A soil microplastic identification and detection method and system based on image recognition

Abstract

The present invention discloses a soil microplastic identification and detection method and system based on image recognition, which belongs to the technical field of microplastic detection, and specifically comprises: mixing microplastics and water to prepare pure soil samples, collecting soil images, pH values ​​and soil ventilation of the soil samples in sequence, and then obtaining the bulk density of the soil samples; converting the soil images into HSV images, extracting soil image features, inputting the soil image features and corresponding bulk density data into an SVR model for training, and constructing a prediction model; constructing a microplastic identification model based on soil bulk density, soil ventilation, pH value and corresponding microplastic quantity; collecting and photographing contaminated soil samples, outputting the corresponding soil bulk density based on the SVR model, and inputting the soil bulk density together with the contaminated soil ventilation and pH value into the microplastic identification model, and outputting the corresponding microplastic quantity; the present invention can quickly and accurately identify and detect microplastics in soil, and provide strong technical support for environmental protection.

Description

Technical Field

[0001] The present invention relates to the technical field of microplastic detection, and in particular to a soil microplastic identification and detection method and system based on image recognition. Background Art

[0002] Microplastics are defined as plastic particles with a diameter of less than 5 mm. Due to their tiny size and wide distribution, they have caused serious impacts on the environment and ecosystem. These microplastic particles are not only found in marine and freshwater systems, but are also commonly found in soil environments. As one of the largest microplastic repositories on Earth, soil microplastic pollution is becoming increasingly prominent, posing a potential threat to soil ecological safety and human health.

[0003] Traditional microplastic detection methods mainly rely on steps such as sampling, separation and microscopic observation. These methods usually involve complex sample pretreatment processes such as density separation, filtration, staining and number counting under a microscope. These steps are not only time-consuming and labor-intensive, but also require high technical skills of operators, making it difficult to achieve large-scale monitoring and evaluation. In addition, due to the diversity and complexity of microplastics, traditional methods often have difficulty in fully and accurately identifying and quantifying microplastic particles.

[0004] In recent years, image recognition technology has been widely used in various fields, but there are few reports on soil microplastic detection. Image recognition technology can quickly and accurately identify specific parameters in an image by analyzing pixel information in an image. Based on image recognition technology and other detection technologies, the present invention proposes a soil microplastic identification and detection method and system based on image recognition. Summary of the invention

[0005] The purpose of the present invention is to provide a soil microplastic identification and detection method and system based on image recognition to solve the following technical problems:

[0006] Traditional microplastic detection methods mainly rely on sampling, which is not only time-consuming and labor-intensive, but also has high technical requirements for operators, making it difficult to achieve large-scale monitoring and evaluation.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A soil microplastic identification and detection method based on image recognition comprises the following steps:

[0009] Pure soil from the area to be tested is collected, divided into several soil samples of the same mass and placed in a perforated container. Different preset amounts of microplastics are mixed into each soil sample, and the same mass of water is added. After waiting for a preset time, the soil image, pH value and soil ventilation of the soil samples are collected in turn, and then the soil is dried to obtain the bulk density of different soil samples;

[0010] The soil image is converted into an HSV image, and soil image features in the HSV image are extracted, wherein the soil image features include color features and texture features, and the color features and texture features are reduced in dimension using principal component analysis, and the soil image features of different soil samples and the corresponding bulk density data are input into an SVR model for training, so as to construct a prediction model of soil image features and bulk density data;

[0011] Based on the convolutional neural network, soil bulk density, soil ventilation, pH value and the corresponding number of microplastics were input into the convolutional neural network to construct a microplastic recognition model, and the microplastic recognition model was verified;

[0012] Collect contaminated soil from the area to be tested, collect contaminated soil samples, add the corresponding mass of water, wait for the preset time, take the contaminated soil image, output the corresponding soil bulk density based on the SVR model, and input it into the microplastic identification model together with the contaminated soil ventilation volume and pH value to output the corresponding number of microplastics.

[0013] As a further solution of the present invention: after water is added to the soil sample until the perforated container stops dripping, the soil image, soil ventilation and pH value data are collected.

[0014] As a further solution of the present invention: the process of converting the soil image into the HSV image is:

[0015] Get the RGB value of each pixel in the soil image, and normalize the RGB values ​​respectively. After normalization, R1=R / 255, G1=G / 255, and B1=B / 255;

[0016] The hue H, saturation S, and brightness V values ​​of any pixel in the HSV color space image are:

[0017] ;

[0018] ;

[0019] V=max(R1,G1,B1);

[0020] Among them, mod6 means modulo 6;

[0021] The mask range is set to select the HSV image, and the boundary values ​​of the mask range are (0, 10, 0) and (30, 60, 25) respectively.

[0022] As a further solution of the present invention: the color features include the HSV components corresponding to the first-order moment mean, the second-order moment standard deviation and the third-order moment slope respectively; the texture features include the energy, entropy, contrast and inverse variance in the image grayscale co-occurrence matrix.

[0023] As a further solution of the present invention: the verification process of the microplastic identification model is:

[0024] The soil bulk density, soil aeration rate and pH value were marked as g, v and p respectively, and rectangular coordinate systems were established. The actual values ​​of soil bulk density g, soil aeration rate v and pH value p of the soil sample were used as the abscissa, and the corresponding number of microplastics N output by the microplastic identification model was used as the ordinate to generate three one-dimensional regression fitting function formulas:

[0025] N=ag+b; N=cv+d; N=ep+f;

[0026] The solutions of parameters a, b, c, d, e, and f are calculated based on the data of the horizontal and vertical coordinates in the formula; and the correlation coefficient r of the microplastic identification model is calculated by the following formula:

[0027]

[0028] The correlation coefficients r1, r2, r3, r4, r5 and r6 are calculated in sequence. If all the correlation coefficients are greater than or equal to 0.7, it means that the microplastic identification model is established. If any correlation coefficient is less than 0.7, the soil parameters are collected again to train the microplastic identification model.

[0029] As a further solution of the present invention: the soil ventilation volume is the total amount of air passing through a unit volume of soil under unit time and unit pressure, including oxygen and carbon dioxide.

[0030] As a further solution of the present invention: in the process of calculating the soil bulk density, only the mass of the water-containing soil and the dried soil is included, and the mass of the microplastics is eliminated.

[0031] A soil microplastic identification and detection system based on image recognition, used to implement the soil microplastic identification and detection method based on image recognition as claimed in the claims, comprising:

[0032] The data acquisition module is used to collect pure soil from the area to be tested, divide it into several soil samples of the same mass and place them in a perforated container, mix different preset amounts of microplastics into each soil sample, and add the same mass of water, wait for a preset time, collect the soil image, pH value and soil ventilation of the soil sample in sequence, and then dry the soil to obtain the bulk density of different soil samples;

[0033] An image training module is used to convert the soil image into an HSV image, extract soil image features in the HSV image, wherein the soil image features include color features and texture features, perform dimensionality reduction processing on the color features and texture features using principal component analysis, and input the soil image features of different soil samples and the corresponding bulk density data into an SVR model for training, so as to construct a prediction model for soil image features and bulk density data;

[0034] A parameter training module is used to input soil bulk density, soil aeration, pH value and the corresponding number of microplastics into a convolutional neural network based on a convolutional neural network, to construct a microplastic recognition model, and to verify the microplastic recognition model;

[0035] The result output module is used to collect contaminated soil from the area to be tested, collect contaminated soil samples, add water of corresponding mass, wait for a preset time, take images of the contaminated soil, output the corresponding soil bulk density based on the SVR model, and input it into the microplastic identification model together with the contaminated soil ventilation volume and pH value to output the corresponding number of microplastics.

[0036] Beneficial effects of the present invention:

[0037] The present invention constructs a prediction model for soil image features and bulk density data by inputting soil image features and corresponding bulk density data of different soil samples into an SVR model for training; the model can quickly and accurately predict the bulk density of soil samples, greatly reducing the time and labor cost of manual operations and improving detection efficiency; and then, through the measured soil pH value and soil ventilation volume, a microplastic recognition model is established based on a convolutional neural network to accurately identify the amount of microplastics; this enables the present invention to achieve comprehensive monitoring of soil microplastics, and provides strong technical support for the identification and detection of microplastics. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below in conjunction with the accompanying drawings.

[0039] Figure 1 It is a schematic diagram of a process of a soil microplastic identification and detection method based on image recognition of the present invention;

[0040] Figure 2 This is the training process of the prediction model of soil image features and bulk density data of the present invention. DETAILED DESCRIPTION

[0041] 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.

[0042] See also Figure 1-2 As shown, the present invention is a soil microplastic identification and detection method and system based on image recognition, comprising the following steps:

[0043] 1. Sample preparation:

[0044] Select representative soil from the area to be tested to ensure that the soil samples collected are sufficiently representative.

[0045] Divide the pure soil into several samples of equal mass and place them in perforated containers.

[0046] Different preset amounts of microplastics (e.g. 0, 10, 20, 30, 40, 50 particles) were added to each soil sample to ensure that the model could learn the characteristics of different levels of pollution.

[0047] The same mass of water was added to each container to simulate soil moisture conditions in the actual environment.

[0048] After waiting for 4-6 hours, allow the water to fully penetrate and mix so that the microplastics can fully contact the soil.

[0049] 2. Data collection:

[0050] The images, pH values, and soil aeration of each soil sample were collected in sequence. A high-resolution camera was used to capture images of the soil surface, while a pH meter and soil aeration meter were used to measure the corresponding indicators.

[0051] The soil samples were dried to obtain soil bulk density (dry soil weight / container volume) at different contamination levels.

[0052] 3. Feature extraction and model training:

[0053] The collected soil images are converted into HSV (hue, saturation, value) images to better extract color and texture features.

[0054] Extract color features (such as hue average, saturation average, etc.) and texture features (such as gray-level co-occurrence matrix, local binary pattern, etc.) of HSV images.

[0055] Principal component analysis (PCA) is used to perform dimensionality reduction on the extracted features to reduce the dimension and noise of the data.

[0056] The reduced-dimensional features and the corresponding soil bulk density data were input into the support vector regression (SVR) model for training to construct a prediction model between soil image features and soil bulk density.

[0057] Based on the convolutional neural network (CNN), soil bulk density, soil aeration, pH value and the corresponding number of microplastics are used as input to build a microplastic recognition model. Through a large amount of training data, the accuracy and generalization ability of the model are improved.

[0058] The microplastic recognition model was validated using an independent validation dataset to ensure its stability and reliability under different conditions.

[0059] 4. Contaminated soil detection:

[0060] Collect contaminated soil samples from the area to be tested, add the corresponding mass of water, wait for 4-6 hours, and then take an image of the contaminated soil.

[0061] Based on the trained SVR model, the corresponding soil bulk density prediction value is output.

[0062] The predicted bulk density of the contaminated soil, soil ventilation and pH value are input into the microplastic identification model together, and the corresponding number of microplastics is output.

[0063] Based on the number of microplastics output by the model, the degree of microplastic pollution in the tested area is evaluated.

[0064] Studies have shown that the presence of microplastics can significantly change the bulk density of the soil. This is because microplastics occupy the pore space originally used to store water, which weakens the soil's ability to retain water and increases the soil's compactness and weight per unit volume. This change directly affects the soil's pore structure, which in turn affects the soil's aeration and water transfer capacity. In addition, the additives released by microplastics and the pollutants adsorbed by them may change the chemical properties of the soil, such as pH and organic matter content. These changes in chemical properties can have an adverse effect on soil organisms.

[0065] Therefore, the present invention combines traditional soil physical and chemical analysis with modern machine learning technology. By converting soil images into HSV format and extracting color and texture features, and then using principal component analysis and SVR models for prediction, information about soil bulk density can be effectively obtained from soil images. This method not only simplifies the traditional cumbersome laboratory analysis steps, but also reduces human errors. Furthermore, a convolutional neural network is used to construct a microplastic recognition model, which is capable of processing complex data relationships and improving the prediction accuracy of microplastic content. CNN excels in image processing and can automatically learn and recognize patterns in images, which is particularly useful when analyzing soil images.

[0066] The SVR (Support Vector Regression) model is a regression method based on the support vector machine (SVM) and is used to solve regression problems with high accuracy and generalization capabilities. SVR achieves the purpose of predicting continuous variables by finding an optimal hyperplane in the data space so that as many data points as possible fall within a preset interval near the hyperplane;

[0067] The present invention uses the color and texture parameters of the soil surface image to characterize the soil roughness and construct an SVR prediction model; by training the image features and corresponding bulk density data of a large number of soil samples, the SVR model can learn the complex relationship between the two, and can quickly predict the corresponding soil bulk density based on the input soil image features. Compared with the traditional bulk density measurement method, the SVR model can quickly predict the soil bulk density through image features, avoiding the cumbersome physical measurement process, thereby significantly improving the overall efficiency of microplastic detection. And the SVR model, with its excellent generalization ability, can maintain good prediction accuracy between different soil samples. This means that even on new, unseen soil samples, the SVR model can provide reliable soil bulk density predictions.

[0068] In another preferred embodiment of the present invention, before data collection, it is necessary to ensure that the soil sample reaches a certain moisture balance state, so when water is added to the soil sample until the perforated container stops dripping, the soil image, soil ventilation and pH value data are collected.

[0069] In another preferred embodiment of the present invention, the process of converting the soil image into the HSV image is:

[0070] Get the RGB value of each pixel in the soil image, and normalize the RGB values ​​respectively. After normalization, R1=R / 255, G1=G / 255, and B1=B / 255;

[0071] The hue H, saturation S, and brightness V values ​​of any pixel in the HSV color space image are:

[0072] ;

[0073] ;

[0074] V=max(R1,G1,B1);

[0075] Among them, mod6 means modulo 6;

[0076] The mask range is set to select the HSV image, and the boundary values ​​of the mask range are (0, 10, 0) and (30, 60, 25) respectively.

[0077] In the present invention, the process of converting the soil image into an HSV image includes obtaining the RGB value of each pixel and normalizing it, and then calculating the values ​​of hue H, saturation S and brightness V. By setting the mask range to select the HSV image, the microplastic features in the soil sample can be effectively identified and analyzed. The advantage of this method is that it can accurately extract useful information from the soil image and enhance the distinguishability of image features through color space conversion, thereby improving the accuracy and efficiency of microplastic detection. In addition, the use of the mask range can further optimize the image analysis process, exclude irrelevant data, and make the model more focused on key visual features, thereby improving the performance of the entire detection system.

[0078] In another preferred embodiment of the present invention, the color features include HSV components corresponding to the first-order moment mean, second-order moment standard deviation and third-order moment slope respectively; the texture features include energy, entropy, contrast and inverse variance in the image grayscale co-occurrence matrix.

[0079] The collected soil images are converted from the original RGB color space to the HSV color space. The HSV color space is closer to the way humans perceive color. This conversion helps to better process and analyze the color information in the soil image. Since soil bulk density is related to the physical structure and color of the soil, these features can serve as an important basis for predicting soil bulk density.

[0080] In the HSV color space, useful features in soil images, especially color features and texture features, can be extracted more accurately. It also helps to reduce the impact of illumination changes on image analysis. In natural environments, the illumination conditions of soil samples may change. The HSV model makes image analysis more stable and reliable by separating color and brightness information.

[0081] In another preferred embodiment of the present invention, the verification process of the microplastic identification model is:

[0082] The soil bulk density, soil aeration rate and pH value were marked as g, v and p respectively, and rectangular coordinate systems were established. The actual values ​​of soil bulk density g, soil aeration rate v and pH value p of the soil sample were used as the abscissa, and the corresponding number of microplastics N output by the microplastic identification model was used as the ordinate to generate three one-dimensional regression fitting function formulas:

[0083] N=ag+b; N=cv+d; N=ep+f;

[0084] The solutions of parameters a, b, c, d, e, and f are calculated based on the data of the horizontal and vertical coordinates in the formula; and the correlation coefficient r of the microplastic identification model is calculated by the following formula:

[0085]

[0086] The correlation coefficients r1, r2, r3, r4, r5 and r6 are calculated in sequence. If all the correlation coefficients are greater than or equal to 0.7, it means that the microplastic identification model is established. If any correlation coefficient is less than 0.7, the soil parameters are collected again to train the microplastic identification model.

[0087] In the process of soil microplastic identification and detection based on image recognition, verifying the accuracy of the model is a crucial step. The process involves taking soil bulk density, soil aeration and pH value as independent variables and the number of microplastics predicted by the model as the dependent variable. By establishing a rectangular coordinate system and using a one-dimensional regression fitting function formula, the relationship between these variables can be quantified. The solution of the parameters is based on the horizontal and vertical coordinate data, which reflects the correlation between the actual measurement value and the model output.

[0088] In order to evaluate the effectiveness of the model, the concept of correlation coefficient r is introduced. The correlation coefficient is a statistical indicator used to measure the strength of the linear relationship between two variables. In this case, it is used to judge the consistency between the model output and the actual observed value. If all correlation coefficients are greater than or equal to 0.7, it indicates that the model has high predictive ability and good generalization performance; conversely, if any correlation coefficient is lower than 0.7, it indicates that the model's predictive ability on this particular variable is insufficient, and it is necessary to collect data again and further train and optimize the model.

[0089] This statistical analysis-based validation method provides an objective evaluation criterion for soil microplastic identification and detection. First, it ensures that the model maintains a high prediction accuracy under different environmental conditions and enhances the robustness of the model. Second, by defining whether the model is qualified through clear numerical thresholds, the model validation process is intuitive and easy to operate. Finally, this process also promotes the continuous improvement of the model because it can point out which specific variables require more attention or adjustment. In short, this method improves the accuracy and reliability of microplastic detection and provides strong technical support for environmental protection and soil health monitoring.

[0090] In another preferred embodiment of the present invention, the soil ventilation volume is the total amount of air passing through a unit volume of soil per unit time and unit pressure, including oxygen and carbon dioxide.

[0091] Soil ventilation refers to the total amount of air that passes through a unit volume of soil per unit time and unit pressure. By constructing a laboratory device to simulate the pressure difference between the soil and the atmosphere, the amount of air that passes through a soil sample in a given time is measured. The commonly used unit is mL / (cm³·s), and the larger the value, the better the soil ventilation. This measurement method helps to evaluate the ventilation capacity of soil under natural conditions, which is of great significance for optimizing farming management and improving soil structure.

[0092] In another preferred embodiment of the present invention, the process of calculating the soil bulk density only includes the mass of water-containing soil and dried soil, and the mass of microplastics is eliminated.

[0093] In the process of calculating soil bulk density, it is usually necessary to consider the dry weight and wet weight of the soil. Soil bulk density refers to the mass of soil per unit volume. In the assessment of microplastic pollution, it is necessary to accurately know the mass of the soil itself, excluding the influence of other foreign substances.

[0094] A soil microplastic identification and detection system based on image recognition, used to implement the above-mentioned soil microplastic identification and detection method based on image recognition, comprising:

[0095] The data acquisition module is used to collect pure soil from the area to be tested, divide it into several soil samples of the same mass and place them in a perforated container, mix different preset amounts of microplastics into each soil sample, and add the same mass of water, wait for a preset time, collect the soil image, pH value and soil ventilation of the soil sample in sequence, and then dry the soil to obtain the bulk density of different soil samples;

[0096] An image training module is used to convert the soil image into an HSV image, extract soil image features in the HSV image, wherein the soil image features include color features and texture features, perform dimensionality reduction processing on the color features and texture features using principal component analysis, and input the soil image features of different soil samples and the corresponding bulk density data into an SVR model for training, so as to construct a prediction model for soil image features and bulk density data;

[0097] A parameter training module is used to input soil bulk density, soil aeration, pH value and the corresponding number of microplastics into a convolutional neural network based on a convolutional neural network, to construct a microplastic recognition model, and to verify the microplastic recognition model;

[0098] The result output module is used to collect contaminated soil from the area to be tested, collect contaminated soil samples, add water of corresponding mass, wait for a preset time, take images of the contaminated soil, output the corresponding soil bulk density based on the SVR model, and input it into the microplastic identification model together with the contaminated soil ventilation volume and pH value to output the corresponding number of microplastics.

[0099] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A soil microplastic identification and detection method based on image recognition, characterized in that: The following steps are involved: Pure soil from the area to be tested is collected, divided into several soil samples of the same mass and placed in a perforated container. Different preset amounts of microplastics are mixed into each soil sample, and the same mass of water is added. After waiting for a preset time, the soil image, pH value and soil ventilation of the soil samples are collected in turn, and then the soil is dried to obtain the bulk density of different soil samples; The soil image is converted into an HSV image, and soil image features in the HSV image are extracted, wherein the soil image features include color features and texture features, and the color features and texture features are reduced in dimension using principal component analysis, and the soil image features of different soil samples and the corresponding bulk density data are input into an SVR model for training, so as to construct a prediction model of soil image features and bulk density data; Based on the convolutional neural network, soil bulk density, soil ventilation, pH value and the corresponding number of microplastics were input into the convolutional neural network to construct a microplastic recognition model, and the microplastic recognition model was verified; Collect contaminated soil from the area to be tested, collect contaminated soil samples, add the corresponding mass of water, wait for the preset time, take the contaminated soil image, output the corresponding soil bulk density based on the SVR model, and input it into the microplastic identification model together with the contaminated soil ventilation volume and pH value to output the corresponding number of microplastics.

2. According to claim 1, a soil microplastic identification and detection method based on image recognition is characterized in that: When water is added to the soil sample until the perforated container stops dripping, the soil image, soil aeration and pH value data are collected.

3. According to the method for identifying and detecting soil microplastics based on image recognition according to claim 1, it is characterized in that: The process of converting soil images into HSV images is: Get the RGB value of each pixel in the soil image, and normalize the RGB values ​​respectively. After normalization, R1=R / 255, G1=G / 255, and B1=B / 255; The hue H, saturation S, and brightness V values ​​of any pixel in the HSV color space image are: ； ； V=max(R1,G1,B1); Among them, mod6 means modulo 6; The mask range is set to select the HSV image, and the boundary values ​​of the mask range are (5, 10, 0) and (40, 55, 255) respectively.

4. A soil microplastic identification and detection method based on image recognition according to claim 3, characterized in that: The color features include HSV components corresponding to the first-order moment mean, second-order moment standard deviation and third-order moment slope respectively; the texture features include energy, entropy, contrast and inverse variance in the image gray-level co-occurrence matrix.

5. The method for identifying and detecting soil microplastics based on image recognition according to claim 1, characterized in that: The verification process of the microplastic identification model is as follows: The soil bulk density, soil aeration rate and pH value were marked as g, v and p respectively, and rectangular coordinate systems were established. The actual values ​​of soil bulk density g, soil aeration rate v and pH value p of the soil sample were used as the abscissa, and the corresponding number of microplastics N output by the microplastic identification model was used as the ordinate to generate three one-dimensional regression fitting function formulas: N=ag+b; N=cv+d; N=ep+f; The solutions of parameters a, b, c, d, e, and f are calculated based on the data of the horizontal and vertical coordinates in the formula; The correlation coefficient r of the microplastic identification model was calculated by the following formula: ； The correlation coefficients r1, r2, r3, r4, r5 and r6 are calculated in sequence. If all the correlation coefficients are greater than or equal to 0.7, it means that the microplastic identification model is established. If any correlation coefficient is less than 0.7, the soil parameters are collected again to train the microplastic identification model.

6. The method for identifying and detecting soil microplastics based on image recognition according to claim 1, characterized in that: The soil ventilation rate is the total amount of air, including oxygen and carbon dioxide, passing through a unit volume of soil per unit time and unit pressure.

7. The method for identifying and detecting soil microplastics based on image recognition according to claim 1, characterized in that: The process of calculating the soil bulk density only includes the mass of the water-containing soil and the dried soil, and excludes the mass of the microplastics.

8. A soil microplastic identification and detection system based on image recognition, used to implement a soil microplastic identification and detection method based on image recognition according to any one of claims 1 to 7, characterized in that: include: The data acquisition module is used to collect pure soil from the area to be tested, divide it into several soil samples of the same mass and place them in a perforated container, mix different preset amounts of microplastics into each soil sample, and add the same mass of water, wait for a preset time, collect the soil image, pH value and soil ventilation of the soil sample in sequence, and then dry the soil to obtain the bulk density of different soil samples; An image training module is used to convert the soil image into an HSV image, extract soil image features in the HSV image, wherein the soil image features include color features and texture features, perform dimensionality reduction processing on the color features and texture features using principal component analysis, and input the soil image features of different soil samples and the corresponding bulk density data into an SVR model for training, so as to construct a prediction model for soil image features and bulk density data; A parameter training module is used to input soil bulk density, soil aeration, pH value and the corresponding number of microplastics into a convolutional neural network based on a convolutional neural network, to construct a microplastic recognition model, and to verify the microplastic recognition model; The result output module is used to collect contaminated soil from the area to be tested, collect contaminated soil samples, add water of corresponding mass, wait for a preset time, take images of the contaminated soil, output the corresponding soil bulk density based on the SVR model, and input it into the microplastic identification model together with the contaminated soil ventilation volume and pH value to output the corresponding number of microplastics.

Images