A remote sensing monitoring method for crop data collection and planting area

Abstract

The present invention specifically relates to a method for remotely sensing and monitoring crop data collection and planting area, including the steps of: constructing multi-dimensional crop data; collecting multi-dimensional crop data; analyzing the relationship between the multi-dimensional crop data and the target variable "vegetation type", and sequentially determining the weight between each type value in the multi-dimensional crop data and the target variable "vegetation type" to obtain a determination model; converting new remote sensing data into a feature vector and inputting it into the determination model to obtain the predicted vegetation type; calculating the planting areas of different types of crops based on the predicted vegetation type for remote sensing monitoring of the planting areas of different types of crops. The present application realizes a deeper and more refined analysis and mining of crop data, thereby providing a more accurate and efficient method for remote sensing monitoring of crop planting areas.

Description

Technical Field

[0001] The present invention belongs to the field of agricultural remote sensing, and specifically relates to a method for remotely monitoring crop data collection and planting area. Background Technique

[0002] In the prior art, NDVI (Normalized Difference Vegetation Index), as an important indicator of vegetation coverage, can effectively reflect the growth status and coverage degree of vegetation. However, NDVI itself is only a relative value, that is, NDVI = (NIR - Red) / (NIR + Red), which is a value obtained by calculating the difference between the reflectance or radiation value of the near-infrared band (NIR, strongly reflected by vegetation) and the red band (Red, relatively more absorbed by vegetation), and dividing by the sum of the reflectance or radiation values of these two bands.

[0003] Therefore, NDVI cannot accurately quantify the specific vegetation coverage. For example, the vegetation coverage of a certain area can only be roughly guessed through the NDVI value, and detailed data such as the type of vegetation cannot be determined. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for remotely monitoring crop data collection and planting area to solve the problems raised in the above background technique.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] A method for remotely monitoring crop data collection and planting area, including the steps of: constructing multi-dimensional crop data;

[0007] The multi-dimensional crop data includes type values such as NDVI value, red band radiation value, near-infrared band radiation value, red band radiation pixel mean value, and near-infrared band radiation pixel mean value;

[0008] Collecting multi-dimensional crop data;

[0009] Analyzing the relationship between the multi-dimensional crop data and the target variable "vegetation type" through the Spearman rank correlation method, and successively determining the weight w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type";

[0010] Assigning the type value with the highest correlation coefficient and the p-value reaching the significant level with the target variable "vegetation type" the maximum weight value, and assigning the type value with the lowest correlation coefficient with the target variable "vegetation type" the minimum weight value;

[0011] Use a Multilayer Perceptron (MLP) to determine the relationship between the NDVI value, the radiation value in the red light band, the radiation value in the near-infrared band, the average radiation pixel value in the red light band, and the average radiation pixel value in the near-infrared band and the vegetation type to obtain a determination model;

[0012] After converting the new remote sensing data into feature vectors, input them into the determination model to obtain the predicted vegetation type;

[0013] Calculate the planting areas of different types of crops based on the predicted vegetation type for remote sensing monitoring of the planting areas of different types of crops.

[0014] Furthermore, the steps to collect multi-dimensional crop data:

[0015] Access the remote sensing image data of crop data publicly released by Earth observation satellites;

[0016] After downloading the original remote sensing data of crop data, radiometric calibration and georegistration are required;

[0017] Extract the corresponding band information from the remote sensing images of crop data respectively.

[0018] Furthermore, use the Spearman rank correlation method to analyze the relationship between multi-dimensional crop data and the target variable "vegetation type", including the steps: collect and organize multi-dimensional crop data; for the target variable "vegetation type", it needs to be converted into a form that can be quantitatively compared and sorted through coding; ensure that all observations of multi-dimensional crop data have complete data records; sort each variable according to the size of the observations and assign corresponding ranks; convert the target variable "vegetation type" into the corresponding rank; calculate the Spearman rank correlation coefficient for each multi-dimensional crop data and the vegetation type rank order respectively; each multi-dimensional crop data will obtain a Spearman rank correlation coefficient ρ and its corresponding p-value; the value range of ρ is between -1 and +1, and the larger the absolute value, the stronger the correlation, and the positive or negative sign indicates positive or negative correlation; the p-value is used to judge the significance of the correlation. If the p-value is less than the pre-set significance level, it is considered that there is a significant rank correlation relationship between the independent variable and the vegetation type; compare the correlation coefficient sizes and their significance levels of each multi-dimensional crop data and the vegetation type, and find the multi-dimensional crop data with the highest correlation coefficient and the p-value reaching the significant level.

[0019] Furthermore, determine the weight w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type". Specifically, the weights can be assigned according to the size of the Spearman rank correlation coefficient and statistical significance (i.e., p-value):

[0020] $w_i = (\rho_i - \min(\rho)) / (\max(\rho)-\min(\rho))$, where $(w_i)$ is the normalized weight of the $i$-th type value, $(\rho_i)$ is the Spearman rank correlation coefficient between the type value and the vegetation type, and $(\min(\rho))$ and $(\max(\rho))$ are the minimum and maximum Spearman rank correlation coefficients among all significantly correlated type values, respectively.

[0021] Furthermore, a Multilayer Perceptron (MLP) is used to determine the relationships between the NDVI value, the red-band radiation value, the near-infrared band radiation value, the mean red-band radiation pixel value, and the mean near-infrared band radiation pixel value and the vegetation type to obtain a determination model. The specific steps are as follows:

[0022] Collect and organize remote sensing data to ensure that each sample contains the NDVI value, the red-band radiation value, the near-infrared band radiation value, the mean red-band radiation pixel value, the mean near-infrared band radiation pixel value, and the corresponding vegetation type label;

[0023] Combine the NDVI value, the red-band radiation value, the near-infrared band radiation value, the mean red-band radiation pixel value, and the mean near-infrared band radiation pixel value into a feature vector, and retain the weight $w_i$ between each type value in the multi-dimensional crop data and the target variable "vegetation type" as the input tensor;

[0024] Build an MLP model to determine the number of neurons in the input layer, the number of hidden layers, the number of neurons in each layer, the activation function, and the number of neurons in the output layer;

[0025] Initialize the weight matrix and bias term of each layer in the model;

[0026] Divide the data set into a training set, a validation set, and a test set;

[0027] Use the training set data to train the model, update the weights through the backpropagation algorithm, and monitor the change of the loss function and the performance metrics on the validation set during the training process.

[0028] This application also discloses a remote sensing monitoring system for crop data collection and planting area, including a computer program product of instructions, which when running on a computer, causes the computer to execute the above-mentioned remote sensing monitoring method for crop data collection and planting area.

[0029] Beneficial effects

[0030] 1. Multi-dimensional data integration: This application no longer solely relies on a single NDVI index. Instead, it constructs a crop dataset that includes multiple dimensions such as NDVI values, red-band radiation values, near-infrared band radiation values, and the average values of corresponding band pixels, enabling a more comprehensive perspective to reflect the growth status of crops and the characteristics of vegetation cover.

[0031] 2. Correlation analysis and weight determination: The Spearman rank correlation analysis method is used to study the relationships between these multi-dimensional data and vegetation types, and different variables are assigned corresponding weights according to the significance level, thereby more accurately quantifying the importance of each variable for vegetation classification.

[0032] 3. Application of deep learning models: The multi-layer perceptron (MLP) model is adopted, using multiple remote sensing band information as input features, and leveraging the powerful non-linear expression ability and learning ability of neural networks to improve the recognition accuracy of vegetation types and the prediction accuracy of planting areas.

[0033] 4. Model optimization and improvement of generalization ability: By reasonably dividing the training set, validation set, and test set, and performing model training, parameter updating, and hyperparameter adjustment, the performance and generalization ability of the model on unknown data are effectively improved.

[0034] The technical solution of this application not only enriches the data dimensions for evaluating vegetation cover and crop planting areas but also realizes a more in-depth and refined analysis and mining of crop data through statistical methods and deep learning techniques, thus providing a more accurate and efficient method for remote sensing monitoring of crop planting areas. Brief Description of the Drawings

[0035] Figure 1 is the flowchart of the method of this application. Detailed Embodiments

[0036] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0037] The remote sensing monitoring method for crop data collection and planting area disclosed in this application, as Figure 1 , includes the following steps:

[0038] Construct multi-dimensional crop data;

[0039] Multi-dimensional crop data includes type values such as NDVI value, red band radiation value, near-infrared band radiation value, red band radiation pixel mean, and near-infrared band radiation pixel mean;

[0040] Steps for collecting multi-dimensional crop data:

[0041] Access the remote sensing image data of crop data publicly released by Earth observation satellites (such as Landsat series, MODIS, Sentinel, etc.). These satellites can capture the electromagnetic radiation of different bands reflected by the crop data surface, including visible light and near-infrared bands;

[0042] After downloading the original remote sensing data of crop data, radiometric correction (to eliminate the influence of atmospheric scattering and absorption) and georegistration (geometric correction) are required to ensure the accuracy of the crop data;

[0043] Extract the corresponding band information from the remote sensing images of crop data respectively; for example, the red band radiation value refers to extracting the digital value of the corresponding red band (Red band); the near-infrared band radiation value refers to extracting the value of the near-infrared band (Near-Infrared, NIR);

[0044] Calculate the NDVI value: Use the formula `NDVI = (NIR - Red) / (NIR + Red)` to calculate the NDVI value of each pixel;

[0045] Statistical pixel mean:

[0046] For the red band radiation pixel mean and the near-infrared band radiation pixel mean, the average value of each pixel value of the corresponding band needs to be calculated within the region of interest (such as a specific farmland, experimental field or research area);

[0047] Analyze the relationship between multi-dimensional crop data and the target variable "vegetation type" through the Spearman rank correlation method: 1. Data preparation:

[0048] Collect and organize multi-dimensional crop data (NDVI value, red band radiation value, near-infrared band radiation value, red band radiation pixel mean, near-infrared band radiation pixel mean); for the target variable "vegetation type", it needs to be converted into a quantifiable and comparable form and sorted through coding (such as 0, 1, 2, etc.);

[0049] 2. Data preprocessing:

[0050] Ensure that all observed values of multi-dimensional crop data (NDVI value, red light band radiation value, near-infrared light band radiation value, red light band radiation pixel mean value, near-infrared light band radiation pixel mean value) have complete data records, and missing values need to be appropriately processed (such as deletion or interpolation); sort each variable according to the size of the observed value and assign corresponding ranks;

[0051] Convert the target variable "vegetation type" into corresponding ranks;

[0052] 3. Calculate the Spearman rank correlation coefficient:

[0053] Calculate the Spearman rank correlation coefficient for each multi-dimensional crop data (NDVI, red light band radiation value, near-infrared light band radiation value, pixel means of red and near-infrared light bands) and the rank order of vegetation types respectively;

[0054] Use statistical software (such as relevant libraries in Python) for calculation. There are built-in functions in the software to directly calculate the Spearman rank correlation coefficient;

[0055] Take the SciPy library in Python as an example to demonstrate how to calculate the Spearman rank correlation coefficient:

[0056]

[0057] Each multi-dimensional crop data will obtain a Spearman rank correlation coefficient ρ and its corresponding p-value;

[0058] The value range of ρ is between -1 and +1. The larger the absolute value, the stronger the correlation, and the positive or negative sign indicates positive or negative correlation;

[0059] The p-value is used to judge the significance of the correlation. If the p-value is less than a pre-set significance level (such as 0.05), it is considered that there is a significant rank correlation relationship between the independent variable and the vegetation type;

[0060] Compare the correlation coefficient sizes and their significance levels between each multi-dimensional crop data and the vegetation type, and find the multi-dimensional crop data with the highest correlation coefficient and a p-value reaching the significant level;

[0061] Assign the maximum weight value to the type value with the highest correlation coefficient and a p-value reaching the significant level with the target variable "vegetation type", and assign the minimum weight value to the type value with the lowest correlation coefficient with the target variable "vegetation type";

[0062] Determine the weight value w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type" in turn:

[0063] When determining the weight value, the weight can be assigned according to the magnitude and statistical significance (i.e., p-value) of the Spearman rank correlation coefficient:

[0064] w_i = (rho_i - min(rho)) / (max(rho) - min(rho)) where (w_i) is the normalized weight of the i-th type value, (rho_i) is the Spearman rank correlation coefficient between this type value and the vegetation type, and (min(rho)) and (max(rho)) are the minimum and maximum Spearman rank correlation coefficients among all significantly correlated type values, respectively.

[0065] For example:

[0066] Suppose the relationships between NDVI, the mean radiation of the near-infrared band and the vegetation type are the closest (with the highest ρ value and a significant p-value), then they will be assigned higher weight values. If the relationship between the radiation value of the red band and the vegetation type is the weakest, a lower weight value will be assigned.

[0067] Use a Multilayer Perceptron (MLP) to determine the relationships between NDVI values, red band radiation values, near-infrared band radiation values, the mean of red band radiation pixels, and the mean of near-infrared band radiation pixels and the vegetation type to obtain a determination model. The specific steps are as follows:

[0068] Collect and organize remote sensing data to ensure that each sample contains NDVI values, red band radiation values, near-infrared band radiation values, the mean of red band radiation pixels, the mean of near-infrared band radiation pixels, and the corresponding vegetation type labels;

[0069] Combine the NDVI values, red band radiation values, near-infrared band radiation values, the mean of red band radiation pixels, and the mean of near-infrared band radiation pixels into a feature vector, and retain the weight value w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type" as the input tensor;

[0070] For example, if each sample is a farmland area, then this vector can be represented as `(NDVI, Red_band_radiation, NIR_band_radiation, Red_mean, NIR_mean)`;

[0071] Build an MLP model to determine the number of neurons in the input layer (equal to the dimension of the feature vector), the number of hidden layers, the number of neurons in each layer, the activation function (such as ReLU, sigmoid, etc.), and the number of neurons in the output layer (equal to the number of classification categories);

[0072] Initialize the weight matrices and bias terms of each layer in the model, usually using random initialization;

[0073] Divide the dataset into a training set, a validation set, and a test set (e.g., 70% for training, 15% for validation, and 15% for testing);

[0074] Use the training set data to train the model, update the weights through the backpropagation algorithm, and monitor the changes in the loss function and performance metrics (such as accuracy, AUC, etc.) on the validation set during the training process;

[0075] Adjust hyperparameters, such as the learning rate, optimizer, regularization strength, etc.;

[0076] Convert the new remote sensing data into feature vectors and then input them into the determined model to obtain the predicted vegetation types;

[0077] The following is an example code for building an MLP model using Python and the Keras library, demonstrating how to combine multi-dimensional crop data into feature vectors and train the model:

[0078]

[0079]

[0080] Calculate the planting areas of different types of crops based on the predicted vegetation types for remote sensing monitoring of the planting areas of different types of crops.

[0081] It can be seen that this application discloses a method for crop data collection and remote sensing monitoring of planting areas, constructs a multi-dimensional crop dataset, uses the Spearman rank correlation analysis method to study the correlation between multi-dimensional crop data and vegetation types, constructs a multi-layer perceptron (MLP) model based on significantly correlated variables, takes feature vectors (such as (NDVI, Red_band_radiation, NIR_band_radiation, Red_mean, NIR_mean)) as inputs, trains and optimizes the model, uses the trained MLP model to process new remote sensing data, predicts the corresponding vegetation types, and then calculates the planting areas of different types of crops according to the prediction results to achieve remote sensing monitoring of crop planting areas.

[0082] Compared with the prior art that only relies on NDVI as an indicator of vegetation coverage, the improvements of this application are mainly reflected in the following aspects:

[0083] 1. Multi-dimensional data integration: This application no longer relies solely on the single NDVI index, but constructs a crop dataset that includes multiple dimensions such as NDVI values, red light band radiation values, near-infrared light band radiation values, and the mean values of corresponding band pixels, which can reflect the growth status of crops and vegetation coverage characteristics from a more comprehensive perspective.

[0084] 2. Correlation analysis and weight determination: The Spearman rank correlation analysis method is used to study the relationship between these multi-dimensional data and vegetation types, and different weights are assigned to different variables according to the significance level, so as to more accurately quantify the importance of each variable to vegetation classification.

[0085] 3. Application of deep learning model: The multi-layer perceptron (MLP) model is adopted, and the information of multiple remote sensing bands is used as input features. By utilizing the powerful non-linear expression ability and learning ability of the neural network, the recognition accuracy of vegetation types and the prediction accuracy of planting areas are improved.

[0086] 4. Model optimization and improvement of generalization ability: By reasonably dividing the training set, validation set and test set, model training, parameter updating and hyperparameter adjustment are carried out, effectively improving the performance and generalization ability of the model on unknown data.

[0087] The technical solution of this application not only enriches the data dimensions for evaluating vegetation cover and crop planting areas, but also realizes a more in-depth and refined analysis and mining of crop data by means of statistical methods and deep learning technologies, thus providing a more accurate and efficient method for remote sensing monitoring of crop planting areas.

[0088] The embodiments to be protected by this application include:

[0089] A remote sensing monitoring method for crop data collection and planting area, including the steps of: constructing multi-dimensional crop data;

[0090] The multi-dimensional crop data includes type values such as NDVI value, red light band radiation value, near-infrared band radiation value, red light band radiation pixel mean value, and near-infrared band radiation pixel mean value;

[0091] Collecting multi-dimensional crop data;

[0092] Analyzing the relationship between the multi-dimensional crop data and the target variable "vegetation type" by the Spearman rank correlation method, and sequentially determining the weight w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type".

[0093] Assigning the type value with the highest correlation coefficient with the target variable "vegetation type" and a p-value reaching the significant level the maximum weight value, and assigning the type value with the lowest correlation coefficient with the target variable "vegetation type" the minimum weight value;

[0094] Use a Multilayer Perceptron (MLP) to determine the relationship between the NDVI value, the radiation value in the red light band, the radiation value in the near-infrared band, the average radiation pixel value in the red light band, and the average radiation pixel value in the near-infrared band and the vegetation type to obtain a determination model;

[0095] After converting the new remote sensing data into a feature vector, input it into the determination model to obtain the predicted vegetation type;

[0096] Calculate the planting areas of different types of crops based on the predicted vegetation type for remote sensing monitoring of the planting areas of different types of crops.

[0097] Preferably, the steps for collecting multi-dimensional crop data are as follows:

[0098] Access the remote sensing image data of crop data publicly released by Earth observation satellites;

[0099] After downloading the original remote sensing data of crop data, radiometric correction and georegistration are required;

[0100] Extract the corresponding band information from the remote sensing images of crop data respectively.

[0101] Preferably, use the Spearman rank correlation method to analyze the relationship between multi-dimensional crop data and the target variable "vegetation type", including the steps: collect and organize multi-dimensional crop data; for the target variable "vegetation type", it needs to be converted into a form that can be quantitatively compared and sorted through coding; ensure that all observed values of multi-dimensional crop data have complete data records; sort each variable according to the size of the observed value and assign the corresponding rank; convert the target variable "vegetation type" into the corresponding rank; calculate the Spearman rank correlation coefficient for each multi-dimensional crop data and the vegetation type rank order respectively; each multi-dimensional crop data will obtain a Spearman rank correlation coefficient ρ and its corresponding p-value; the range of the ρ value is between -1 and +1, and the larger the absolute value, the stronger the correlation, and the positive or negative sign indicates positive or negative correlation; the p-value is used to judge the significance of the correlation. If the p-value is less than the pre-set significance level, it is considered that there is a significant rank correlation relationship between the independent variable and the vegetation type; compare the correlation coefficient sizes and their significance levels of each multi-dimensional crop data and the vegetation type, and find the multi-dimensional crop data with the highest correlation coefficient and the p-value reaching the significant level.

[0102] Preferably, determine the weight w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type". Specifically, the weight can be assigned according to the size and statistical significance (i.e., the p-value) of the Spearman rank correlation coefficient:

[0103] $w_i = (\rho_i - \min(\rho)) / (\max(\rho) - \min(\rho))$, where $(w_i)$ is the normalized weight of the $i$-th type value, $(\rho_i)$ is the Spearman rank correlation coefficient between this type value and the vegetation type, and $(\min(\rho))$ and $(\max(\rho))$ are the minimum and maximum Spearman rank correlation coefficients among all significantly correlated type values, respectively.

[0104] Preferably, a Multilayer Perceptron (MLP) is used to determine the relationships between the NDVI value, the red-band radiation value, the near-infrared band radiation value, the mean red-band radiation pixel value, and the mean near-infrared band radiation pixel value and the vegetation type to obtain a determination model. The specific steps are as follows:

[0105] Collect and organize remote sensing data to ensure that each sample contains the NDVI value, the red-band radiation value, the near-infrared band radiation value, the mean red-band radiation pixel value, the mean near-infrared band radiation pixel value, and the corresponding vegetation type label;

[0106] Combine the NDVI value, the red-band radiation value, the near-infrared band radiation value, the mean red-band radiation pixel value, and the mean near-infrared band radiation pixel value into a feature vector, and retain the weight $w_i$ between each type value in the multi-dimensional crop data and the target variable "vegetation type" as the input tensor;

[0107] Construct an MLP model to determine the number of neurons in the input layer, the number of hidden layers, the number of neurons in each layer, the activation function, and the number of neurons in the output layer;

[0108] Initialize the weight matrix and bias term of each layer in the model;

[0109] Divide the data set into a training set, a validation set, and a test set;

[0110] Use the training set data to train the model, update the weights through the backpropagation algorithm, and monitor the change of the loss function and the performance metrics on the validation set during the training process.

[0111] The embodiment of the present application also provides a computer program product including instructions, which when running on a computer, causes the computer to execute the method provided in the above embodiment.

[0112] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including those of the above method embodiments. The foregoing storage medium can be at least one of the following media: read-only memory (ROM), RAM, magnetic disk, or optical disk, etc., which can store program codes.

[0113] Therefore, the present application also discloses a remote sensing monitoring system for crop data collection and planting area, including a computer program product of instructions. When it runs on a computer, it causes the computer to execute the above remote sensing monitoring method for crop data collection and planting area.

[0114] It should be noted that the various embodiments in this specification are described in a progressive manner. For the same or similar parts among the various embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. As described above, only a specific implementation manner of the present application is provided, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered by the present application.

Claims

1. A method for remote sensing monitoring of crop data collection and planting area, characterized in that: The method includes the following steps: constructing multi-dimensional crop data; Multi-dimensional crop data includes type values ​​such as NDVI value, red light band radiation value, near infrared band radiation value, red light band radiation pixel mean, near infrared band radiation pixel mean; Collect multi-dimensional crop data; The Spearman rank correlation method is used to analyze the relationship between the multi-dimensional crop data and the target variable "vegetation type", and the weight w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type" is determined in turn; The type value with the highest correlation coefficient with the target variable "vegetation type" and a significant p-value is assigned the maximum weight value, and the type value with the lowest correlation coefficient with the target variable "vegetation type" is assigned the minimum weight value; A multi-layer perceptron was used to determine the relationship between NDVI value, red light band radiation value, near infrared band radiation value, red light band radiation pixel mean, near infrared band radiation pixel mean and vegetation type to obtain a determined model; The new remote sensing data is converted into feature vectors and then input into the determination model to obtain the predicted vegetation type; Calculate the planting area of ​​different types of crops based on the predicted vegetation types, and remotely monitor the planting area of ​​different types of crops; Determine the weight w_i between each type value in the multi-dimensional crop data and the target variable "vegetation type". The weight is assigned according to the size and statistical significance of the Spearman rank correlation coefficient, that is, the p value: w_i=(rho_i-min(rho)) / (max(rho)-min(rho)) where w_i is the standardized weight of the i-th type value, rho_i is the Spearman rank correlation coefficient between the type value and the vegetation type, min(rho) and max(rho) are the minimum and maximum Spearman rank correlation coefficients of all significantly correlated type values, respectively; The multi-layer perceptron is used to determine the relationship between NDVI value, red light band radiation value, near infrared band radiation value, red light band radiation pixel mean, near infrared band radiation pixel mean and vegetation type. The specific steps of the determination model are as follows: Collect and organize remote sensing data to ensure that each sample contains NDVI value, red light band radiation value, near infrared band radiation value, red light band radiation pixel mean and near infrared band radiation pixel mean, as well as the corresponding vegetation type label; Combine the NDVI value, red light band radiation value, near infrared band radiation value, red light band radiation pixel mean and near infrared band radiation pixel mean into a feature vector, and retain the weight w_i between each type value and the target variable "vegetation type" in the multi-dimensional crop data as the input tensor; Construct the MLP model, determine the number of neurons in the input layer, the number of hidden layers, the number of neurons in each layer, the activation function, and the number of neurons in the output layer; Initialize the weight matrix and bias terms of each layer in the model; Divide the dataset into training set, validation set and test set; Use the training set data to train the model, update the weights through the back-propagation algorithm, and monitor the changes in the loss function and the performance indicators on the validation set during the training process.

2. The method for crop data collection and planting area remote sensing monitoring according to claim 1, characterized in that: Steps to collect multi-dimensional crop data: Access crop data remote sensing image data publicly released by earth observation satellites; After downloading the original remote sensing data of crop data, radiometric correction and georeferencing are required; The corresponding band information is extracted from the crop data remote sensing images respectively.

3. The method for crop data collection and planting area remote sensing monitoring according to claim 1, characterized in that: The Spearman rank correlation method is used to analyze the relationship between multi-dimensional crop data and the target variable "vegetation type", including the following steps: collecting and organizing multi-dimensional crop data; for the target variable "vegetation type", it is necessary to convert it into a form that can be quantified and compared, and sort it by coding; ensuring that all multi-dimensional crop data observations have complete data records; sorting each variable according to the size of the observation value and assigning a corresponding rank; converting the target variable "vegetation type" into a corresponding rank; calculating the Spearman rank correlation coefficient for each multi-dimensional crop data and vegetation type rank order; each multi-dimensional crop data will obtain a Spearman rank correlation coefficient ρ and its corresponding p value; the ρ value range is between -1 and +1, the larger the absolute value, the stronger the correlation, and the positive and negative signs indicate positive or negative correlation; The p-value is used to judge the significance of the correlation. If the p-value is less than the pre-set significance level, it is considered that there is a significant rank correlation between the independent variable and the vegetation type. Compare the correlation coefficient and significance level of each multi-dimensional crop data and vegetation type to find the multi-dimensional crop data with the highest correlation coefficient and p-value reaching a significant level.

4. A system for remote sensing monitoring of crop data collection and planted area, comprising a computer program product of instructions, characterized in that: When it is run on a computer, the computer executes the method for remote sensing monitoring of crop data collection and planting area as described in claim 1.

Images