (NIR) regions. It can capture the characteristic peaks of oil stain fluorescence in the short-wavelength band and the broad-spectrum information of the reflection spectrum. Its spectral resolution is controlled between 0.2nm and 1nm, which can be flexibly set according to actual detection needs, ensuring both the ability to identify fine spectral structures and balancing signal-to-noise ratio and data acquisition speed. To adapt to the field measurement needs under different terrain, lighting, and oil stain distribution conditions, the module is mounted on a programmable gimbal, supporting pitch angle rotation from 0° to 90°, thereby flexibly adjusting the detection direction and incident / receiver geometry to ensure stable and reliable spectral data acquisition even in complex environments.

[0027] This spectrometer has the following technical advantages: 1. Full-band coverage: The 200-1000nm wavelength range fully covers the spectral information of the ultraviolet (200-400nm), visible (400-700nm), and near-infrared (700-1000nm) regions. The ultraviolet region can effectively detect the characteristic fluorescence of polycyclic aromatic hydrocarbons in oil spills, the visible region can observe the color and morphological characteristics of oil spills, and the near-infrared region can acquire the reflectance spectral information of water and oil spills. 2. Complete Extraction of Fluorescent Features from Oil Spills: Different types of oil spills exhibit characteristic fluorescence in different wavelengths. The fluorescence peaks of crude oil and fuel oil are mainly distributed in the 400-600 nm range, while some oils also show characteristic reflections in the near-infrared region (700-900 nm). A wide-band design of 200-1000 nm can simultaneously capture these characteristic signals, enabling accurate identification of oil spill types; 3. Reflectance detection compatibility: Reflected light signals from the target area can be acquired within the 200-1000nm wavelength range. When used in conjunction with a reference spectrum detection module, reflectance across the entire wavelength range can be calculated to assist in oil stain determination. 4. High-resolution design: The spectral resolution of 0.2-1nm can finely distinguish the subtle spectral features of oil stains, such as fluorescence peaks and absorption valleys at specific wavelengths, which helps to improve the accuracy of oil stain type identification.

[0028] Technical advantages of gimbal design: 1. Multi-angle detection capability: By adjusting the pitch angle of the measurement spectrum detection module via the gimbal, it can adapt to different detection environments and application scenarios. When conducting large-area inspections, the detection angle can be set to near horizontal (0°) to cover a larger ground area; when conducting detailed inspections of key areas, the detection angle can be set to near vertical (90°) to obtain more accurate spectral data; 2. Sunlight Avoidance: By adjusting the detection angle, direct sunlight interference can be effectively avoided. When the sun's altitude angle is high, the detection angle is adjusted downwards to reduce the angle at which sunlight enters the field of view, thereby improving the signal-to-noise ratio. 3. Oil film thickness assessment: By comparing and analyzing detection data from different angles, the thickness of the oil film can be estimated, thus achieving a quantitative assessment of the degree of oil contamination. 4. Multi-angle data fusion: Spectral data acquired from different angles can be fused to improve the reliability and accuracy of detection.

[0029] The reference spectrum detection module 103 is installed vertically upwards, facing the sky, and is used to synchronously collect reference spectrum data, including atmospheric path radiation and downward solar spectral irradiance over the target area. The reference spectrum detection module and the measurement spectrum detection module use the same spectral sensing unit and perform synchronous data acquisition based on a unified time base.

[0030] This module is installed vertically upwards, facing the sky, to simultaneously acquire atmospheric path radiation and downlink solar spectral irradiance over the target area. Both serve as reference spectra for subsequent processing, thereby obtaining high-precision reference spectral data. Specific functions include, but are not limited to: first, quantitatively calculating and effectively separating the contribution of atmospheric path radiation to the surface radiation signal using the measured reference spectrum; second, assisting in the inversion or imposing physical constraints on the downlink solar spectral irradiance at the current time and place to correct for radiation deviations caused by changes in sky illumination conditions. Through these methods, the module can significantly eliminate the interference of time-varying sky illumination conditions such as clouds, aerosols, and water vapor, as well as atmospheric path radiation effects, on the extraction of true surface reflectance spectral information, ensuring the accuracy and stability of surface parameter (such as reflectivity and emissivity) inversion under changing atmospheric environments.

[0031] This module uses the same or rigorously cross-radiometrically calibrated spectral sensing unit as the measurement optical path, and performs synchronous data acquisition based on a unified time base, thereby continuously acquiring a sequence of sky spectral radiance data that is completely synchronized in time and co-located in space with the target measurement data.

[0032] The airborne communication and control module 104 is used to receive measurement spectral data from the measurement spectral detection module, reference spectral data from the reference spectral detection module, the GNSS position of the target area, the IMU attitude of the UAV platform, and the laser timing synchronization signal to obtain the data to be transmitted; inject the timestamp generated by its own clock source into the data to be transmitted, and send the data to be transmitted containing the timestamp to the cloud platform so that the cloud platform aligns the data to be transmitted according to the timestamp, reconstructs all parameters collected at the same physical moment into a fused data frame with spatiotemporal consistency, and removes abnormal data frames.

[0033] In one example, the airborne communication and control module is also used to control the pulse timing of laser emission to ensure that laser excitation and measurement spectral data acquisition are synchronized; continuously listen to and parse instructions from the cloud platform, including remote dynamic adjustment commands for the UAV flight path, laser emission parameters, spectral detection module integration time and acquisition frequency, and UAV gimbal pointing angle, and distribute the instructions to the UAV platform's flight control system and corresponding payload actuators.

[0034] The airborne communication and control module, serving as the system's local hub, integrates a system-on-a-chip, high-speed data buffer memory, and a 4G / 5G or broadband wireless transmission module. This module connects to the high-frequency modulated laser emission module, the measurement spectrum detection module, the reference spectrum detection module, and the high-precision GNSS (Global Navigation Satellite System) via a dedicated data interface. The SatelliteSystem (Global Navigation Satellite System) module and attitude measurement unit (IMU) are electrically connected and synchronized via communication protocols. The core operation process is as follows: First, the unit receives raw data streams from various sensors and injects a unified, precise timestamp generated by its own highly stable clock source, completing millisecond-level time alignment and encapsulation of the multi-source data. Simultaneously, it synchronously controls the pulse timing of the high-frequency modulated laser emission module to ensure precise synchronization between laser excitation and spectral acquisition. Then, a low-latency, high-bandwidth communication link is established with the ground / cloud platform through the wireless transmission module, and the encapsulated fusion data packet (including measurement spectrum, reference spectrum, GNSS position, IMU attitude, laser timing synchronization signal, etc.) is streamed in real time. At the same time, the unit continuously monitors and parses downlink commands from the platform. These commands may include remote dynamic adjustment commands for the UAV flight path, laser emission parameters, spectral detection module integration time and acquisition frequency, and the gimbal pointing angle, and reliably distributes these commands to the flight control system and corresponding payload actuators.

[0035] Its core function is to build a two-way, real-time, and command-intervention-enabled data transmission and control channel, thereby physically realizing the integrated collaboration between the airborne payload and the ground computing platform, and providing underlying communication and control guarantees for the "real-time online" monitoring capability of the entire system.

[0036] The cloud platform processes data through a data receiving and synchronization engine. This engine, serving as the data entry point and preprocessing core for both the ground and cloud platforms, is deployed on a server with a public IP address. It continuously monitors and maintains a wireless network connection with the UAV's onboard communication unit. Its operating mechanism involves capturing raw data stream packets from the air in real-time via a high-concurrency network service interface. Based on a predefined communication protocol, it parses, verifies, and decapsulates the data packets, separating them into independent data such as embedded measurement spectral data, reference spectral data, GNSS positioning data, IMU attitude data, laser timing synchronization signals, and device status data. The core function of this engine is to perform millisecond- or even microsecond-level time alignment across the aforementioned multi-source heterogeneous data streams using a high-precision hardware timestamp extracted from the data packet and generated by an onboard precision clock as the sole reference. This reconstructs all parameters collected at the same physical moment into a fused data frame with strict spatiotemporal consistency. This process ensures that in subsequent processing, each spectral data point used for analysis can be unambiguously associated with its corresponding precise geographical location, platform attitude, sky reference background, and laser excitation timing, laying an accurate data fusion foundation for the real-time analysis and spatial mapping of the entire system.

[0037] The dual-modal fusion analysis and oil contamination determination module 105 runs on a cloud platform. It is used to perform background subtraction processing on the corresponding spectral data of the off laser and the corresponding spectral data of the on laser. Based on the laser timing synchronization signal, it uses digital lock-in amplification technology to extract fluorescence intensity from the background-subtracted on laser spectral data. It calculates reflectivity based on the corresponding spectral data of the off laser, reference spectral data, and solar zenith angle. It calculates the comprehensive oil contamination index using fluorescence intensity and reflectivity, and generates the oil contamination determination result based on the comparison result of the comprehensive oil contamination index and the preset threshold.

[0038] Let the spectrum when the laser is turned on be The spectrum when the laser is off is Firstly, in the characteristic wavelength range of oil stain fluorescence [ , The two values ​​are numerically summed separately to obtain the total spectral intensity under the corresponding state: , ; in, This represents the cumulative spectral value when the laser is on. This represents the cumulative spectral value when the laser is off. and These represent the lower and upper limits of the characteristic wavelength range of oil stain fluorescence, respectively. Background subtraction processing of the corresponding spectral data of the on-laser using the off-laser corresponding spectral data is expressed as follows: .

[0039] Real-time atmospheric correction is performed on the measured spectral data using reference spectral data to calculate surface reflectance. ; in: To measure the value of the laser spectrum in the spectral data, Atmospheric path radiation, This refers to the downward solar spectral irradiance. This is the solar zenith angle.

[0040] The reflectivity algorithm uses only the reflectivity of the laser spectrum for feature extraction to obtain a pure reflected light signal.

[0041] Min-Max normalization of reflectivity: Where: λ is the wavelength. R (λ) represents the spectral reflectance at wavelength λ. The normalized spectral reflectance, and These are the minimum and maximum values ​​of spectral reflectance, respectively. Quantification of normalized spectral reflectance using Pearson correlation coefficient oil stain reference spectral reflectance The degree of linear correlation in waveform morphology; Pearson correlation coefficient ; The mean is: , ; Here, n represents the band value, and n is the number of bands. Calculate Pearson similarity S ; The fluorescence intensity F and Pearson similarity S are fused for analysis, and the comprehensive oil pollution index is calculated as α×F+β×S, where α and β are fusion weight coefficients.

[0042] In bright daylight environments, the background light from the sun suppresses fluorescence, leading to a decrease in the reliability of the fluorescence signal. In this case, the system will automatically lower α and raise β. In low-light environments at night or on cloudy days, there is no sunlight interference, and the fluorescence signal is very pure and strong. However, the reflectivity deteriorates due to the weak incident light, resulting in a poor signal-to-noise ratio. In this case, the system will automatically lower β and raise α.

[0043] This application provides an oil spill detection system based on a drone, comprising: a high-frequency modulated laser emitting module deployed on a drone platform, used to emit a high-frequency modulated laser with a repetition frequency of 1kHz to 10kHz towards a target area during flight following the drone platform, to excite the oil spill to produce fluorescence; a measurement spectrum detection module used to collect measurement spectrum data of the target area during the switching cycle of the high-frequency modulated laser, wherein the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal; a reference spectrum detection module installed vertically upwards and towards the sky, used to synchronously collect reference spectrum data, including atmospheric path radiation and downward solar spectral irradiance above the target area; the reference spectrum detection module and the measurement spectrum detection module use the same spectral sensing unit and perform synchronous data acquisition based on a unified time base; and an airborne communication and control module used to receive measurement spectrum data from the measurement spectrum detection module and reference spectrum data from the reference spectrum detection module. The system obtains the data to be transmitted from spectral data, the GNSS position of the target area, the IMU attitude of the UAV platform, and the laser timing synchronization signal. A timestamp generated by its own clock source is injected into the data to be transmitted, and the data containing the timestamp is sent to the cloud platform so that the cloud platform aligns the data to be transmitted according to the timestamp. All parameters collected at the same physical moment are reconstructed into a fused data frame with spatiotemporal consistency, and abnormal data frames are eliminated. A dual-modal fusion analysis and oil contamination determination module, running on the cloud platform, is used to perform background subtraction processing on the spectral data corresponding to the off-laser using the off-laser corresponding spectral data. Based on the laser timing synchronization signal, digital lock-in amplification technology is used to extract fluorescence intensity from the background-subtracted on-laser spectral data. Reflectivity is calculated based on the off-laser corresponding spectral data, reference spectral data, and the solar zenith angle. A comprehensive oil contamination index is calculated using fluorescence intensity and reflectivity, and an oil contamination determination result is generated based on the comparison result of the comprehensive oil contamination index with a preset threshold. This application comprehensively acquires oil contamination information from two dimensions: chemical characteristics (fluorescence spectral characteristics) and physical characteristics (reflectivity characteristics). The two methods complement each other and synergistically enhance each other, and the comprehensive determination result after fusion is significantly better than that of a single detection mode.

[0044] In one example, the system further includes a terminal output module, which dynamically calculates and generates a geometric unit representing the spatial coverage of the current data collection based on the GNSS coordinates within each fused data frame and the instantaneous field of view radius on the associated digital geographic base map. The module dynamically binds the comprehensive oil pollution index from the dual-modal fusion analysis and oil pollution determination module as an attribute field to the geometric unit. The system then uses a graphics processor to render the discrete geometric units with oil pollution attributes, stitching and merging them in real time into a spatially continuous thematic map of oil pollution distribution. The rendering result is then pushed to the user monitoring interface as a low-latency video stream or a dynamically updated interactive layer.

[0045] The terminal output module of the data processing link is deployed on a high-performance graphics computing node on the cloud platform. This engine takes the standardized data frames, which have undergone spatiotemporal fusion and are output by the data receiving and synchronization engine, as direct input. Its core algorithm first uses the high-precision GNSS coordinates within each data frame, combined with the instantaneous field-of-view radius pre-calculated or calibrated in real-time using sensor optical parameters and flight altitude, to dynamically calculate and generate a geometric unit representing the spatial coverage of this data collection within the associated digital geographic base map or custom coordinate system. This unit is preferably a circle centered on the collection point, or defined by its minimum bounding rectangle. Subsequently, the engine integrates data from dual-modal fusion analysis and oil spill determination... The corresponding results of the module—namely, the oil pollution determination result, the oil pollution concentration estimate, or the oil pollution type classification—are dynamically bound to the graphic unit as one or more attribute fields. Finally, the engine uses the graphics processor to perform high-speed parallel rendering based on the preset or user-specified visualization scheme (such as color map, transparency, and rendering order). This process stitches and merges the continuously input discrete graphic units with oil pollution attributes into a spatially continuous thematic map of oil pollution distribution in real time. The rendering results are then pushed to the user's monitoring interface in the form of a low-latency video stream or a dynamically updated interactive layer, thus completing the final transformation from non-imaging point source spectral data to intuitive spatial distribution information of oil pollution.

[0046] In one example, the system also includes a mission planning and monitoring interface, which runs on the user terminal and maintains bidirectional communication with the cloud platform. This interface is used to set flight altitude, speed, spectral acquisition trigger conditions, gimbal observation angle, laser emission parameters, and sensor operating parameters, as well as to issue intervention commands based on real-time monitoring information to correct flight paths, start and stop payloads, adjust acquisition parameters, and change gimbal pointing.

[0047] The mission planning and monitoring interface serves as the system's human-computer interaction and central control interface. Provided as a web application or local client software, it runs on the user terminal and maintains bidirectional communication with the cloud processing server. The interface's functional architecture is divided into three main logical areas: First, the mission planning area, which provides digital map loading, region-of-interest selection, automated flight path generation tools, and an integrated parameter configuration panel for users to set flight altitude, speed, spectral acquisition trigger conditions, gimbal observation angle, laser emission parameters, and sensor operating parameters. Second, the comprehensive monitoring dashboard area, which visualizes in real-time the multi-dimensional status stream from the UAV platform and processing engine in the form of a dashboard. This includes, but is not limited to, UAV position and trajectory, communication link strength, raw data reception rate, status indicators and progress of each processing stage, and dynamic oil spill distribution thematic layers generated by the terminal output module. Third, the interactive control area, which allows users to issue intervention commands during mission execution based on real-time monitoring information, such as flight path correction, payload start / stop, acquisition parameter adjustment, and gimbal pointing change.

[0048] Its core function is to provide users with an integrated task design, full-process status awareness and real-time control entry point, and to encapsulate the complex backend data flow and processing into intuitive visual feedback and simple operation instructions, thereby logically closing the complete control chain from user intent to task execution and result perception. It is the key upper-level architecture for realizing flexible application and efficient operation of the system.

[0049] Secondly, embodiments of this application provide an oil spill detection method based on an unmanned aerial vehicle (UAV), comprising: A high-frequency modulated laser with a repetition frequency of 1 kHz to 10 kHz is emitted toward the target area to excite the oil stain to produce fluorescence; The target area is collected by measuring spectral data during the switching cycle of a high-frequency modulated laser, where the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal. Simultaneously acquire reference spectral data, including atmospheric path radiation and downward solar spectral irradiance over the target area; The system receives measurement spectral data, reference spectral data, GNSS position of the target area, IMU attitude of the UAV platform, and laser timing synchronization signal to obtain the data to be transmitted. It injects timestamps into the data to be transmitted and sends the data to be transmitted containing timestamps to the cloud platform so that the cloud platform aligns the data to be transmitted according to timestamps. It reconstructs all parameters collected at the same physical moment into a fused data frame with spatiotemporal consistency and removes abnormal data frames. Background subtraction is performed on the corresponding spectral data of the off laser by using the corresponding spectral data of the on laser. Based on the laser timing synchronization signal, the fluorescence intensity is extracted from the background-subtracted on laser spectral data using digital lock-in amplification technology. The reflectance is calculated based on the corresponding spectral data of the off laser, the reference spectral data, and the solar zenith angle. The comprehensive oil pollution index is calculated using the fluorescence intensity and reflectance. Based on the comparison between the comprehensive oil pollution index and the preset threshold, the oil pollution judgment result is generated.

[0050] In one possible implementation, the method further includes: Control the pulse timing of laser emission to ensure that laser excitation and measurement spectral data acquisition are synchronized; It continuously listens for and parses commands from the cloud platform, including remote dynamic adjustment commands for the UAV's flight path, laser emission parameters, spectral detection module integration time and acquisition frequency, and the UAV gimbal pointing angle, and distributes the commands to the UAV platform's flight control system and corresponding payload actuators.

[0051] In one possible implementation, the method further includes: Based on the GNSS coordinates within each fused data frame and combined with the instantaneous field of view radius, a geometric unit representing the spatial coverage of the data collection is dynamically calculated and generated on the associated digital geographic base map. The comprehensive oil pollution index is used as an attribute field and dynamically bound to this geometric unit. The graphics processor is used for rendering, and the discrete graphic units with oil pollution attributes are stitched and merged in real time into a thematic map of oil pollution distribution that is continuously distributed in space. The rendering result is pushed to the user monitoring interface in the form of a low-latency video stream or a dynamically updated interactive layer.

[0052] In one possible implementation, the method further includes: The system sets flight altitude, speed, spectral acquisition trigger conditions, gimbal observation angle, laser emission parameters, and sensor operating parameters, and issues intervention commands based on real-time monitoring information to correct flight paths, start and stop payloads, adjust acquisition parameters, and change gimbal pointing.

[0053] The method in this application achieves comprehensive and accurate detection of oil stains by simultaneously and collaboratively detecting and fusing two methods: high-frequency modulated laser-induced fluorescence detection and reflectivity detection. (See [link to relevant documentation]). Figure 2 Specifically, it includes the following steps: S1: Task planning and parameter setting; Based on the monitoring target and area, the drone's flight path, altitude, and speed are planned. Simultaneously, laser emission parameters (wavelength, power, modulation frequency), spectral acquisition parameters (integration time, sampling frequency), and gimbal observation angle are set according to the oil spill detection requirements to ensure that the dual-modal detection system can effectively excite and acquire the fluorescence and reflectivity signals of the oil spill within the target area.

[0054] S2: Dual-modal synchronous data acquisition and real-time transmission; The drone flies along a flight path, simultaneously executing two detection modes: high-frequency modulated laser-induced fluorescence detection and reflectivity detection, combined with... Figure 3 The schematic diagram of the working principle of dual-modal fusion detection shows that it includes: (1) The high-frequency modulated laser emission module emits ultraviolet modulated pulsed laser to the target area according to the set pulse sequence; (2) The measurement spectrum detection module simultaneously acquires measurement spectra containing fluorescence signals (400nm-600nm) and reflected light signals (200nm-1000nm); (3) The reference spectrum detection module synchronously acquires the sky background spectrum as a reference spectrum; (4) The GNSS module records real-time location information, and the IMU module records platform attitude data; (5) The airborne communication and control unit stamps all data with a unified timestamp and uploads it to the cloud platform in real time via wireless network.

[0055] The two detection modes are performed simultaneously at the same time and in the same spatial location, comprehensively acquiring oil pollution information from both chemical and physical characteristics.

[0056] S3: Platform-side dual-modal real-time data processing; a. Time synchronization and initial quality screening: Align multi-source data by timestamp, determine the stability of data acquisition time, and remove abnormal data frames; b. Dual-modal signal processing: High-frequency modulation fluorescence signal extraction: Based on the laser timing synchronization signal, the laser-induced fluorescence component is extracted within a precise time window, and digital lock-in amplification technology is used to further suppress environmental background noise; Reflectivity calculation: Dark current subtraction and radiometric calibration are performed according to a standardized process, and surface reflectivity is calculated in real time using dual-path data; c. Dual-modal feature extraction: Based on the oil pollution detection model, simultaneously calculate fluorescence characteristic parameters (fluorescence intensity, peak wavelength, full width at half maximum, etc.) and reflectance characteristic parameters (reflectance value, oil pollution diagnostic spectral index, etc.). d. Dual-modal fusion oil pollution determination: The fluorescence detection results and reflectance detection results are fused and analyzed to calculate the comprehensive oil pollution index, which is compared with the preset threshold to output the oil pollution presence determination result and oil pollution type classification.

[0057] S4: Spatial-spectral fusion and dynamic visualization; The oil pollution determination results obtained from each data collection point are bound to graphic units determined by its GNSS coordinates and instantaneous field of view. As data streams continuously arrive, the platform dynamically stitches together and renders these discrete graphic units with oil pollution attributes into a continuous spatial distribution thematic map covering the entire aerial survey area, and updates the display in real time.

[0058] The thematic map of oil pollution distribution displays the distribution of oil pollution in the detection area in the form of a pseudo-color image; it also outputs the estimated value of oil pollution concentration, the statistics of oil pollution area, and timestamp and location information.

[0059] Compared with the prior art, this application has the following significant advantages: Breakthrough innovation of dual-modal fusion detection: This application is the first to combine high-frequency modulated laser-induced fluorescence detection and reflectivity detection methods for simultaneous, coordinated detection and fusion analysis. It comprehensively obtains oil pollution information from two dimensions: chemical characteristics (fluorescence spectral characteristics) and physical characteristics (reflectivity characteristics), achieving complementary advantages and synergistic effects. The comprehensive judgment result after fusion is significantly better than that of a single detection mode. Breaking through daytime detection bottlenecks: The dual-modal fusion architecture effectively overcomes the detection challenges under daytime sunlight. In strong sunlight, fluorescence signals may be interfered with, but reflectivity detection is less affected by sunlight. Reflectivity detection can identify the presence of oil stains, but it is difficult to accurately distinguish the type of oil stain, while fluorescence detection can provide chemical characteristic information about oil stains. The two methods work together to achieve all-weather oil stain detection capabilities. Improved timeliness: By adopting a non-imaging point source detection mode and combining real-time transmission with cloud processing, the output delay of data results is shortened from "hours-days" to "seconds-minutes", achieving true near real-time monitoring and meeting emergency response needs; Accurate identification of oil stain types: Fluorescent detection provides chemical characteristic information of oil stains, enabling accurate identification and classification of different types of oil stains such as crude oil, fuel oil, and lubricating oil; Data quality optimization at the source: The reference optical path design provides a co-located and synchronized sky reference, enabling more accurate real-time atmospheric correction; the gimbal's adjustable measurement optical path allows for active avoidance of unfavorable observation geometry, reducing noise interference such as flares at the source; Lightweight and intelligent system: The airborne end does not require a powerful computing unit, reducing the weight and power consumption of the payload; complex computing tasks are handled by the cloud, which facilitates algorithm upgrades and maintenance; High application flexibility: The output is essentially a discrete point cloud of "geographic coordinates + oil pollution attributes", which is easy to integrate with GIS (Geographic Information System or Geo-Information system) systems, and facilitates spatial interpolation, statistical analysis and dynamic change detection.

[0060] The following examples of emergency detection of oil spills in large lakes further illustrate this application, but the invention is not limited to this scenario.

[0061] Preparation for implementation: (1) Determine the monitoring lake area, plan the UAV route, set the flight altitude to 50 meters and the flight speed to 4 meters / second; (2) Set the laser emission parameters: wavelength 385nm, power 50mW, modulation frequency 1kHz; (3) In order to avoid water surface flare and optimize fluorescence signal acquisition, the platform presets the measurement optical path gimbal tilt angle to 45°.

[0062] Implementation process: (1) The UAV takes off and starts the oil spill detection system after arriving at the starting point of the flight path. The system first automatically collects dark spectrum and performs initial whiteboard calibration; (2) The UAV cruises at a constant speed along the predetermined flight path. The system executes two detection modes at the same time: the high-frequency modulated laser emission module emits ultraviolet pulse laser to the water body to excite the oil spill to generate fluorescence signal; the measurement spectrum detection module collects the measurement spectrum containing fluorescence signal and reflected light signal; the reference spectrum detection module collects the sky background spectrum; GNSS and IMU record position and attitude data; (3) After all data is timestamped, it is transmitted to the cloud platform in real time through the 4G network.

[0063] Real-time processing on the cloud platform: (1) Receive data, perform time alignment and initial quality screening; (2) Dual-modal signal processing, based on laser time synchronization signal, extract laser-induced fluorescence component, and use digital lock-in amplification technology to denoise; use dark spectrum and calibration coefficient to convert the original DN value (Digital Number, remote sensing image pixel brightness value) into radiance; use reference optical path data, combined with radiative transfer model, to calculate the reflectance spectrum of water surface in real time (200-1000nm, resolution 0.4nm); (3) Dual-modal feature extraction: extract fluorescence features, fluorescence intensity (400-600nm integral), peak wavelength, etc.; extract reflectance features, reflectance value, oil pollution diagnostic spectral index, etc.; (4) Dual-modal fusion oil pollution judgment: calculate comprehensive oil pollution index = α×F + β×S; compare with the judgment threshold, output oil pollution existence judgment and oil pollution type classification.

[0064] Dynamic mapping and display: (1) The platform generates a circular area representing the detection range of each point, centered on the GNSS coordinates of each collection point; (2) The oil pollution judgment result calculated for that point is assigned to this circular area and rendered with color gradient (e.g., green represents clean and red represents oil pollution); (3) As the drone flies, thousands of circular areas with color attributes are drawn on the electronic map in real time and automatically stitched into a thematic map reflecting the spatial distribution of oil pollution in the entire lake; (4) Throughout the process, users can watch the generation process of the thematic map and the changes in oil pollution distribution in real time through the platform, and can also issue flight path correction or parameter adjustment commands.

[0065] The core innovation of this application lies in constructing a dual-modal fusion detection architecture, which simultaneously and collaboratively detects and fuses two methods: high-frequency modulated laser-induced fluorescence detection and reflectivity detection. Through the following technical means, it achieves all-weather, high-sensitivity, and high-reliability oil stain detection.

[0066] The principle of high-frequency modulated laser-induced fluorescence detection is that when a high-frequency modulated laser irradiates an oily surface, polycyclic aromatic hydrocarbons and other fluorescent substances in the oil absorb the laser energy and transition from the ground state to the excited state. When these substances return from the excited state to the ground state, they release a fluorescence signal with a characteristic wavelength. By acquiring the signal within a precise time window after the laser pulse and using digital lock-in amplification technology to extract the signal component related to the laser modulation frequency, interference from ambient background light can be effectively suppressed. Different types of oil stains have different fluorescence spectral characteristics; by analyzing the fluorescence spectrum, the type of oil stain and its concentration can be identified and estimated. This method provides chemical characteristic information of oil stains. The specific processing procedure is as follows: Step 1: Spectral accumulation calculation; Let the spectrum when the laser is turned on be The spectrum when the laser is off is First, in the characteristic wavelength range [ , The two values ​​are numerically summed separately to obtain the total spectral intensity under the corresponding state: , ;in, This represents the cumulative spectral value when the laser is on. This represents the cumulative spectral value when the laser is off. and and represent the lower and upper limits of the characteristic wavelength range, respectively; and and represent the spectral intensity values ​​at the corresponding wavelengths, respectively.

[0067] Step 2: Calculate the shape factor; Furthermore, the shape factor (SF) is defined as a normalized difference measure between the two sums to quantify the relative intensity of the fluorescence signal induced by laser excitation: The shape factor SF ranges from [-1, 1]. The closer the SF value is to 1, the stronger the laser-induced fluorescence signal and the higher the probability of the presence of oil.

[0068] Step 3: Oil stain determination; Based on the comparison between the shape factor and the preset threshold θ, the presence of oil can be determined: when SF>θ, the oil fluorescence signal is considered to be effectively detected; when SF≤θ, it is considered that there is no significant oil response. The threshold θ is determined by experimental calibration and can be adaptively adjusted according to the detection environment, noise level, and false alarm rate requirements.

[0069] The basic principle of reflectance detection lies in the significant difference in reflectance characteristics between oil spills and clean water. This difference provides a physical basis for the quantitative or qualitative detection of oil spills. Specifically, data is simultaneously collected by a measurement spectral detection module and a reference spectral detection module to calculate the reflectance of the target area. The reflectance characteristics of oil spills differ significantly from those of clean water, allowing for effective identification of oil-contaminated areas using reflectance data. This method provides physical characteristic information about oil spills.

[0070] After calculating the reflectivity, an algorithm is used to identify and compare oil stains based on the reflectivity.

[0071] .

[0072] Spectral preprocessing: Min-Max normalization. To eliminate the influence of light intensity differences on the absolute value of reflectance, the spectrum to be measured is normalized using Min-Max. ; , Where λ is the wavelength. R (λ) represents the spectral reflectance at wavelength λ. The normalized spectral reflectance, and These represent the minimum and maximum values ​​of spectral reflectance, respectively. Physically, this means mapping reflectance to the [0,1] interval, preserving the morphological characteristics of the spectral curve, and eliminating amplitude drift caused by light intensity.

[0073] To construct the baseline spectrum, N typical samples were collected from both water and oil spills, and the average spectrum of each was calculated as the baseline. Water reference spectrum Oil stain reference spectrum

[0074] Similarity is measured using the Pearson correlation coefficient, which quantifies the linear correlation in waveform morphology between the measured spectrum and the reference spectrum. Pearson correlation coefficient: ; The mean is: , ; Here, n represents the band value, and n is the number of bands. Pearson similarity (for r∈[ Linear mapping from [1,1] to [0,1]): . =1 indicates that the waveforms are completely positively correlated (the shapes are completely identical). =0 indicates that it is completely unrelated; =0.5 indicates neutral.

[0075] Classification decision-making involves calculating the Pearson similarity between the measured spectrum and water and oil pollution standards, respectively. S water and S oil The following rules apply to the determination:

[0076] in This is the confidence threshold, used to suppress false positives in areas with blurred boundaries. Confidence index: C = | - |, C∈[0,1], the larger C is, the higher the classification confidence.

[0077] The dual-modal fusion detection principle combines high-frequency modulated laser-induced fluorescence detection and reflectivity detection simultaneously, collaboratively, and fused for analysis, achieving complementary advantages: fluorescence detection provides chemical characteristics (oil contamination type identification), while reflectivity detection provides physical characteristics (oil contamination presence determination). Both methods comprehensively acquire oil contamination information from different dimensions. In strong sunlight, fluorescence signals may be interfered with, but reflectivity detection is less affected by sunlight. While reflectivity detection can identify the presence of oil contamination, it struggles to accurately distinguish oil contamination types, whereas fluorescence detection provides chemical characteristic information. The two methods work together to achieve all-weather detection. The fusion analysis of the two detection results significantly improves the accuracy and reliability of detection, reducing the false positive and false negative rates of single detection modes.

[0078] The oil spill detection method based on UAV provided in this application can be extended to the following application scenarios: (1) Marine oil spill emergency monitoring: conduct large-scale oil spill patrols on the sea surface, detect and locate the source of oil film leakage in real time, and assess the trend of oil spill spread; (2) Port oil spill detection: conduct routine oil spill monitoring in port terminals, waterways and other areas, and promptly detect ship oil spills and oil leaks; (3) River oil spill monitoring: conduct oil spill detection in freshwater bodies such as rivers, lakes and reservoirs, and assess oil pollution caused by industrial emissions and traffic accidents; (4) Industrial area drainage outlet monitoring: conduct real-time monitoring of industrial enterprise drainage outlets, and quickly detect illegal discharge of oily wastewater; (5) Monitoring around oil depots and gas stations: conduct regular inspections of the soil and water around oil storage facilities to prevent oil spill pollution.

[0079] Thirdly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the aforementioned oil spill detection methods based on unmanned aerial vehicles.

[0080] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (SSD)).

[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0082] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the method embodiments are described more simply because they are similar to the system embodiments; relevant parts can be referred to the descriptions in the system embodiments.

[0083] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. An oil spill detection system based on a drone, characterized in that, include: A high-frequency modulated laser emitting module, deployed on a drone platform, is used to emit a high-frequency modulated laser with a repetition frequency of 1kHz to 10kHz into a target area during flight following the drone platform, in order to excite the oil stains to produce fluorescence. The measurement spectrum detection module is used to collect measurement spectrum data of the target area during the switching cycle of the high-frequency modulated laser, where the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal. The reference spectral detection module is installed vertically upwards, facing the sky, and is used to synchronously acquire reference spectral data, including atmospheric path radiation and downward solar spectral irradiance over the target area. The reference spectral detection module and the measurement spectral detection module use the same spectral sensing unit and perform synchronous data acquisition based on a unified time base. The airborne communication and control module is used to receive measurement spectral data from the measurement spectral detection module, reference spectral data from the reference spectral detection module, the GNSS position of the target area, the IMU attitude of the UAV platform, and the laser timing synchronization signal to obtain the data to be transmitted. Inject timestamps generated by its own clock source into the data to be sent, and send the data to be sent containing the timestamps to the cloud platform so that the cloud platform can align the data to be sent according to the timestamps, reconstruct all parameters collected at the same physical moment into a fused data frame with spatiotemporal consistency, and remove abnormal data frames. The dual-modal fusion analysis and oil contamination determination module runs on a cloud platform. It is used to perform background subtraction processing on the corresponding spectral data of the on-laser using the off-laser corresponding spectral data. Based on the laser timing synchronization signal, it uses digital lock-in amplification technology to extract fluorescence intensity from the background-subtracted on-laser spectral data. It calculates reflectivity based on the off-laser corresponding spectral data, reference spectral data, and solar zenith angle. It calculates the comprehensive oil contamination index using fluorescence intensity and reflectivity, and generates the oil contamination determination result based on the comparison result of the comprehensive oil contamination index and the preset threshold.

2. The system according to claim 1, characterized in that, The airborne communication and control module is also used to control the pulse timing of laser emission to ensure that laser excitation and measurement spectral data acquisition are synchronized; it continuously listens to and parses instructions from the cloud platform, including remote dynamic adjustment commands for the UAV flight path, laser emission parameters, spectral detection module integration time and acquisition frequency, and UAV gimbal pointing angle, and distributes the instructions to the UAV platform's flight control system and corresponding payload actuators.

3. The system according to claim 1, characterized in that, The background subtraction processing of the corresponding spectral data of the on-laser using the off-laser corresponding spectral data includes: Let the spectrum when the laser is turned on be The spectrum when the laser is off is Firstly, in the characteristic wavelength range of oil stain fluorescence [ , The two values ​​are numerically summed separately to obtain the total spectral intensity under the corresponding state: ， ； in, This represents the cumulative spectral value when the laser is on. This represents the cumulative spectral value when the laser is off. and These represent the lower and upper limits of the characteristic wavelength range of oil stain fluorescence, respectively. Background subtraction processing of the corresponding spectral data of the on-laser using the off-laser corresponding spectral data is expressed as follows: .

4. The system according to claim 1, characterized in that, The calculation of reflectivity based on the corresponding laser spectral data, reference spectral data, and solar zenith angle includes: Real-time atmospheric correction is performed on the measured spectral data using reference spectral data to calculate surface reflectance. ； in: To measure the value of the laser spectrum in the spectral data, Atmospheric path radiation, This refers to the downward solar spectral irradiance. This is the solar zenith angle.

5. The system according to claim 4, characterized in that, The calculation of the comprehensive oil pollution index using fluorescence intensity and reflectance includes: Min-Max normalization of reflectivity: Where: λ is the wavelength. R (λ) represents the spectral reflectance at wavelength λ. The normalized spectral reflectance, and These are the minimum and maximum values ​​of spectral reflectance, respectively. Quantification of normalized spectral reflectance using Pearson correlation coefficient oil stain reference spectral reflectance The degree of linear correlation in waveform morphology; Pearson correlation coefficient ; The mean is: , ; Here, n represents the band value, and n is the number of bands. Calculate Pearson similarity S ; The fluorescence intensity F and Pearson similarity S are fused for analysis, and the comprehensive oil pollution index is calculated as α×F+β×S, where α and β are fusion weight coefficients.

6. The system according to claim 3, characterized in that, The system also includes a terminal output module, which is used to dynamically calculate and generate a geometric unit representing the spatial coverage of this acquisition based on the GNSS coordinates in each fused data frame and the instantaneous field of view radius in the associated digital geographic base map, and dynamically bind the comprehensive oil pollution index from the dual-modal fusion analysis and oil pollution determination module as an attribute field to the geometric unit. Using a graphics processor for rendering, discrete graphic units with oil stain attributes are stitched and merged in real time into a thematic map of oil stain distribution that is continuously distributed in space. The rendering results are then pushed to the user monitoring interface in the form of a low-latency video stream or a dynamically updated interactive layer.

7. The system according to claim 1, characterized in that, The system also includes a mission planning and monitoring interface, which runs on the user terminal and maintains two-way communication with the cloud platform. It is used to set flight altitude, speed, spectral acquisition trigger conditions, gimbal observation angle, laser emission parameters and sensor operating parameters, as well as to issue intervention commands based on real-time monitoring information to correct flight paths, start and stop payloads, adjust acquisition parameters and change gimbal pointing.

8. A method for detecting oil spills based on a drone, characterized in that, include: A high-frequency modulated laser with a repetition frequency of 1 kHz to 10 kHz is emitted toward the target area to excite the oil stain to produce fluorescence; The target area is collected by measuring spectral data during the switching cycle of a high-frequency modulated laser, where the laser-on phase corresponds to the fluorescence signal and the laser-off phase corresponds to the reflected light signal. Simultaneously acquire reference spectral data, including atmospheric path radiation and downward solar spectral irradiance over the target area; Receive measurement spectral data, reference spectral data, GNSS position of the target area, IMU attitude of the UAV platform, and laser timing synchronization signal to obtain the data to be transmitted; Inject timestamps into the data to be sent, and send the data to be sent containing timestamps to the cloud platform so that the cloud platform can align the data to be sent according to timestamps, reconstruct all parameters collected at the same physical moment into a fused data frame with spatiotemporal consistency, and remove abnormal data frames. Background subtraction is performed on the corresponding spectral data of the off laser by using the corresponding spectral data of the on laser. Based on the laser timing synchronization signal, the fluorescence intensity is extracted from the background-subtracted on laser spectral data using digital lock-in amplification technology. The reflectance is calculated based on the corresponding spectral data of the off laser, the reference spectral data, and the solar zenith angle. The comprehensive oil pollution index is calculated using the fluorescence intensity and reflectance. Based on the comparison between the comprehensive oil pollution index and the preset threshold, the oil pollution judgment result is generated.

9. The method according to claim 8, characterized in that, The method further includes: Control the pulse timing of laser emission to ensure that laser excitation and measurement spectral data acquisition are synchronized; It continuously listens for and parses commands from the cloud platform, including remote dynamic adjustment commands for the UAV's flight path, laser emission parameters, spectral detection module integration time and acquisition frequency, and the UAV gimbal pointing angle, and distributes the commands to the UAV platform's flight control system and corresponding payload actuators.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in claim 8 or 9.

Images