Microplastic polluted water area spatial distribution identification method based on remote sensing

Abstract

The invention relates to a micro-plastic polluted water area space distribution identification method based on remote sensing, particularly relates to the technical field of environment monitoring, and solves the problems that various characteristic data of micro-plastic are difficult to analyze through image processing software, and the characteristics are difficult to comprehensively construct a model to output a water area comprehensive pollution degree value. The method comprises the following steps: extracting 70% of sample data from a target water area, and preprocessing the sample data; screening out samples with micro-plastic abundance not smaller than the abundance average value of the micro-plastic in the target water area; extracting feature data from the remote sensing image, wherein the feature data comprises a total color adsorption capability degree, a total shape coverage capability degree and a total water turbidity degree; constructing a model by using the feature data to obtain a water area comprehensive pollution degree predicted value; spatial distribution of comprehensive pollution degree predicted values is generated through the GIS technology, and different colors are used for representing different pollution degrees.

Description

Technical Field

[0001] The present invention belongs to the technical field of environmental monitoring, and in particular relates to a method for identifying the spatial distribution of microplastic-polluted waters based on remote sensing. Background Art

[0002] With the continuous increase in global plastic production, microplastics have become one of the main sources of water pollution. For this reason, a method for identifying the spatial distribution of microplastic-polluted waters based on remote sensing has emerged. The present invention constructs a comprehensive water pollution degree prediction model by extracting a variety of characteristic data through comprehensive utilization of remote sensing technology, neural network models and multi-dimensional data analysis methods, thereby achieving efficient identification of the distribution of microplastic pollution in water bodies. The model prediction value point data is converted into surface data and then imported into GIS software to generate the spatial distribution of pollution degree prediction values.

[0003] Although the existing methods for identifying the spatial distribution of microplastic-contaminated waters have achieved the identification of the spatial distribution of polluted waters to a certain extent, the traditional water quality monitoring method requires sampling one by one, which is inefficient. It is difficult to extract features from microplastic images through image processing software, and to construct corresponding feature degree data for analysis based on the features. It is also difficult to comprehensively construct a model based on the feature degree data to predict the comprehensive pollution degree value of the water area. There is a lack of importing the comprehensive pollution degree value of the water area into GIS software to obtain the spatial distribution, and using different colors to represent different pollution degrees. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a method for identifying the spatial distribution of microplastic polluted waters based on remote sensing, which is used to solve the following technical problems: Although the existing methods for identifying the spatial distribution of microplastic-contaminated waters have achieved the identification of the spatial distribution of polluted waters to a certain extent, the traditional water quality monitoring method requires sampling one by one, which is inefficient. It is difficult to extract features from microplastic images through image processing software, and to construct corresponding feature degree data for analysis based on the features. It is also difficult to comprehensively construct a model based on the feature degree data to predict the comprehensive pollution degree value of the water area. There is a lack of importing the comprehensive pollution degree value of the water area into GIS software to obtain the spatial distribution, and using different colors to represent different pollution degrees.

[0005] To solve the above problems, the first aspect of the present invention provides a method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing, comprising the following steps: S1: Collect several water samples in the target waters and extract 70% of the sample data for preprocessing; S2: Detect the pre-treated water samples and screen out water samples whose microplastic abundance is greater than or equal to the average abundance of microplastics in the target waters; S3: remote sensing technology is used to obtain remote sensing images of the screened water samples, and the boundaries of the water bodies are extracted by the Canny edge detection algorithm and then preprocessed to extract and identify feature data from the remote sensing images, including the total color adsorption capacity, the total shape coverage capacity and the total water turbidity; S4: Use the collected feature data to build a model and train it, deploy the trained model to the actual environment, extract features from new images and use the model to predict, and obtain the predicted value of the comprehensive pollution degree of the water area; S5: Use GIS technology to visualize the spatial distribution of the predicted values ​​of the comprehensive pollution level of water bodies obtained by the model on the map. According to the predicted values ​​of the comprehensive pollution level of water bodies and different colors representing different pollution levels, use GIS tools to analyze the spatial distribution of microplastic water pollution and identify hot spots.

[0006] As a further solution of the present invention: Step S1 comprises the following steps: According to the preset sampling point coordinates, the width of the target water area is measured, and several sampling points are arranged at equal distances along the width of the water area. The sampling depth is between 0.5 and 1 meter of the depth of the target water area. Several water samples are collected in the target water area, and 70% of the collected water samples are randomly selected as processing objects, and pre-processing operations are performed on the selected processing objects.

[0007] As a further solution of the present invention: the preprocessing operation on the extracted processing object includes the following steps: The extracted treatment objects were subjected to pretreatment operations, including adding 30% hydrogen peroxide with a volume ratio of 10:1 to the samples, and digesting them at a constant temperature of 60°C for 12 hours; mixing the digested water samples with a high-density saturated sodium chloride salt solution for density separation, stirring them evenly and letting them stand for 24 hours to collect the upper solution; filtering the upper solution, and placing the obtained filter paper in a culture dish and drying it at 30°C for 24 hours.

[0008] As a further solution of the present invention: the step S2 comprises the following steps: The pretreated water samples were tested for microplastics using a microscope visual identification method, and the number, color, size and shape characteristics of the microplastics in each sample were recorded. Based on the test data, the average abundance of microplastics in all samples in the target waters was calculated, and the abundance of microplastics in each sample was compared with the average abundance of the target waters. Water samples with microplastic abundance greater than or equal to the average abundance of microplastics in the target waters were screened out and analyzed.

[0009] As a further solution of the present invention: Step S3 comprises the following steps: Remote sensing technology is used to obtain remote sensing images of screened water samples. The boundaries of the water bodies are extracted using the Canny edge detection algorithm and then preprocessed, including denoising and image enhancement. Feature data are extracted and identified from the remote sensing images, including the total color adsorption capacity, the total shape coverage capacity and the total water turbidity.

[0010] As a further solution of the present invention: the total color adsorption capacity comprises the following steps: Collect image datasets containing microplastics and other objects. The model uses a convolutional neural network to extract image features, including color, size, and shape. The Mask R-CNN mask region convolutional neural network uses a region proposal network to generate potential microplastic locations, classify and segment the locations, assign a category label to each potential microplastic location, and generate a corresponding mask. Use the mouse to draw a polygon to mark the edge of the target object, use the annotation tool to annotate the microplastics and other objects in the image, and save it as a .json file. Convert the .json format to a format that matches the Mask R-CNN mask region convolutional neural network training. The converted data package includes image files and corresponding annotation files. Use the converted data to train the Mask R-CNN mask region convolutional neural network, and use the trained model to distinguish microplastics from other objects in the image. According to the color characteristics, the microplastics are divided into three categories, including black, blue and colorless microplastics. The images containing black, blue and colorless microplastics are annotated, and the annotated data set is used to train the classifier to automatically identify and classify microplastics. The object technology method is used to count the microplastics of each color category, and the proportion of the number of microplastics of each color to the total number of microplastics is calculated. The proportion value is used as the weight coefficient. According to the adsorption intensity coefficient set for microplastics of different colors based on the research data, the weight coefficient of each color of microplastics is combined with the adsorption intensity coefficient of the corresponding color, and the combined values ​​of the three colors are added to obtain the total color adsorption capacity of the water area; The total color adsorption capacity calculation formula is: in, is the degree of total color adsorption capacity, The amount of black microplastics, The amount of blue microplastics, is the amount of colorless microplastics, is the total amount of microplastics of three colors, , and is the set adsorption intensity coefficient.

[0011] As a further solution of the present invention: the overall shape coverage capability degree comprises the following steps: The collected images are converted into grayscale images, and the pattern recognition algorithm in the image analysis software is used to identify and classify the shapes of microplastics. The edge detection algorithm detects the changes in pixel intensity in the image to determine the edge position, extracts the edges of microplastics from the image, and uses the contour tracking algorithm to connect the edge points to form the object contour; a Fourier transform infrared microscope based on a focal plane array is used to image the microplastic particles to obtain high-resolution two-dimensional images, and laser ranging is used to obtain the depth information of the microplastics. The depth information is combined with the two-dimensional image, and a three-dimensional reconstruction algorithm is used to generate a three-dimensional model of the microplastics. The surface of each microplastic is obtained through the three-dimensional model. The data of microplastics were collected by 3D model. The microplastics were divided into different shape types, including fiber, granular and film, and the number of microplastics in each shape category was counted. The surface area of ​​microplastics of different shapes was calculated. The surface area of ​​microplastics of each shape was multiplied by the corresponding number to obtain the total surface area of ​​the corresponding shape. The ratio of the total surface area of ​​each shape to the combined surface area of ​​the three shapes was calculated. The ratio was used as the weight coefficient to set the coverage breadth coefficient for each shape of microplastics. After multiplying the weight coefficient by the coverage breadth coefficient, the combined values ​​of the three shapes were added to obtain the total shape coverage capacity of the water area. The total shape coverage degree calculation formula is: in, is the combined surface area of ​​the three shapes, is the amount of fibrous microplastics, is the surface area of ​​fibrous microplastics, is the amount of granular microplastics, is the surface area of ​​granular microplastics, is the amount of film-like microplastics, is the surface area of ​​film-like microplastics, is the total shape coverage capability, , and is the set coverage breadth coefficient.

[0012] As a further solution of the present invention: the total water turbidity degree comprises the following steps: Import the collected images into ImageJ open source software, set the ruler length and unit in the software, pre-process the images, including adjusting the contrast and brightness, use the Rectangle tool in the toolbar to manually mark the microplastic particles in the image, and for each marked particle, use the function provided by the software to draw a straight line through the center of the particle, calculate and record the maximum size of each particle, and classify the microplastics according to the measured particle size, including nano-scale microplastics, micro-scale microplastics, and millimeter-scale microplastics. Collect the particle size data of all particles, perform statistical analysis to obtain the average particle size of the microplastics, count the microplastics in each particle size range, count the number of microplastics in each particle size range, and jointly establish the shape ratios of three different shapes of microplastics to construct a weighted average particle size. According to the average particle size of each shape of microplastics and its ratio, the total water turbidity of the water area is obtained; The calculation formula of total water turbidity is: in, is the overall water turbidity level, is the particle size and For the The particle size of fibrous microplastics is is the amount of fibrous microplastics, is the total number of microplastics in three shapes, is the particle size and For the The particle size of microplastic particles is is the amount of granular microplastics, is the particle size and For the The particle size of film-like microplastics is is the amount of film-like microplastics.

[0013] As a further solution of the present invention: the step S4 comprises the following steps: Based on the feature data extracted from the image, including the total color adsorption capacity, the total shape covering capacity and the total water turbidity, a machine learning model that can predict the comprehensive pollution level of the water area is constructed after cleaning and standardization operations. The model is trained using the collected feature data, and the trained model is deployed in the actual environment. Features are extracted from new images and predicted using the model to obtain the predicted value of the comprehensive pollution level of the water area.

[0014] As a further solution of the present invention: the step S5 comprises the following steps: The model prediction value point data is converted into surface data and imported into GIS software to generate the spatial distribution of the pollution degree prediction value; the water area comprehensive pollution degree prediction value is set as low pollution when ≤1, medium pollution when 1<water area comprehensive pollution degree prediction value ≤2, and heavy pollution when 2<water area comprehensive pollution degree prediction value. Different colors are used to represent different pollution degrees, among which green represents low pollution degree, yellow represents medium pollution degree, and red represents high pollution degree. The mapping function of GIS software is used to add annotations and legends; the spatial distribution of microplastic water pollution is analyzed and hot spots are identified based on GIS tools.

[0015] Beneficial effects of the present invention: The present invention uses remote sensing technology to obtain remote sensing images of screened water samples, extracts features of microplastics from samples to construct a neural network model that can distinguish microplastics from other objects, combines multiple features of microplastics with corresponding quantities and coefficients to obtain feature data of the total color adsorption capacity, total shape coverage capacity and total water turbidity of the water area, uses hyperspectral remote sensing technology and machine learning algorithms to improve the efficiency and accuracy of identifying microplastics, and realizes large-scale monitoring of microplastic pollution in water bodies. The multiple feature data obtained by comprehensive analysis of multiple features of microplastics provide a scientific basis for comprehensive assessment of water pollution conditions. The present invention extracts characteristic data from the image, including the total color adsorption capacity, the total shape coverage capacity and the total water turbidity, and pre-processes the characteristic data to construct a machine learning model that can predict the comprehensive pollution degree of the water area and train it. The trained model is used to predict the data in the new image to obtain the predicted value of the comprehensive pollution degree of the water area, thereby achieving real-time acquisition of water quality monitoring data and improving the efficiency and accuracy of water quality monitoring. The present invention uses GIS technology to visualize the spatial distribution of the predicted values ​​of the comprehensive pollution degree of water areas obtained by the model on a map, and uses different colors to represent different pollution degrees according to the threshold range of the predicted values ​​of the comprehensive pollution degree of water areas, thereby achieving more accurate identification of pollution areas and spatial distribution of different degrees. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] See also Figure 1 As shown, the present invention is a method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing, comprising the following steps: S1: Collect several water samples in the target waters and extract 70% of the sample data for preprocessing; S2: Detect the pre-treated water samples and screen out water samples whose microplastic abundance is greater than or equal to the average abundance of microplastics in the target waters; S3: remote sensing technology is used to obtain remote sensing images of the screened water samples, and the boundaries of the water bodies are extracted by the Canny edge detection algorithm and then preprocessed to extract and identify feature data from the remote sensing images, including the total color adsorption capacity, the total shape coverage capacity and the total water turbidity; S4: Use the collected feature data to build a model and train it, deploy the trained model to the actual environment, extract features from new images and use the model to predict, and obtain the predicted value of the comprehensive pollution degree of the water area; S5: Use GIS technology to visualize the spatial distribution of the predicted values ​​of the comprehensive pollution level of water bodies obtained by the model on the map. According to the predicted values ​​of the comprehensive pollution level of water bodies and different colors representing different pollution levels, use GIS tools to analyze the spatial distribution of microplastic water pollution and identify hot spots.

[0020] Specifically, several water samples are collected in the target waters, 70% of the collected water samples are randomly selected as treatment objects, and pretreatment operations are performed on the selected treatment objects; 30% hydrogen peroxide with a volume ratio of 10:1 is added to each water sample, and the mixture is placed at a constant temperature of 60°C for digestion for 12 hours, and the digested water sample is mixed with a high-density saturated sodium chloride salt solution for density separation, stirred evenly and allowed to stand for 24 hours, the upper solution is collected, the upper solution is filtered, and the obtained filter paper is placed in a culture dish The pretreated water samples were dried at 30°C for 24 hours; the microplastics were detected by microscope visual recognition method, the number, color, size and shape characteristics of microplastics in each sample were recorded, the average abundance of microplastics in all samples of the target water area was calculated, the abundance of microplastics in each sample was compared with the average abundance of microplastics in the target water area, and water samples with microplastic abundance not less than the average abundance of microplastics in the target water area were screened out; the satellite remote sensing platform was used to obtain remote sensing images of the target water area, and the Canny edge detection algorithm was used to process the remote sensing images to extract the boundaries of the water body, and the extracted images were processed to obtain the boundaries of the water body. The extracted water body boundaries are preprocessed, including denoising and image enhancement processing, and feature data are extracted and identified from the preprocessed remote sensing images. The pollution coverage degree is obtained through the pollution adsorption degree of microplastics of different colors and the surface area of ​​microplastics of different shapes, and the total water turbidity degree of the water area is obtained through microplastics of different particle sizes. After cleaning and standardizing various feature data, the feature data of different dimensions are converted to a unified scale, and the total color adsorption capacity, total shape coverage capacity and total water turbidity are used as model inputs, and the comprehensive pollution degree of the water area is used as the model output. A machine learning model that can predict the comprehensive pollution degree of the water area is constructed, and the model is trained using the collected feature data. The trained model is deployed in a new actual environment, and the newly acquired remote sensing images are subjected to the same preprocessing and feature extraction. The feature data of the new image is used as input using the deployed model to obtain the predicted value of the comprehensive pollution degree of the water area. The model prediction value point data is converted into surface data and imported into the GIS software to generate the spatial distribution of the pollution degree prediction value, and different colors are used to represent the different pollution degrees obtained according to the threshold range of the predicted value of the comprehensive pollution degree of the water area.

[0021] In one embodiment of the present invention, the step S1 comprises the following steps: According to the preset sampling point coordinates, the width of the target water area is measured, and several sampling points are arranged at equal distances along the width of the water area. The sampling depth is between 0.5 and 1 meter of the depth of the target water area. Several water samples are collected in the target water area, and 70% of the collected water samples are randomly selected as processing objects, and pre-processing operations are performed on the selected processing objects.

[0022] Specifically, the width of the target water area is obtained by measuring, and several sampling points are arranged at equal distances along the width of the water area. The sampling depth is set to between 0.5 and 1 meter of the depth of the target water area. Several water samples are collected in the target water area, and 70% of the collected water samples are randomly selected as processing objects, and pre-processing operations are performed on the selected processing objects.

[0023] In one embodiment of the present invention, the preprocessing operation on the extracted processing object includes the following steps: The extracted treatment objects were subjected to pretreatment operations, including adding 30% hydrogen peroxide with a volume ratio of 10:1 to the samples, and digesting them at a constant temperature of 60°C for 12 hours; mixing the digested water samples with a high-density saturated sodium chloride salt solution for density separation, stirring them evenly and letting them stand for 24 hours to collect the upper solution; filtering the upper solution, and placing the obtained filter paper in a culture dish and drying it at 30°C for 24 hours.

[0024] Specifically, 30% hydrogen peroxide at a volume ratio of 10:1 was added to each water sample, and the mixture was digested at a constant temperature of 60°C for 12 hours. The digested water sample was mixed with a high-density saturated sodium chloride salt solution for density separation. After stirring evenly, the solution was allowed to stand and settle for 24 hours. The upper solution was collected and filtered, and the obtained filter paper was placed in a culture dish and dried at 30°C for 24 hours to ensure that the sample had undergone appropriate chemical treatment and physical separation process before analysis and research.

[0025] In one embodiment of the present invention, the step S2 comprises the following steps: The pretreated water samples were tested for microplastics using a microscope visual identification method, and the number, color, size and shape characteristics of the microplastics in each sample were recorded. Based on the test data, the average abundance of microplastics in all samples in the target waters was calculated, and the abundance of microplastics in each sample was compared with the average abundance of the target waters. Water samples with microplastic abundance greater than or equal to the average abundance of microplastics in the target waters were screened out and analyzed.

[0026] Specifically, all pretreated water samples were prepared, and each sample underwent density separation, filtration and drying steps to facilitate observation and identification of microplastic particles under a microscope. Each water sample was observed using a high-resolution microscope, and the color, size and shape characteristics of the microplastics in each sample were recorded. Based on the observed data, the average abundance of microplastics in all samples in the target waters was calculated, with an average abundance of 6 pieces / L. The plastic abundance of each sample was compared with the average abundance of the target waters, and water samples with microplastic abundance greater than or equal to the average abundance of microplastics in the target waters were screened out, and the screened samples were analyzed.

[0027] In one embodiment of the present invention, step S3 comprises the following steps: Remote sensing technology is used to obtain remote sensing images of screened water samples. The boundaries of the water bodies are extracted using the Canny edge detection algorithm and then preprocessed, including denoising and image enhancement. Feature data are extracted and identified from the remote sensing images, including the total color adsorption capacity, the total shape coverage capacity and the total water turbidity.

[0028] Specifically, a satellite remote sensing platform is used to obtain remote sensing images of the target water area, and the Canny edge detection algorithm is applied to process the remote sensing images to extract the boundaries of the water body. The extracted water body boundaries are preprocessed, including denoising and image enhancement. Feature data is extracted and identified from the preprocessed remote sensing images, and the pollution adsorption degree of microplastics of different colors and the pollution coverage degree are obtained through the surface area of ​​microplastics of different shapes. The overall water turbidity of the water area is obtained through microplastics of different particle sizes.

[0029] In one embodiment of the present invention, the total color adsorption capacity comprises the following steps: Collect image datasets containing microplastics and other objects. The model uses a convolutional neural network to extract image features, including color, size, and shape. The Mask R-CNN mask region convolutional neural network uses a region proposal network to generate potential microplastic locations, classify and segment the locations, assign a category label to each potential microplastic location, and generate a corresponding mask. Use the mouse to draw a polygon to mark the edge of the target object, use the annotation tool to annotate the microplastics and other objects in the image, and save it as a .json file. Convert the .json format to a format that matches the Mask R-CNN mask region convolutional neural network training. The converted data package includes image files and corresponding annotation files. Use the converted data to train the Mask R-CNN mask region convolutional neural network, and use the trained model to distinguish microplastics from other objects in the image. According to the color characteristics, the microplastics are divided into three categories, including black, blue and colorless microplastics. The images containing black, blue and colorless microplastics are annotated, and the annotated data set is used to train the classifier to automatically identify and classify microplastics. The object technology method is used to count the microplastics of each color category, and the proportion of the number of microplastics of each color to the total number of microplastics is calculated. The proportion value is used as the weight coefficient. According to the adsorption intensity coefficient set for microplastics of different colors based on the research data, the weight coefficient of each color of microplastics is combined with the adsorption intensity coefficient of the corresponding color, and the combined values ​​of the three colors are added to obtain the total color adsorption capacity of the water area; The total color adsorption capacity calculation formula is: in, is the degree of total color adsorption capacity, The amount of black microplastics, The amount of blue microplastics, is the amount of colorless microplastics, is the total amount of microplastics of three colors, , and is the set adsorption intensity coefficient.

[0030] In one embodiment of the present invention, the overall shape coverage capability comprises the following steps: The collected images are converted into grayscale images, and the pattern recognition algorithm in the image analysis software is used to identify and classify the shapes of microplastics. The edge detection algorithm detects the changes in pixel intensity in the image to determine the edge position, extracts the edges of microplastics from the image, and uses the contour tracking algorithm to connect the edge points to form the object contour; a Fourier transform infrared microscope based on a focal plane array is used to image the microplastic particles to obtain high-resolution two-dimensional images, and laser ranging is used to obtain the depth information of the microplastics. The depth information is combined with the two-dimensional image, and a three-dimensional reconstruction algorithm is used to generate a three-dimensional model of the microplastics. The surface of each microplastic is obtained through the three-dimensional model. The data of microplastics were collected by 3D model. The microplastics were divided into different shape types, including fiber, granular and film, and the number of microplastics in each shape category was counted. The surface area of ​​microplastics of different shapes was calculated. The surface area of ​​microplastics of each shape was multiplied by the corresponding number to obtain the total surface area of ​​the corresponding shape. The ratio of the total surface area of ​​each shape to the combined surface area of ​​the three shapes was calculated. The ratio was used as the weight coefficient to set the coverage breadth coefficient for each shape of microplastics. After multiplying the weight coefficient by the coverage breadth coefficient, the combined values ​​of the three shapes were added to obtain the total shape coverage capacity of the water area. The total shape coverage degree calculation formula is: in, is the combined surface area of ​​the three shapes, is the amount of fibrous microplastics, is the surface area of ​​fibrous microplastics, is the amount of granular microplastics, is the surface area of ​​granular microplastics, is the amount of film-like microplastics, is the surface area of ​​film-like microplastics, is the total shape coverage capability, , and is the set coverage breadth coefficient.

[0031] In one embodiment of the present invention, the total water turbidity degree comprises the following steps: Import the collected images into ImageJ open source software, set the ruler length and unit in the software, pre-process the images, including adjusting the contrast and brightness, use the Rectangle tool in the toolbar to manually mark the microplastic particles in the image, and for each marked particle, use the function provided by the software to draw a straight line through the center of the particle, calculate and record the maximum size of each particle, and classify the microplastics according to the measured particle size, including nano-scale microplastics, micro-scale microplastics, and millimeter-scale microplastics. Collect the particle size data of all particles, perform statistical analysis to obtain the average particle size of the microplastics, count the microplastics in each particle size range, count the number of microplastics in each particle size range, and jointly establish the shape ratios of three different shapes of microplastics to construct a weighted average particle size. According to the average particle size of each shape of microplastics and its ratio, the total water turbidity of the water area is obtained; The calculation formula of total water turbidity is: in, is the overall water turbidity level, is the particle size and For the The particle size of fibrous microplastics is is the amount of fibrous microplastics, is the total number of microplastics in three shapes, is the particle size and For the The particle size of microplastic particles is is the amount of granular microplastics, is the particle size and For the The particle size of film-like microplastics is is the amount of film-like microplastics.

[0032] Specifically, an image dataset containing microplastics and other objects is collected, and the model extracts image features, including color, size and shape data, through a convolutional neural network; the Mask R-CNN mask region convolutional neural network uses a region proposal network to generate potential microplastic locations and then classifies and segments them, assigns a category label to each potential microplastic location and generates a corresponding mask; a polygon is drawn with the mouse to mark the edge of the target object, and the microplastics and other objects in the image are marked with a marking tool, and saved as a .json file; the .json format is converted to a format that matches the training of the Mask R-CNN mask region convolutional neural network. The converted data package includes an image file and a corresponding annotation file. The Mask R-CNN mask region convolutional neural network is trained using the converted data, and the trained model is used to distinguish microplastics from other objects in the image; after extracting the color, shape and size features of the microplastics, the features are calculated using a variety of methods and algorithms to obtain the total color adsorption capacity, total shape coverage capacity and total water turbidity, respectively, providing a basis for building a model.

[0033] In one embodiment of the present invention, the step S4 comprises the following steps: Based on the feature data extracted from the image, including the total color adsorption capacity, the total shape covering capacity and the total water turbidity, a machine learning model that can predict the comprehensive pollution level of the water area is constructed after cleaning and standardization operations. The model is trained using the collected feature data, and the trained model is deployed in the actual environment. Features are extracted from new images and predicted using the model to obtain the predicted value of the comprehensive pollution level of the water area.

[0034] Specifically, a variety of feature data are extracted from the image, including the total color adsorption capacity, the total shape covering capacity and the total water turbidity. After cleaning and standardization operations are performed on the various feature data, the feature data of different dimensions are converted to a unified scale. The total color adsorption capacity, the total shape covering capacity and the total water turbidity are used as model inputs, and the comprehensive pollution degree of the water area is used as the model output. A machine learning model that can predict the comprehensive pollution degree of the water area is constructed, the model is trained using the collected feature data, and the trained model is deployed in a new actual environment. The newly acquired remote sensing image is subjected to the same preprocessing and feature extraction, and the feature data of the new image is used as input using the deployed model to obtain the predicted value of the comprehensive pollution degree of the water area.

[0035] In one embodiment of the present invention, the step S5 comprises the following steps: The model prediction value point data is converted into surface data and imported into GIS software to generate the spatial distribution of the pollution degree prediction value; the water area comprehensive pollution degree prediction value is set as low pollution when ≤1, medium pollution when 1<water area comprehensive pollution degree prediction value ≤2, and heavy pollution when 2<water area comprehensive pollution degree prediction value. Different colors are used to represent different pollution degrees, among which green represents low pollution degree, yellow represents medium pollution degree, and red represents high pollution degree. The mapping function of GIS software is used to add annotations and legends; the spatial distribution of microplastic water pollution is analyzed and hot spots are identified based on GIS tools.

[0036] Specifically, the model is used to obtain the predicted value of the comprehensive pollution degree of the water area, and the predicted value is sorted according to the longitude and latitude coordinates to ensure that each location has a corresponding pollution prediction value. The point data is converted into surface data through the tools in the GIS software, and the pollution degree data predicted by the model is imported into the GIS software to generate a spatial distribution. According to the range of the predicted value, colors are selected to represent different levels of pollution degree, green represents low pollution degree, yellow represents medium pollution degree, and red represents high pollution degree.

[0037] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing, characterized in that: The following steps are involved: S1: Collect several water samples in the target waters and extract 70% of the sample data for preprocessing; S2: Detect the pre-treated water samples and screen out water samples whose microplastic abundance is greater than or equal to the average abundance of microplastics in the target waters; S3: remote sensing technology is used to obtain remote sensing images of the screened water samples, and the boundaries of the water bodies are extracted by the Canny edge detection algorithm and then preprocessed to extract and identify feature data from the remote sensing images, including the total color adsorption capacity, the total shape coverage capacity and the total water turbidity; S4: Use the collected feature data to build a model and train it, deploy the trained model to the actual environment, extract features from new images and use the model to predict, and obtain the predicted value of the comprehensive pollution degree of the water area; S5: Use GIS technology to visualize the spatial distribution of the predicted values ​​of the comprehensive pollution level of water bodies obtained by the model on the map. According to the predicted values ​​of the comprehensive pollution level of water bodies and different colors representing different pollution levels, use GIS tools to analyze the spatial distribution of microplastic water pollution and identify hot spots.

2. According to the method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing in claim 1, it is characterized in that: The step S1 comprises the following steps: According to the preset sampling point coordinates, the width of the target water area is measured, and several sampling points are arranged at equal distances along the width of the water area. The sampling depth is between 0.5 and 1 meter of the depth of the target water area. Several water samples are collected in the target water area, and 70% of the collected water samples are randomly selected as processing objects, and pre-processing operations are performed on the selected processing objects.

3. According to claim 2, a method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing is characterized in that: The preprocessing operation on the extracted processing object includes the following steps: The extracted treatment objects were subjected to pretreatment operations, including adding 30% hydrogen peroxide with a volume ratio of 10:1 to the samples, and digesting them at a constant temperature of 60°C for 12 hours; mixing the digested water samples with a high-density saturated sodium chloride salt solution for density separation, stirring them evenly and letting them stand for 24 hours to collect the upper solution; filtering the upper solution, and placing the obtained filter paper in a culture dish and drying it at 30°C for 24 hours.

4. According to the method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing in claim 1, it is characterized in that: The step S2 comprises the following steps: The pretreated water samples were tested for microplastics using a microscope visual identification method, and the number, color, size and shape characteristics of the microplastics in each sample were recorded. Based on the test data, the average abundance of microplastics in all samples in the target waters was calculated, and the abundance of microplastics in each sample was compared with the average abundance of the target waters. Water samples with microplastic abundance greater than or equal to the average abundance of microplastics in the target waters were screened out and analyzed.

5. According to the method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing in claim 1, it is characterized in that: The step S3 comprises the following steps: Remote sensing technology is used to obtain remote sensing images of screened water samples. The boundaries of the water bodies are extracted using the Canny edge detection algorithm and then preprocessed, including denoising and image enhancement. Feature data are extracted and identified from the remote sensing images, including the total color adsorption capacity, the total shape coverage capacity and the total water turbidity.

6. The method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing according to claim 5 is characterized in that: The total color adsorption capacity degree comprises the following steps: Collect image datasets containing microplastics and other objects. The model uses a convolutional neural network to extract image features, including color, size, and shape. The Mask R-CNN mask region convolutional neural network uses a region proposal network to generate potential microplastic locations, classify and segment the locations, assign a category label to each potential microplastic location, and generate a corresponding mask. Use the mouse to draw a polygon to mark the edge of the target object, use the annotation tool to annotate the microplastics and other objects in the image, and save it as a .json file. Convert the .json format to a format that matches the Mask R-CNN mask region convolutional neural network training. The converted data package includes image files and corresponding annotation files. Use the converted data to train the Mask R-CNN mask region convolutional neural network, and use the trained model to distinguish microplastics from other objects in the image. According to the color characteristics, the microplastics are divided into three categories, including black, blue and colorless microplastics. The images containing black, blue and colorless microplastics are annotated, and the annotated data set is used to train the classifier to automatically identify and classify microplastics. The object technology method is used to count the microplastics of each color category, and the proportion of the number of microplastics of each color to the total number of microplastics is calculated. The proportion value is used as the weight coefficient. According to the adsorption intensity coefficient set for microplastics of different colors based on the research data, the weight coefficient of each color of microplastics is combined with the adsorption intensity coefficient of the corresponding color, and the combined values ​​of the three colors are added to obtain the total color adsorption capacity of the water area; The total color adsorption capacity calculation formula is: in, is the degree of total color adsorption capacity, The amount of black microplastics, The amount of blue microplastics, is the amount of colorless microplastics, is the total amount of microplastics of three colors, , and is the set adsorption strength coefficient.

7. The method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing according to claim 5 is characterized in that: The overall shape coverage capability degree comprises the following steps: The collected images are converted into grayscale images, and the pattern recognition algorithm in the image analysis software is used to identify and classify the shapes of microplastics. The edge detection algorithm detects the changes in pixel intensity in the image to determine the edge position, extracts the edges of microplastics from the image, and uses the contour tracking algorithm to connect the edge points to form the object contour; a Fourier transform infrared microscope based on a focal plane array is used to image the microplastic particles to obtain high-resolution two-dimensional images, and laser ranging is used to obtain the depth information of the microplastics. The depth information is combined with the two-dimensional image, and a three-dimensional reconstruction algorithm is used to generate a three-dimensional model of the microplastics. The surface of each microplastic is obtained through the three-dimensional model. The data of microplastics were collected by 3D model. The microplastics were divided into different shape types, including fiber, granular and film, and the number of microplastics in each shape category was counted. The surface area of ​​microplastics of different shapes was calculated. The surface area of ​​microplastics of each shape was multiplied by the corresponding number to obtain the total surface area of ​​the corresponding shape. The ratio of the total surface area of ​​each shape to the combined surface area of ​​the three shapes was calculated. The ratio was used as the weight coefficient to set the coverage breadth coefficient for each shape of microplastics. After multiplying the weight coefficient by the coverage breadth coefficient, the combined values ​​of the three shapes were added to obtain the total shape coverage capacity of the water area. The total shape coverage degree calculation formula is: in, is the combined surface area of ​​the three shapes, is the amount of fibrous microplastics, is the surface area of ​​fibrous microplastics, is the amount of granular microplastics, is the surface area of ​​granular microplastics, is the amount of film-like microplastics, is the surface area of ​​film-like microplastics, is the total shape coverage capability, , and is the set coverage breadth coefficient.

8. The method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing according to claim 5 is characterized in that: The total water turbidity degree comprises the following steps: Import the collected images into ImageJ open source software, set the ruler length and unit in the software, pre-process the images, including adjusting the contrast and brightness, use the Rectangle tool in the toolbar to manually mark the microplastic particles in the image, and for each marked particle, use the function provided by the software to draw a straight line through the center of the particle, calculate and record the maximum size of each particle, and classify the microplastics according to the measured particle size, including nano-scale microplastics, micro-scale microplastics, and millimeter-scale microplastics. Collect the particle size data of all particles, perform statistical analysis to obtain the average particle size of the microplastics, count the microplastics in each particle size range, count the number of microplastics in each particle size range, and jointly establish the shape ratios of three different shapes of microplastics to construct a weighted average particle size. According to the average particle size of each shape of microplastics and its ratio, the total water turbidity of the water area is obtained; The calculation formula of total water turbidity is: in, is the overall water turbidity level, is the particle size and For the The particle size of fibrous microplastics is is the amount of fibrous microplastics, is the total number of microplastics in three shapes, is the particle size and For the The particle size of microplastic particles is is the amount of granular microplastics, is the particle size and For the The particle size of film-like microplastics is is the amount of film-like microplastics.

9. The method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing according to claim 1, characterized in that: The step S4 comprises the following steps: Based on the feature data extracted from the image, including the total color adsorption capacity, the total shape covering capacity and the total water turbidity, a machine learning model that can predict the comprehensive pollution level of the water area is constructed after cleaning and standardization operations. The model is trained using the collected feature data, and the trained model is deployed in the actual environment. Features are extracted from new images and predicted using the model to obtain the predicted value of the comprehensive pollution level of the water area.

10. The method for identifying the spatial distribution of microplastic-contaminated waters based on remote sensing according to claim 1, characterized in that: The step S5 comprises the following steps: The model prediction value point data is converted into surface data and imported into GIS software to generate the spatial distribution of the pollution degree prediction value; the water area comprehensive pollution degree prediction value is set as low pollution when ≤1, medium pollution when 1<water area comprehensive pollution degree prediction value ≤2, and heavy pollution when 2<water area comprehensive pollution degree prediction value. Different colors are used to represent different pollution degrees, among which green represents low pollution degree, yellow represents medium pollution degree, and red represents high pollution degree. The mapping function of GIS software is used to add annotations and legends; the spatial distribution of microplastic water pollution is analyzed and hot spots are identified based on GIS tools.

Images