An invasive aquatic plant monitoring method, system, device and medium

Abstract

The application discloses an invasive aquatic plant monitoring method, system, device and medium, and relates to the field of biological safety invasive aquatic plant monitoring. The method comprises the following steps: acquiring a multispectral image of a water area to be measured; the multispectral image is obtained by using a multispectral imager carried by a drone; determining a segmentation result of a pixel point in the multispectral image of the water area to be measured according to the multispectral image of the water area to be measured and a classification model; the classification model is obtained by training a random forest model; the segmentation result of the pixel point in the multispectral image comprises a pixel point of a water surface and a pixel point of an invasive aquatic plant of hydrilla verticillata; generating an invasive damage level of the hydrilla verticillata in the water area to be measured according to the segmentation result of the pixel point in the multispectral image of the water area to be measured; and alarming according to the invasive damage level. The application does not require investigators to enter the water area, improves monitoring accuracy and reduces monitoring cost.

Description

Technical Field

[0001] This application relates to the field of biosafety monitoring of invasive aquatic plants, and in particular to a method, system, equipment and medium for monitoring invasive aquatic plants. Background Technology

[0002] *Pistia stratiotes* L., an invasive aquatic plant, is the only species in the genus *Pistia* of the family Aracoae. It is a perennial, floating aquatic herbaceous plant. In the 1950s, it was widely cultivated in southern China as pig feed. The lack of effective management strategies after its introduction led to its proliferation, seriously threatening biodiversity throughout southern China. With increasing emphasis on ecological security, *Pistia stratiotes* has been included in the list of key invasive alien species for management. *Pistia stratiotes* spreads and reproduces rapidly, exhibits strong environmental adaptability, and often forms dense communities after adaptation, thus encroaching on the growth environment of other organisms and making it impossible for other aquatic plants to survive, seriously impacting the stability of aquatic ecosystems.

[0003] Currently, *Isatis tinctoria* is widely distributed, especially in areas with dense water networks. In recent years, the effects of eutrophication and climate change have accelerated its spread, and its distribution range shows a northward expansion trend. Monitoring *Isatis tinctoria* is of great significance for protecting ecological balance, water environment management, protecting agriculture and aquaculture, and accumulating scientific research data. Because *Isatis tinctoria* mainly grows on the water surface and has a floating characteristic, traditional manual survey techniques require boat-based quadrat surveys, which are difficult, time-consuming, and labor-intensive, and the personal safety of survey personnel cannot be guaranteed. High-altitude remote sensing and satellite remote sensing images have insufficient resolution to identify specific invasive plants in the images, resulting in low monitoring accuracy and high monitoring costs.

[0004] Therefore, there is an urgent need to develop a new monitoring technology and alarm system to monitor the level of harm caused by invasive aquatic plants in water areas and guide the prevention and control of invasive aquatic plants in water areas. Summary of the Invention

[0005] The purpose of this application is to provide a method, system, equipment, and medium for monitoring invasive aquatic plants, enabling monitoring personnel to monitor and issue alarms for invasive aquatic plants in water bodies without entering the water, thereby improving monitoring accuracy and reducing monitoring costs.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] Firstly, this application provides a method for monitoring invasive aquatic plants, including:

[0008] Acquire multispectral images of the water body to be measured; the multispectral images are obtained by taking pictures using a multispectral imager mounted on a UAV;

[0009] Based on the multispectral image of the water area to be tested and the classification model, the segmentation results of the pixels in the multispectral image of the water area to be tested are determined; the classification model is obtained by training a random forest model; the segmentation results of the pixels in the multispectral image include: pixels on the water surface and pixels of *Isatis tinctoria* among invasive aquatic plants;

[0010] The invasion hazard level of *Ichthyophthirius multifiliis* in the water body is generated based on the pixel segmentation results in the multispectral image of the water body to be tested.

[0011] An alarm will be triggered based on the level of intrusion hazard.

[0012] Optionally, based on the multispectral image of the water body to be measured and the classification model, the segmentation results of the pixels in the multispectral image of the water body to be measured are determined, specifically including:

[0013] Feature data of the multispectral image of the water body to be measured is extracted; the feature data includes: original spectral features, vegetation index features, and texture features; the original spectral features include: red band features, green band features, blue band features, and near-infrared band features;

[0014] The feature data of the multispectral image of the water body to be tested is input into the classification model to obtain the segmentation results of the pixels in the multispectral image of the water body to be tested.

[0015] Optionally, the method for determining the classification model specifically includes:

[0016] Acquire sample data; the sample data includes: a multispectral image of the training water area and the segmentation results of pixels in the multispectral image of the training water area;

[0017] Extract feature data from multispectral images of the water area used for training;

[0018] The feature data of the multispectral images of the training water area and the segmentation results of the pixels in the multispectral images of the training water area are determined as the classification sample set;

[0019] The classification sample set is divided into a test set and a training set;

[0020] The random forest model is trained using the training set mentioned above;

[0021] The trained random forest model is tested using the test set to obtain test results; the test results are the accuracy of the output results of the trained random forest model.

[0022] If the test results reach the set overall classification accuracy, the trained random forest model will be determined as the classification model.

[0023] If the test results do not reach the set overall classification accuracy, the number of training sets is increased, and the random forest model is retrained until the test results reach the set overall classification accuracy, thus obtaining a classification model.

[0024] Optionally, feature data from the multispectral images of the training water bodies are extracted, specifically including:

[0025] The normalized water index is calculated based on the multispectral image of the training water area, and the multispectral image of the training water area is masked based on the normalized water index to obtain the masked multispectral image of the training water area.

[0026] A superpixel segmentation algorithm is used to perform superpixel segmentation on the multispectral image after the water area mask is used for training, generating superpixel clusters; the superpixel clusters have spectral consistency.

[0027] The superpixel clusters are classified and labeled to obtain a labeled dataset;

[0028] The labeled dataset is statistically averaged to obtain the feature data of the multispectral images of the water area used for training.

[0029] Optionally, the superpixel segmentation algorithm is a simple non-iterative algorithm.

[0030] Optionally, the invasive hazard level of *Ichthyophthirius multifiliis* in the water body is generated based on the pixel segmentation results in the multispectral image of the water body to be tested, specifically including:

[0031] The coverage of invasive aquatic plants is generated based on the segmentation results of pixels in the multispectral image of the water area to be tested.

[0032] Invasive hazard levels are generated based on the coverage of invasive aquatic plants.

[0033] Optionally, the expression for the invasive aquatic plant cover is:

[0034] A = P i / (P i +P a +P w );

[0035] Where A represents the coverage of the water area invaded by *Pyrrosia lingua*, and P represents the coverage of the water area in the overall water body. i P represents the number of pixels labeled as *Isodon spp.*, an invasive aquatic plant, in the multispectral image of the water area under test. a P represents the number of pixels in the multispectral image of the water body being tested that are labeled as aquatic plants other than *Pyrrosia lingua*. wThis indicates the number of pixels marked as water surface in the multispectral image of the water area to be measured.

[0036] Secondly, this application provides a monitoring system for invasive aquatic plants, comprising:

[0037] The system comprises an image acquisition platform, a computing platform, and a monitoring and alarm platform connected in sequence. The image acquisition platform includes a drone and a multispectral imager. The drone carries the multispectral imager, which is used to capture multispectral images of the water area under test.

[0038] The computing platform includes:

[0039] The image acquisition module is used to acquire multispectral images of the water body to be measured.

[0040] The model determination module is used to determine the classification model; the classification model is obtained by training a random forest model.

[0041] The segmentation result determination module is used to determine the segmentation result of pixels in the multispectral image of the water body to be tested based on the multispectral image of the water body to be tested and the classification model; the segmentation result of pixels in the multispectral image includes: pixels on the water surface and pixels of *Pistia stratiotes* among invasive aquatic plants;

[0042] The monitoring and alarm platform includes:

[0043] The invasion hazard level generation module is used to generate the invasion hazard level of the water hyacinth in the water body under test based on the segmentation results of the pixels in the multispectral image of the water body under test;

[0044] An alarm module is used to issue an alarm based on the level of intrusion hazard.

[0045] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the invasive aquatic plant monitoring method described in any one of the above.

[0046] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the invasive aquatic plant monitoring method described above.

[0047] According to the specific embodiments provided in this application, this application has the following technical effects:

[0048] This application provides a method, system, device, and medium for monitoring invasive aquatic plants. It uses drones to capture images of the area to be monitored, eliminating the need for investigators to enter the water surface and ensuring their safety. The multispectral images captured by the drone, equipped with a multispectral imager, can more accurately identify invasive aquatic plants, reducing the high monitoring costs caused by the low resolution of remote sensing images. Furthermore, it uses machine learning algorithms to efficiently and accurately identify invasive aquatic plants, other aquatic plants, and water area, resulting in more accurate monitoring data and enabling timely prevention of invasive aquatic plant outbreaks. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating the invasive aquatic plant monitoring method provided in this application;

[0051] Figure 2 This is a schematic diagram of the invasive aquatic plant monitoring system provided in this application;

[0052] Figure 3 This is a schematic diagram of the structure of a computer device provided in this application. Detailed Implementation

[0053] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0055] In one exemplary embodiment, such as Figure 1 As shown, a method for monitoring invasive aquatic plants is provided, including:

[0056] Acquire multispectral images of the water body to be measured; the multispectral images are obtained by taking pictures using a multispectral imager mounted on a drone.

[0057] Based on the multispectral image of the water body to be tested and the classification model, the pixel segmentation results in the multispectral image of the water body to be tested are determined; the classification model is obtained by training a random forest model; the pixel segmentation results in the multispectral image include: pixels on the water surface and pixels of *Isodon spp.*, an invasive aquatic plant.

[0058] The invasion hazard level of water hyacinth in the water body is generated based on the pixel segmentation results in the multispectral image of the water body to be tested.

[0059] An alarm will be triggered based on the level of intrusion hazard.

[0060] In another exemplary embodiment of this application, the segmentation result of pixels in the multispectral image of the water body to be measured is determined based on the multispectral image of the water body to be measured and the classification model, specifically including:

[0061] Feature data from the multispectral image of the water body to be measured is extracted. This feature data includes: raw spectral features, vegetation index features, and texture features. Raw spectral features include: red band features, green band features, blue band features, and near-infrared band features. Vegetation index features include: Normalized Difference Vegetation Index (NDVI), Normalized Green Difference Vegetation Index (GNDVI), Enhanced Vegetation Index (EVI), Visual Greenness Index (VARI), Ratio Vegetation Index (RVI), Red-Green Ratio Index (RGRI), Normalized Difference Blue-Green Index (NDBGI), and Normalized Difference Vegetation Index with Red Edge (NDRE). Texture features include mean, standard deviation, skewness, kurtosis, contrast, energy, entropy, correlation, and evenness. Specifically, the normalized water index is calculated based on the green band features and near-infrared band features, and vegetation index features and texture features are calculated based on the raw spectral features.

[0062] The feature data of the multispectral image of the water body to be tested is input into the classification model to obtain the segmentation results of the pixels in the multispectral image of the water body to be tested.

[0063] In another exemplary embodiment of this application, the method for determining the classification model specifically includes:

[0064] Acquire sample data; the sample data includes: multispectral images of the training water area and the segmentation results of pixels in the multispectral images of the training water area.

[0065] Extract feature data from multispectral images of the water area used for training.

[0066] The feature data of the multispectral images of the training water area and the segmentation results of the pixels in the multispectral images of the training water area are determined as the classification sample set.

[0067] The classification sample set is divided into a test set and a training set; the number of pixels for each of the following in the training set is at least 200: water hyacinth, other aquatic plants, water surface, and background. 80% of the samples in the classification sample set are used as the training set, and the remaining 20% ​​are used as the validation set.

[0068] The random forest model is trained using the training set.

[0069] The trained random forest model is tested using the test set to obtain the test results; the test results represent the accuracy of the output results of the trained random forest model.

[0070] If the test results reach the set overall classification accuracy, the trained random forest model will be determined as the classification model.

[0071] If the test results do not reach the set overall classification accuracy, increase the number of training sets and retrain the random forest model until the test results reach the set overall classification accuracy to obtain the classification model.

[0072] In another exemplary embodiment of this application, extracting feature data from a multispectral image of the training water area specifically includes:

[0073] The normalized water index is calculated based on the multispectral image of the training water area, and then a mask is removed from the multispectral image of the training water area based on the normalized water index to obtain a masked multispectral image of the training water area. The normalized water index is a remote sensing index used to enhance the detection and identification of water bodies. The threshold for extracting water bodies using this index is 0; that is, when the index is greater than 0, the target object is identified as a water body.

[0074] A superpixel segmentation algorithm is used to segment the multispectral image after masking the water area used for training, generating superpixel clusters; the superpixel clusters have spectral consistency. Specifically, a superpixel cluster is a small region unit with spectral consistency.

[0075] Superpixel clusters were classified and labeled to obtain a labeled dataset. The labeled dataset contains superpixel clusters of three types of land cover elements: invasive aquatic plant *Pistachio spp.*, other aquatic plants, and background.

[0076] Statistical averaging is performed on the labeled dataset to obtain feature data from the multispectral images of the training water bodies. Specifically, the original spectral features, vegetation index features, and texture features within each cluster of the labeled dataset are statistically averaged to assign characteristic attributes to these clusters. These clusters containing characteristic attributes are then used as the classification sample set.

[0077] This application inputs the training set into a random forest machine learning classifier to construct a trained random forest model for image classification, identifying the target invasive plant *Isatis tinctoria*, other aquatic plants, and the background. The trained random forest model is then tested using a test set. If the test results meet expectations, the model can be used to effectively classify and extract the target invasive plant *Isatis tinctoria*. If the test results do not meet the standards, the superpixel segmentation parameters and the number of samples are adjusted, and the model is retrained until the accuracy of the model's classification on the test set is greater than or equal to the expected standard. The resulting new model is then used for the analysis of UAV multispectral images of this water area.

[0078] In another exemplary embodiment of this application, the superpixel segmentation algorithm is a simple non-iterative algorithm.

[0079] In another exemplary embodiment of this application, the invasive hazard level of *Ichthyophthirius multifiliis* in the water body to be tested is generated based on the segmentation results of pixels in the multispectral image of the water body to be tested, specifically including:

[0080] The coverage (area percentage) of invasive aquatic plants is generated based on the segmentation results of pixels in the multispectral image of the water area to be tested.

[0081] Invasive hazard levels are generated based on the coverage of invasive aquatic plants.

[0082] In another exemplary embodiment of this application, the expression for the coverage of invasive aquatic plants is:

[0083] A = P i / (P i +P a +P w ).

[0084] Where A represents the coverage (ratio) of the water area invaded by *Paspalum notatum* among invasive aquatic plants relative to the total water area, and P... i P represents the number of pixels labeled as *Isodon spp.*, an invasive aquatic plant, in the multispectral image of the water area under test. a P represents the number of pixels in the multispectral image of the water body being tested that are labeled as aquatic plants other than *Pyrrosia lingua*. w This indicates the number of pixels marked as water surface in the multispectral image of the water area to be measured.

[0085] This application relates to invasive plant monitoring and alarm systems, specifically disclosing a monitoring and alarm system for the invasive aquatic plant *Isatis tinctoria*. This application utilizes a drone equipped with a multispectral imager to capture images of the area to be monitored. Based on superpixel segmentation and random forest machine learning algorithms, it efficiently and accurately identifies the invasive aquatic plant *Isatis tinctoria*, other aquatic plants, and water area. The invasion level is assessed based on the proportion of *Isatis tinctoria* on the water surface. Alarms are issued via email or SMS regarding the extent of *Isatis tinctoria* invasion, enabling timely intervention to prevent outbreaks.

[0086] Based on the same inventive concept, this application also provides an invasive aquatic plant monitoring system for implementing the aforementioned invasive aquatic plant monitoring method. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more embodiments of the invasive aquatic plant monitoring system provided below can be found in the limitations of the invasive aquatic plant monitoring method described above, and will not be repeated here.

[0087] In one exemplary embodiment, such as Figure 2 As shown, a monitoring system for invasive aquatic plants is provided, comprising:

[0088] The system consists of an image acquisition platform, a computing platform, and a monitoring and alarm platform connected sequentially. The image acquisition platform includes a drone and a multispectral imager. The drone carries the multispectral imager, which is used to capture multispectral images of the water area under test. The computing platform is typically a smartphone. The drone, connected to the multispectral imager, captures multispectral images of the water area from the air, ensuring that both the forward and lateral overlap are no less than 60%, and records information such as GPS coordinates and image capture time.

[0089] The computing platform includes:

[0090] The image acquisition module is used to acquire multispectral images of the water body to be measured.

[0091] The model determination module is used to determine the classification model; the classification model is obtained by training the random forest model.

[0092] The segmentation result determination module is used to determine the segmentation results of pixels in the multispectral image of the water body to be tested based on the multispectral image and classification model of the water body to be tested. The segmentation results of pixels in the multispectral image include: pixels on the water surface and pixels of *Isodon spp.*, an invasive aquatic plant.

[0093] The monitoring and alarm platform includes:

[0094] The invasion hazard level generation module is used to generate the invasion hazard level of water hyacinth in the water body under test based on the segmentation results of pixels in the multispectral image of the water body under test.

[0095] The alarm module is used to issue alarms based on the level of intrusion hazard.

[0096] According to the agricultural industry standard "Technical Specification for Monitoring Invasive Alien Plants (NY / T 3076-2017)", the invasion damage of Iris pseuda is divided into three levels: mild occurrence (coverage <5%), moderate occurrence (coverage 5% to 20%), and severe occurrence (coverage >20%). When the invasion damage level is severe, an alarm will be sent to the management personnel via SMS or email.

[0097] As an optional implementation, the invasive aquatic plant monitoring system also includes: a (ground) control platform; the (ground) control platform is connected to both the image acquisition platform and the computing platform; the drone can connect to the (ground) control platform via WiFi, and the (ground) control platform can connect to the computing platform via Bluetooth or WiFi.

[0098] The ground control platform is used to control the flight of the UAV, the multispectral imager to capture multispectral images of the water area to be measured, and to receive multispectral images of the water area to be measured.

[0099] As an alternative implementation, the drone is also used to record GPS positioning coordinates and image capture time.

[0100] The computing platform also includes:

[0101] The distribution heatmap generation unit is used to generate a distribution heatmap of water hyacinth based on the segmentation results of pixels in the multispectral image of the water body to be measured.

[0102] The transmitting unit is used to send GPS positioning coordinates, image capture time, pixel segmentation results in the multispectral image of the water area to be measured, multispectral image of the water area to be measured, coverage of invasive aquatic plants, and heat map of the distribution of *Ipomoea aquatica* to the monitoring and alarm platform.

[0103] The image determination module is specifically used to stitch together the multispectral images of the water body under test to obtain a complete multispectral image of the water body under test.

[0104] The monitoring and alarm platform is located on a cloud server. Managers can manually review the information sent by the computing platform. For example, if the hazard level of pied lees is severe or above, the received severe occurrence and severe occurrence data can be used to send SMS or email alarms to relevant managers via the network.

[0105] The computing platform uses the trained model to analyze the new multispectral images and sends the multispectral images, distribution heatmaps, GPS location coordinates, image capture time, bulrush coverage, and bulrush intrusion hazard level to the monitoring and alarm platform located in the cloud.

[0106] The beneficial effects of this application are:

[0107] 1. This application provides high-resolution multispectral images containing more image information in addition to visible light, which can more accurately identify the invasive aquatic plant *Pistachio spp.*

[0108] 2. This application uses a random forest model for machine learning, which produces a smaller model that does not require large-scale computing power and can achieve local multispectral image recognition and segmentation using a smartphone.

[0109] 3. This application can calculate the hazard level of water bodies invaded by water hyacinth and report it to the monitoring and alarm platform, which facilitates the monitoring and management of water hyacinth invasion.

[0110] 4. The monitoring system of this application does not require investigators to enter the water surface for testing, ensuring the personal safety of investigators and providing more accurate monitoring data.

[0111] The implementation of the invasive aquatic plant monitoring in this application requires the following steps:

[0112] Set a flight path for the drone (DJI Mavic 3M) at an altitude of at least 20 meters, with both directional and lateral overlap at least 60%. The multispectral imager should include at least four bands: red, green, blue, and near-infrared. Based on the planned flight path, hover over the water area to capture a large number of multispectral images of the water, and record GPS coordinates, image capture time, and other information. Ensure the ground sampling distance resolution of the captured images is at least better than 10cm.

[0113] Image stitching: Acquiring high-resolution multispectral imagery covering the entire water area. First, keypoint detection is performed between images using the scale-invariant feature transform principle. For each detected keypoint, a feature vector is generated. By calculating the Euclidean distance between keypoints in the images, similar keypoint pairs are found. Based on the matched keypoint pairs, the geometric transformation matrix between the two images is estimated. The calculated geometric transformation matrix is ​​then used to project one image onto the coordinate system of the other image, achieving image alignment. Overlapping areas are then fused (e.g., using weighted averaging, smooth transition) to eliminate seams. Finally, a complete image of the entire water area is obtained.

[0114] Based on the original green and near-infrared bands of multispectral imagery, a normalized water index is calculated using the formula: (green band - near-infrared band) / (green band + near-infrared band). A threshold of 0 is set, and areas exceeding this threshold are considered water bodies. Water bodies are then masked for removal.

[0115] Vegetation index features and texture features were calculated based on the original bands. The calculated vegetation index features include eight types: Normalized Difference Vegetation Index (NDVI), Normalized Green Vegetation Index (NDR), Enhanced Vegetation Index (EGI), Visual Greenness Index (VLE), Ratio Vegetation Index (RRI), Red-Green Ratio Index (RRI), Normalized Difference Blue-Green Index (NDBL), and Red-Edge Normalized Vegetation Index (REDI). Texture features include nine types: Mean, Standard Deviation, Skewness, Kurtosis, Entropy, Contrast, Energy, Correlation, and Evenness. Because texture features were extracted separately for each original band, a total of 36 features (4*9) were generated from the four bands.

[0116] Superpixel segmentation is performed on the image behind the water mask to generate superpixel clusters with spectral consistency. Cluster boundary information is obtained. The original bands are segmented using a Simple Non-Iterative Algorithm (SNIC), and a unified cluster center is derived based on spatial heterogeneity and spatial autocorrelation. Specifically, initial cluster centers are first defined, i.e., seed initialization, and pixels are assigned to the nearest cluster center based on color similarity and spatial proximity. Then, the seed positions are refined iteratively by calculating the average color and spatial coordinates of the pixels assigned to each seed. The parameters involved in superpixel segmentation include superpixel size, compactness, connectivity, and neighborhood size, which are set to 50, 10, 8, and 15, respectively.

[0117] For superpixel clusters, classification and annotation are performed. At least 500 labeled dataset points containing three types of land cover elements—invasive aquatic plant *Potamogeton crispus*, other aquatic plants, and background—are randomly selected through visual interpretation and saved as point vector files.

[0118] The feature information of each cluster is obtained from the labeled dataset. Specifically, the original spectral features (4), vegetation index features (10), and texture features (36) within each cluster are statistically averaged, and the average value of these features is assigned as the feature attribute of these clusters. These clusters containing feature attributes are used as the classification sample set. 80% of the samples are used as the training set, and the remaining 20% ​​are used as the validation set.

[0119] The evaluation metric for classification accuracy is the overall classification accuracy (OA), calculated using the following formula:

[0120]

[0121] n ii represents the number of correctly classified samples (diagonal elements) of class i in the confusion matrix. k is the total number of classes. N is the total number of samples (the sum of all elements in the confusion matrix).

[0122] A random forest machine learning classifier was used as the classification model. Training samples were input into the random forest classifier to build a classification model that classifies the entire image, identifying the target invasive plant *Physalis alkekengi*, other aquatic plants, and the background. Training model parameters: number of decision trees ntree = 500, maximum number of nodes per tree maxnodes = 50, and feature sampling parameter mtry was set to the square root of the number of features (4 + 10 + 36).

[0123] After setting the model parameters, the trained model is tested using a test set. If the test results meet expectations (overall classification accuracy OA ≥ 75%), the model can be used to effectively classify and extract the target invasive plant, *Isatis tinctoria*. If the test results do not meet the standards, the superpixel segmentation parameters and the number of samples are adjusted, and the model is retrained until the accuracy of the model's classification on the test set is greater than or equal to the expected standard. The resulting new model is then used for the analysis of UAV multispectral images of this water area.

[0124] Based on the classification results, a heat map of the distribution of water hyacinth is generated, and the coverage (area percentage) of water hyacinth invasion in water areas is calculated using the number of pixels.

[0125] The new multispectral images are analyzed using a computing platform (usually a smartphone) with a trained model, and the multispectral images, distribution heatmaps, GPS location coordinates, image capture time, porphyria coverage, and porphyria intrusion hazard level are sent to a monitoring and alarm platform located in the cloud.

[0126] Administrators manually review the information sent by the computing platform. Upon receiving data on severe occurrences and severe occurrences, they can send SMS or email alerts to administrators via the network.

[0127] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores monitoring data on invasive aquatic plants. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for monitoring invasive aquatic plants.

[0128] Those skilled in the art will understand that Figure 3 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0129] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0130] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0131] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0132] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0133] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0134] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. An invasive aquatic plant monitoring method characterized by, The invasive aquatic plant monitoring method comprises: acquiring a multispectral image of a water area to be measured; the multispectral image is obtained by using a multi-spectral imager carried by a drone to take pictures; determining a segmentation result of a pixel point in the multispectral image of the water area to be measured according to the multispectral image of the water area to be measured and a classification model; the classification model is obtained by training a random forest model; the segmentation result of the pixel point in the multispectral image comprises a pixel point of a water surface and a pixel point of an invasive aquatic plant of hydrilla verticillata; generating an invasive hazard level of hydrilla verticillata in the water area to be measured according to the segmentation result of the pixel point in the multispectral image of the water area to be measured; performing an alarm according to the invasive hazard level; the method for determining the classification model specifically comprises: acquiring sample data; the sample data comprises a multispectral image of a training water area and a segmentation result of a pixel point in the multispectral image of the training water area; extracting feature data of the multispectral image of the training water area; determining the feature data of the multispectral image of the training water area and the segmentation result of the pixel point in the multispectral image of the training water area as a classification sample set; dividing the classification sample set into a test set and a training set; training a random forest model by using the training set; testing the trained random forest model according to the test set to obtain a test result; the test result is an accuracy rate of an output result of the trained random forest model; if the test result reaches a set overall classification accuracy, the trained random forest model is determined as the classification model; if the test result does not reach the set overall classification accuracy, the number of the training set is increased, the random forest model is retrained until the test result reaches the set overall classification accuracy, and the classification model is obtained; the extraction of the feature data of the multispectral image of the training water area specifically comprises: calculating a normalized water body index based on the multispectral image of the training water area, and removing a mask based on the normalized water body index to obtain a multispectral image of the training water area after mask removal; performing superpixel segmentation on the multispectral image of the training water area after mask removal by using a superpixel segmentation algorithm to generate a superpixel cluster; the superpixel cluster has spectral consistency; classifying and labeling the superpixel cluster to obtain a labeled data set; statistically averaging the labeled data set to obtain the feature data of the multispectral image of the training water area.

2. The invasive aquatic plant monitoring method according to claim 1, characterized by, determining the segmentation result of the pixel point in the multispectral image of the water area to be measured according to the multispectral image of the water area to be measured and the classification model specifically comprises: extracting feature data of the multispectral image of the water area to be measured; the feature data comprises original spectral features, vegetation index features and texture features; the original spectral features comprise red waveband features, green waveband features, blue waveband features and near-infrared waveband features; inputting the feature data of the multispectral image of the water area to be measured into the classification model to obtain the segmentation result of the pixel point in the multispectral image of the water area to be measured.

3. The invasive aquatic plant monitoring method according to claim 1, characterized by, The superpixel segmentation algorithm is a simple non-iterative algorithm.

4. The invasive aquatic plant monitoring method according to claim 1, characterized by, generating the invasive hazard level of hydrilla verticillata in the water area to be measured according to the segmentation result of the pixel point in the multispectral image of the water area to be measured specifically comprises: Generate the invasive aquatic plant coverage according to the segmentation result of the pixel points in the multispectral image of the water area to be measured; Generate the invasive hazard level according to the invasive aquatic plant coverage.

5. The invasive aquatic plant monitoring method according to claim 4, characterized by, The expression of the invasive aquatic plant coverage is: A = P i (P i + P a + P w ); Wherein, A represents the coverage of the area invaded by the water plant Alternanthera philoxeroides in the overall water area, P i represents the number of pixel points marked as the water plant Alternanthera philoxeroides in the multispectral image of the water area to be tested, P a represents the number of pixel points marked as other water plants except Alternanthera philoxeroides in the multispectral image of the water area to be tested, P w represents the number of pixel points marked as the water surface in the multispectral image of the water area to be tested.

6. An invasive aquatic plant monitoring system characterized by, The invasive aquatic plant monitoring system comprises an image acquisition platform, a calculation platform and a monitoring and alarm platform connected in sequence; the image acquisition platform comprises a drone and a multispectral imager; the drone carries the multispectral imager; the multispectral imager is used for photographing the water area to be measured to obtain the multispectral image of the water area to be measured; The calculation platform comprises: An image acquisition module is configured to acquire the multispectral image of the water area to be measured; A model determination module is configured to determine a classification model; the classification model is obtained by training a random forest model; The determination method of the classification model specifically comprises: Acquiring sample data; the sample data comprises the multispectral image of a training water area and the segmentation result of the pixel points in the multispectral image of the training water area; Extracting feature data of the multispectral image of the training water area; Determining the feature data of the multispectral image of the training water area and the segmentation result of the pixel points in the multispectral image of the training water area as a classification sample set; Dividing the classification sample set into a test set and a training set; Training a random forest model by using the training set; Testing the trained random forest model according to the test set to obtain a test result; the test result is the accuracy rate of the output result of the trained random forest model; If the test result reaches a set overall classification accuracy, the trained random forest model is determined as the classification model; If the test result does not reach the set overall classification accuracy, the number of the training set is increased, the random forest model is retrained, and the classification model is obtained until the test result reaches the set overall classification accuracy; The extraction of the feature data of the multispectral image of the training water area specifically comprises: Calculating a normalized water body index based on the multispectral image of the training water area, and removing the mask based on the normalized water body index to obtain the multispectral image of the training water area after mask removal; Performing superpixel segmentation on the multispectral image of the training water area after mask removal by using a superpixel segmentation algorithm to generate a superpixel cluster; the superpixel cluster has spectral consistency; Classifying and labeling the superpixel cluster to obtain a labeled data set; Statistically averaging the labeled data set to obtain the feature data of the multispectral image of the training water area; A segmentation result determination module is configured to determine the segmentation result of the pixel points in the multispectral image of the water area to be measured according to the multispectral image of the water area to be measured and the classification model; the segmentation result of the pixel points in the multispectral image comprises the pixel points of the water surface and the pixel points of the invasive aquatic plant of Hygroryza aromaticus; The monitoring and alarm platform comprises: An invasive hazard level generation module is configured to generate the invasive hazard level of Hygroryza aromaticus in the water area to be measured according to the segmentation result of the pixel points in the multispectral image of the water area to be measured; An alarm module is configured to alarm according to the invasive hazard level.

7. A computer device comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that the processor executes the computer program to implement the invasive aquatic plant monitoring method of any one of claims 1-5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the invasive aquatic plant monitoring method of any one of claims 1-5.

Images