A visible light unmanned aerial vehicle-based intelligent monitoring and early warning method for urban small and micro water bodies

Abstract

This invention discloses an intelligent monitoring and early warning method for small urban water bodies based on visible light drones, belonging to the field of water quality prediction and management technology. The method acquires water body images using a drone equipped with a visible light sensor, automatically stitches them together to generate orthophotos, automatically segments water body areas and outputs boundary vector layers, extracts band DN values ​​and constructs a water quality parameter feature dataset through normalization, band combination, and chromaticity angle calculation, uses a machine learning model to invert chlorophyll a concentration, turbidity, and phycocyanin concentration, and optimizes the model accuracy by combining measured data, performs statistical analysis and future trend prediction on the inversion results, and generates thematic maps, completes water quality evaluation according to standards, and automatically outputs reports and pushes them to relevant units when early warning conditions are met. This invention uses low-cost visible light equipment to achieve fully automated intelligent monitoring, effectively improving inversion accuracy and regulatory efficiency, and adapting to the needs of large-scale, routine monitoring and early warning of small urban water bodies.

Description

Technical Field

[0001] This invention belongs to the field of water quality prediction and management technology, specifically relating to an intelligent monitoring and early warning method for small urban water bodies based on visible light drones. Background Technology

[0002] As aquatic ecosystems closely linked to residents' production and daily life, urban small water bodies currently lack long-term, efficient, and low-cost monitoring and management methods. Among existing water quality monitoring technologies, multispectral and hyperspectral UAV remote sensing monitoring methods are relatively mature, but the equipment costs are high and the operation is highly specialized, making it difficult to promote and apply them on a large scale in urban small water body scenarios.

[0003] Drones equipped with visible light sensors can only acquire RGB broadband images, making it difficult to accurately capture the spectral characteristics of water quality parameters. Furthermore, the accuracy of inversion is easily affected by factors such as lighting conditions, shooting angle, and water body shadows, leading to large inversion errors. At the same time, existing technologies lack methods for screening water quality features from visible light RGB images, high-precision machine learning inversion models, and automated, intelligent monitoring, early warning, and reporting processes, making it difficult to meet the needs for rapid, low-cost, and routine monitoring of small urban water bodies.

[0004] Therefore, developing a monitoring method based on visible light drones that can achieve high-precision inversion of water quality parameters, intelligent analysis, and automatic early warning is of great practical significance for improving the regulatory efficiency of small urban water bodies. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes an intelligent monitoring and early warning method for small urban water bodies based on visible light drones. This method employs low-cost visible light equipment to achieve fully automated intelligent monitoring, effectively improving inversion accuracy and regulatory efficiency, and adapting to the needs of large-scale and routine monitoring and early warning of small urban water bodies.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for intelligent monitoring and early warning of small urban water bodies based on visible light unmanned aerial vehicles (UAVs) includes the following steps: S1. Data Acquisition: Using drones equipped with visible light sensors to acquire images of the monitored water bodies; S2. Automated image stitching: Correction, alignment, feature matching, color equalization and fusion processing are performed on images collected by UAVs, and the stitched images generate complete orthophotos of the monitored water bodies. S3. Automatic water body extraction: Identify and segment water bodies in the stitched images, extract water body masks, and output boundary vector layers. S4. Color Feature Extraction: Extract the DN values ​​of image bands, and construct a water quality parameter feature dataset by normalization and band combination calculation. S5. Water quality parameter inversion: Based on the machine learning model, the concentration of chlorophyll a, turbidity, and phycocyanin are inverted and predicted, and the model is optimized by combining the measured data. S6. Intelligent Analysis and Trend Prediction: Perform statistical analysis on the inversion results, predict the future trend of water quality parameters, and generate thematic maps of the spatial distribution of water quality parameters. S7. Water quality assessment and early warning: Conduct water quality assessments in accordance with standards, and automatically generate water quality reports and push them to relevant units when early warning conditions are met.

[0007] Preferably, the specific process of step S1 is as follows: S11. Select a time period with sufficient sunlight and shadows less than the threshold, plan the drone flight path to cover all monitored water bodies, set the flight altitude to 100m, and the image overlap rate to 80%; S12. Fix the camera's ISO, shutter speed, and aperture parameters, keep the camera vertical for shooting, ensure consistent brightness across multiple images, and obtain RGB images of the monitored water body.

[0008] Preferably, the specific process of step S2 is as follows: S21. Image Correction: Orthorectify the images acquired by the UAV using flight attitude data and elevation data to eliminate geometric distortion. S22. Feature Extraction: Use the ORB algorithm or SURF algorithm to extract feature points in the overlapping areas of adjacent images; S23. Image alignment: Based on the spatial transformation relationship between images, the RANSAC algorithm is used to select reliable matching points to achieve image alignment. S24. Image stitching: Image spatial stitching is completed based on feature point matching; S25, Image Color Equalization: Uses a color equalization algorithm to eliminate color and brightness differences between images; S26. Image Fusion: Use a weighted fusion algorithm to eliminate seam lines and generate a complete orthophoto of the monitored water body.

[0009] Preferably, the specific process of step S4 is as follows: S41. Extract the DN values ​​of the three bands B1, B2, and B3 for each effective pixel in the water boundary vector layer, perform gamma correction and normalization to obtain r, g, and b, and calculate the chromaticity angle using the following formula: , , , In the formula, These are the normalized redness coordinates; These are the normalized greenness coordinates; B1 represents the normalized blue hue coordinates; B2 represents the red band DN value; B3 represents the green band DN value; and B4 represents the blue band pixel value. S42. Perform band combination calculations on the data from each band to form characteristic variables. The calculation formula is: F1 = F2= F3= F4= F5= In the formula, F1 is the first band combination; F2 is the second band combination; F3 is the third band combination; F4 is the fourth band combination; and F5 is the fifth band combination. , , All values ​​are normalized chromaticity coordinates for different bands; calculations are performed on different combinations of bands to obtain a water quality parameter feature dataset.

[0010] Preferably, the specific process of step S5 is as follows: S51. Call the pre-trained machine learning model, input r, g, b, chromaticity angle and band combination features, and perform inversion of the monitoring water body chlorophyll a concentration, turbidity and phycocyanin concentration data. S52. On-site sampling is used to obtain measured data, and paired datasets are constructed with image features to obtain paired data; S53. Divide the paired dataset into training and testing sets in an 80%:20% ratio for model training and evaluation, using R... 2 The hyperparameters of the indicator optimization model are calculated using the following formula: In the formula, R 2 The coefficient of determination; It is the first Measured values ​​at each monitoring point; No. Predicted values ​​for each monitoring point; It is the average of the measured data; It is the predicted mean of the data; For monitoring points; This represents the total number of monitoring points. S54. Use the optimized machine learning model to carry out routine monitoring and inversion.

[0011] Preferably, the specific process of step S6 is as follows: S61. Automatically perform statistical analysis of water quality parameters, including average, variance, maximum and minimum values, analyze the time trend of the statistical data, predict the average value of each water quality parameter for the next seven days based on machine learning methods, and display the time trend of each water quality parameter for the next seven days; the water quality parameters include chromaticity angle, chlorophyll a concentration, turbidity and phycocyanin concentration. S62. Sort the historical measured data of each water quality parameter of the monitored water body by magnitude, and calculate the interquartile range. The calculation formula is as follows: In the formula, IQR is the interquartile range; This represents the value at the 25th percentile of historical measured data. This represents the value at the 75th percentile of historical measured data. The anomaly detection threshold is set as follows: In the formula, This is the adjustment coefficient; when When the threshold value is 1.5, the water quality is judged as abnormal if it exceeds the judgment threshold; when... If the water quality exceeds the judgment threshold, the water quality will be judged as extremely abnormal; the average measured data of each water quality parameter will be compared with the abnormal judgment threshold. S63. Based on the boundary vector layer and the optimized machine learning model, generate a thematic map of the spatial distribution of water quality parameters.

[0012] Preferably, the process for determining the warning conditions in step S7 is as follows: After data accumulation, it is determined whether the predicted average values ​​of each water quality parameter of the monitored water body are in an abnormal water quality state. If the predicted average values ​​of the water quality parameters of the monitored water body are normal, the warning step is skipped; if the predicted average values ​​of the water quality parameters of the monitored water body are abnormal, the predicted trend change rate for the next seven days and the trend change rate for the past seven days are calculated. The formula for calculating the trend change rate is: In the formula, The rate of change of the trend; Indexed by the sequence number of the time point; For the predicted time range; For time sequence number; For the corresponding water quality parameter Daily average; The trend ratio is calculated using the following formula: In the formula, R is the trend ratio; Predict the rate of change of the trend over the next seven days; This represents the rate of change of the trend over the past seven days. If any one of the water quality parameters simultaneously satisfies the conditions of water quality abnormality and If the warning is triggered, then the corresponding water quality parameter warning procedure will be performed; otherwise, the warning procedure will be skipped.

[0013] After adopting the above technical solution, the present invention has the following beneficial effects: The present invention uses a drone equipped with a visible light sensor, which eliminates the need for expensive multispectral or hyperspectral equipment, has a low operating threshold and is easy to deploy, and can be applied on a large scale in urban small water body scenarios, with low cost and easy promotion; by constructing a feature set through RGB band normalization, chromaticity angle calculation and multi-form band combination, and iteratively optimizing the machine learning model with measured data, the inversion error caused by factors such as illumination and shadow is significantly reduced; from image acquisition, stitching, water body extraction, parameter inversion, trend analysis to early warning report output, no manual intervention is required throughout the entire process, greatly improving monitoring efficiency; the interquartile range method is used to identify water quality anomalies, and the early warning conditions are determined by the ratio of trend change rates, with rigorous early warning logic and timely response, which can support water environment management units to deal with the situation quickly; the spatial distribution thematic map of water quality parameters and water quality report are automatically generated, the data is intuitive, and the results can be directly used for regulatory decision-making, improving the efficiency of small water body governance. Attached Figure Description

[0014] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] like Figure 1 As shown, a method for intelligent monitoring and early warning of small urban water bodies based on visible light unmanned aerial vehicles (UAVs) includes the following steps: S1. Data Acquisition: Using drones equipped with visible light sensors to acquire images of the monitored water bodies; The specific process of step S1 is as follows: S11. Select a time period with sufficient sunlight and shadows less than the threshold, plan the drone flight path to cover all monitored water bodies, set the flight altitude to 100m, and the image overlap rate to 80%; S12. Fix the camera's ISO, shutter speed, and aperture parameters, keep the camera vertical for shooting, ensure the brightness consistency of multiple images, and obtain RGB images of the monitored water body; S2. Automated image stitching: Correction, alignment, feature matching, color equalization and fusion processing are performed on images collected by UAVs, and the stitched images generate complete orthophotos of the monitored water bodies. The specific process of step S2 is as follows: S21. Image Correction: Orthorectify the images acquired by the UAV using flight attitude data and elevation data to eliminate geometric distortion. S22. Feature Extraction: Use the ORB algorithm or SURF algorithm to extract feature points in the overlapping areas of adjacent images; S23. Image alignment: Based on the spatial transformation relationship between images, the RANSAC algorithm is used to select reliable matching points to achieve image alignment. S24. Image stitching: Image spatial stitching is completed based on feature point matching; S25, Image Color Equalization: Uses a color equalization algorithm to eliminate color and brightness differences between images; S26. Image Fusion: Use a weighted fusion algorithm to eliminate seam lines and generate a complete orthophoto of the monitored water body; S3. Automatic water body extraction: Identify and segment water bodies in the stitched images, extract water body masks, and output boundary vector layers. S4. Color Feature Extraction: Extract the DN values ​​of image bands, and construct a water quality parameter feature dataset by normalization and band combination calculation. The specific process of step S4 is as follows: S41. Extract the DN values ​​of the three bands B1, B2, and B3 for each effective pixel in the water boundary vector layer, perform gamma correction and normalization to obtain r, g, and b, and calculate the chromaticity angle using the following formula: , , , In the formula, These are the normalized redness coordinates; These are the normalized greenness coordinates; B1 represents the normalized blue hue coordinates; B2 represents the red band DN value; B3 represents the green band DN value; and B4 represents the blue band pixel value. S42. Perform band combination calculations on the data from each band to form characteristic variables. The calculation formula is: F1 = F2= F3= F4= F5= In the formula, F1 is the first band combination; F2 is the second band combination; F3 is the third band combination; F4 is the fourth band combination; and F5 is the fifth band combination. , , ; All values ​​are normalized chromaticity coordinates for different bands; calculations are performed on different band combinations to obtain a water quality parameter feature dataset. S5. Water quality parameter inversion: Based on the machine learning model, the concentration of chlorophyll a, turbidity, and phycocyanin are inverted and predicted, and the model is optimized by combining the measured data. The specific process of step S5 is as follows: S51. Call the pre-trained machine learning model, input r, g, b, chromaticity angle and band combination features, and perform inversion of the monitoring water body chlorophyll a concentration, turbidity and phycocyanin concentration data. S52. On-site sampling is used to obtain measured data, and paired datasets are constructed with image features to obtain paired data; S53. Divide the paired dataset into training and testing sets in an 80%:20% ratio for model training and evaluation, using R... 2 The hyperparameters of the indicator optimization model are calculated using the following formula: In the formula, R 2 The coefficient of determination; It is the first Measured values ​​at each monitoring point; No. Predicted values ​​for each monitoring point; It is the average of the measured data; It is the predicted mean of the data; For monitoring points; This represents the total number of monitoring points. S54. Use the optimized machine learning model to carry out routine monitoring and inversion; S6. Intelligent Analysis and Trend Prediction: Perform statistical analysis on the inversion results, predict the future trend of water quality parameters, and generate thematic maps of the spatial distribution of water quality parameters. The specific process of step S6 is as follows: S61. Automatically perform statistical analysis of water quality parameters, including average, variance, maximum and minimum values, analyze the time trend of the statistical data, predict the average value of each water quality parameter for the next seven days based on machine learning methods, and display the time trend of each water quality parameter for the next seven days; the water quality parameters include chromaticity angle, chlorophyll a concentration, turbidity and phycocyanin concentration. S62. Sort the historical measured data of each water quality parameter of the monitored water body by magnitude, and calculate the interquartile range. The calculation formula is as follows: In the formula, IQR is the interquartile range; This represents the value at the 25th percentile of historical measured data. This represents the value at the 75th percentile of historical measured data. The anomaly detection threshold is set as follows: In the formula, This is the adjustment coefficient; when When the threshold value is 1.5, the water quality is judged as abnormal if it exceeds the judgment threshold; when... If the water quality exceeds the judgment threshold, the water quality will be judged as extremely abnormal; the average measured data of each water quality parameter will be compared with the abnormal judgment threshold. S63. Based on the boundary vector layer and the optimized machine learning model, generate a thematic map of the spatial distribution of water quality parameters; S7. Water quality assessment and early warning: Conduct water quality assessments in accordance with standards, and automatically generate water quality reports and push them to relevant units when early warning conditions are met; The process of determining the warning conditions in step S7 is as follows: After data accumulation, it is determined whether the predicted average values ​​of each water quality parameter of the monitored water body are in an abnormal water quality state. If the predicted average values ​​of the water quality parameters of the monitored water body are normal, the warning step is skipped; if the predicted average values ​​of the water quality parameters of the monitored water body are abnormal, the predicted trend change rate for the next seven days and the trend change rate for the past seven days are calculated. The formula for calculating the trend change rate is: In the formula, The rate of change of the trend; Indexed by the sequence number of the time point; For the predicted time range; For time sequence number; For the corresponding water quality parameter Daily average; The trend ratio is calculated using the following formula: In the formula, R is the trend ratio; Predict the rate of change of the trend over the next seven days; This represents the rate of change of the trend over the past seven days. If any one of the water quality parameters simultaneously satisfies the conditions of water quality abnormality and If the warning is triggered, then the corresponding water quality parameter warning procedure will be performed; otherwise, the warning procedure will be skipped.

[0017] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligent monitoring and early warning of small urban water bodies based on visible light unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1. Data Acquisition: Using drones equipped with visible light sensors to acquire images of the monitored water bodies; S2. Automated image stitching: Correction, alignment, feature matching, color equalization and fusion processing are performed on images collected by UAVs, and the stitched images generate complete orthophotos of the monitored water bodies. S3. Automatic water body extraction: Identify and segment water bodies in the stitched images, extract water body masks, and output boundary vector layers. S4. Color Feature Extraction: Extract the DN values ​​of image bands, and construct a water quality parameter feature dataset by normalization and band combination calculation. S5. Water quality parameter inversion: Based on the machine learning model, the concentration of chlorophyll a, turbidity, and phycocyanin are inverted and predicted, and the model is optimized by combining the measured data. S6. Intelligent Analysis and Trend Prediction: Perform statistical analysis on the inversion results, predict the future trend of water quality parameters, and generate thematic maps of the spatial distribution of water quality parameters. S7. Water quality assessment and early warning: Conduct water quality assessments in accordance with standards, and automatically generate water quality reports and push them to relevant units when early warning conditions are met.

2. The intelligent monitoring and early warning method for small urban water bodies based on visible light unmanned aerial vehicles as described in claim 1, characterized in that, The specific process of step S1 is as follows: S11. Select a time period with sufficient sunlight and shadows less than the threshold, plan the drone flight path to cover all monitored water bodies, set the flight altitude to 100m, and the image overlap rate to 80%; S12. Fix the camera's ISO, shutter speed, and aperture parameters, keep the camera vertical for shooting, ensure consistent brightness across multiple images, and obtain RGB images of the monitored water body.

3. The intelligent monitoring and early warning method for small urban water bodies based on visible light unmanned aerial vehicles as described in claim 1, characterized in that, The specific process of step S2 is as follows: S21. Image Correction: Orthorectify the images acquired by the UAV using flight attitude data and elevation data to eliminate geometric distortion. S22. Feature Extraction: Use the ORB algorithm or SURF algorithm to extract feature points in the overlapping areas of adjacent images; S23. Image alignment: Based on the spatial transformation relationship between images, the RANSAC algorithm is used to select reliable matching points to achieve image alignment. S24. Image stitching: Image spatial stitching is completed based on feature point matching; S25, Image Color Equalization: Uses a color equalization algorithm to eliminate color and brightness differences between images; S26. Image Fusion: Use a weighted fusion algorithm to eliminate seam lines and generate a complete orthophoto of the monitored water body.

4. The intelligent monitoring and early warning method for small urban water bodies based on visible light unmanned aerial vehicles as described in claim 1, characterized in that, The specific process of step S4 is as follows: S41. Extract the DN values ​​of the three bands B1, B2, and B3 for each effective pixel in the water boundary vector layer, perform gamma correction and normalization to obtain r, g, and b, and calculate the chromaticity angle using the following formula: , , , In the formula, These are the normalized redness coordinates; These are the normalized greenness coordinates; B1 represents the normalized blue hue coordinates; B2 represents the red band DN value; B3 represents the green band DN value; and B4 represents the blue band pixel value. S42. Perform band combination calculations on the data from each band to form characteristic variables. The calculation formula is: F1 = F2= F3= F4= F5= In the formula, F1 is the first band combination; F2 is the second band combination; F3 is the third band combination; F4 is the fourth band combination; and F5 is the fifth band combination. , , All values ​​are normalized chromaticity coordinates for different bands; calculations are performed on different combinations of bands to obtain a water quality parameter feature dataset.

5. The intelligent monitoring and early warning method for small urban water bodies based on visible light unmanned aerial vehicles as described in claim 1, characterized in that, The specific process of step S5 is as follows: S51. Call the pre-trained machine learning model, input r, g, b, chromaticity angle and band combination features, and perform inversion of the monitoring water body chlorophyll a concentration, turbidity and phycocyanin concentration data. S52. On-site sampling is used to obtain measured data, and paired datasets are constructed with image features to obtain paired data; S53. Divide the paired dataset into training and testing sets in an 80%:20% ratio for model training and evaluation, using R... 2 The hyperparameters of the indicator optimization model are calculated using the following formula: In the formula, R 2 The coefficient of determination; It is the first Measured values ​​at each monitoring point; No. Predicted values ​​for each monitoring point; It is the average of the measured data; It is the predicted mean of the data; For monitoring points; This represents the total number of monitoring points. S54. Use the optimized machine learning model to carry out routine monitoring and inversion.

6. The intelligent monitoring and early warning method for small urban water bodies based on visible light unmanned aerial vehicles as described in claim 1, characterized in that, The specific process of step S6 is as follows: S61. Automatically perform statistical analysis of water quality parameters, including average, variance, maximum and minimum values, analyze the time trend of the statistical data, predict the average value of each water quality parameter for the next seven days based on machine learning methods, and display the time trend of each water quality parameter for the next seven days; the water quality parameters include chromaticity angle, chlorophyll a concentration, turbidity and phycocyanin concentration. S62. Sort the historical measured data of each water quality parameter of the monitored water body by magnitude, and calculate the interquartile range. The calculation formula is as follows: In the formula, IQR is the interquartile range; This represents the value at the 25th percentile of historical measured data. This represents the value at the 75th percentile of historical measured data. The anomaly detection threshold is set as follows: In the formula, This is the adjustment coefficient; when When the threshold value is 1.5, the water quality is judged as abnormal if it exceeds the judgment threshold; when... If the water quality exceeds the judgment threshold, the water quality will be judged as extremely abnormal; the average measured data of each water quality parameter will be compared with the abnormal judgment threshold. S63. Based on the boundary vector layer and the optimized machine learning model, generate a thematic map of the spatial distribution of water quality parameters.

7. The intelligent monitoring and early warning method for small urban water bodies based on visible light unmanned aerial vehicles as described in claim 1, characterized in that, The process of determining the warning conditions in step S7 is as follows: After data accumulation, it is determined whether the predicted average value of each water quality parameter of the monitored water body is in an abnormal water quality state. If the predicted average value of the water quality parameter of the monitored water body is normal, the warning step is skipped. If the predicted average value of the water quality parameters of the monitored water body is abnormal, then calculate the predicted trend change rate for the next seven days and the trend change rate for the past seven days. The formula for calculating the trend change rate is: In the formula, The rate of change of the trend; Indexed by the sequence number of the time point; For the predicted time range; For time sequence number; For the corresponding water quality parameter Daily average; The trend ratio is calculated using the following formula: In the formula, R is the trend ratio; Predict the rate of change of the trend over the next seven days; This represents the rate of change of the trend over the past seven days. If any one of the water quality parameters simultaneously satisfies the conditions of water quality abnormality and If the warning is triggered, then the corresponding water quality parameter warning procedure will be performed; otherwise, the warning procedure will be skipped.

Images