Remote sensing monitoring method for farmland film mulching age limit

Abstract

The invention discloses a remote sensing monitoring method for farmland film mulching age limit, and particularly relates to the technical field of farmland film mulching. By using multispectral and hyperspectral remote sensing technologies, combining farmland environment data such as temperature, humidity and illumination intensity, and by establishing a spectrum database, comparing current remote sensing data with mulching film spectrum characteristics in a preliminary database, and accurately distinguishing normal and abnormal spectrum characteristics; in the identification process, if the distinguishing accuracy is high, the method is directly used for farmland management decision making; if the accuracy is low, the distinguishing capability is enhanced through environment data auxiliary analysis; the accuracy of monitoring the mulching film state is improved, misjudgment is reduced, and a scientific basis is provided for replacement and management of the mulching film, so that farmland management is optimized, the service life of the mulching film is prolonged, and the purposes of reducing cost and increasing efficiency are finally achieved.

Description

Technical Field

[0001] The invention relates to the technical field of farmland film laying, and in particular to a remote sensing monitoring method for the years of farmland film laying. Background Art

[0002] The life of agricultural film covering refers to the effective time or duration of the film used on the field when the film covering technology is used in agricultural production. Generally, the service life of the film is related to factors such as its material quality, thickness, environmental conditions, and the types of crops planted. After reaching the life of the film covering, the film may age, break, or lose its functionality and need to be replaced or removed to ensure the normal growth of crops and prevent long-term adverse effects on the soil.

[0003] Remote sensing monitoring of the life of agricultural land mulch refers to the use of satellites, drones and other remote sensing technologies to monitor and evaluate the coverage and service life of agricultural land mulch. By collecting agricultural image data and analyzing the reflection characteristics, spectral changes and other information of the mulch, remote sensing technology can identify the aging degree, damage and usage time of the mulch, thereby providing a scientific basis for agricultural management, effectively extending the service life of the mulch or replacing it in time, and optimizing agricultural land management. However, when using remote sensing technology to monitor the service life of agricultural land mulch, the spectral characteristics of the reflection of mulch at different aging stages or damage levels (i.e., the reflection pattern of light captured by remote sensing equipment) may be very similar to the spectral characteristics of certain types of soil or plant residues in the field. This similarity makes it difficult for remote sensing data to accurately distinguish between mulch and other substances, resulting in incorrect judgment of the state or usage time of the mulch, affecting the correct decision-making of agricultural land management. Summary of the invention

[0004] The purpose of the present invention is to provide a remote sensing monitoring method for the years of film laying in farmland to solve the shortcomings of the background technology.

[0005] In order to achieve the above object, the present invention provides the following technical solution: a remote sensing monitoring method for the years of film laying in farmland, comprising the following steps:

[0006] S1: Use multispectral and hyperspectral remote sensing equipment to continuously collect spectral data of mulch, soil and plant residues in farmland, and obtain farmland environmental data at different time periods, wherein the farmland environmental data includes temperature, humidity and light intensity;

[0007] S2: Classify the collected spectral data, establish spectral databases of mulch, soil and plant residues respectively, compare the spectral features of mulch in the current remote sensing data with the mulch features in the preliminary spectral database, and divide them into normal spectral features and abnormal spectral features;

[0008] S3: Analyze the abnormal spectral characteristics and determine the accuracy of remote sensing data in distinguishing between mulch and soil or plant residues according to the degree of spectral confusion between mulch and soil or plant residues at different aging stages;

[0009] S4: If the accuracy of distinction is high, remote sensing data will be directly used for farmland management decisions. If the accuracy of distinction is low, farmland environmental data will be used for auxiliary judgment to enhance the accuracy of distinction.

[0010] Preferably, in S2, the spectral features are compared by using a support vector machine to classify the spectral data into normal features and abnormal features, specifically:

[0011] The spectral data of mulch, soil and plant residues in farmland are collected, and the preprocessed spectral data are divided into normal spectral features and abnormal spectral features; the data set is divided into a training set and a test set; feature vectors are extracted from each spectral sample, and the SVM kernel function is selected according to the characteristics of the spectral data; the SVM model is trained using the training set data, and the trained SVM model is applied to the test set to evaluate its classification accuracy, and the confusion matrix is ​​used to analyze the specific performance of the model classification results, including the number of correctly classified and misclassified samples, and the spectral feature vector currently collected by remote sensing is input into the trained SVM model; the SVM model outputs the classification result of each spectral feature and gives the confidence of the classification.

[0012] Preferably, according to the classification result output by SVM, the current spectral feature is classified as a normal feature or an abnormal feature, specifically:

[0013] If the confidence is greater than or equal to the preset confidence standard threshold, it is classified as a normal feature and determined as a normal spectrum; if the confidence is less than the preset confidence standard threshold, it is classified as an abnormal feature and determined as an abnormal spectrum.

[0014] Preferably, in S3, a peak-to-valley displacement fluctuation index is generated according to the change of the peak and trough positions of the spectrum curve, and the peak-to-valley displacement fluctuation index is obtained by:

[0015] The spectral data of ground film at different aging stages are standardized. On the spectral curve, the peak position is determined by identifying the local maximum point, and the peak position of the i-th band is λ p,i , the position of the trough is determined by identifying the local minimum point, and the trough position of the i-th band is set to λ v,i ; For the spectral curve of the aged mulch film, calculate the displacement of the peak of each band relative to the peak position of the standard mulch film, the expression is: Δλ p,i =|λ p,i,老化 -λ p,i,标准|; Calculate the displacement of the trough of each band relative to the trough position of the standard ground film, the expression is: Δλ v,i =|λ v,i,老化 -λ v,i,标准 |; Calculate the standard deviation of the peak displacement of all bands and the standard deviation of the trough displacement of all bands respectively, and calculate the peak-trough displacement fluctuation index PVSFI after comprehensive analysis. The expression is: In the formula, σ p and σ v is the standard deviation of the peak and trough displacements.

[0016] Preferably, in S3, a light intensity sensitivity index is generated according to the influence of different light intensities on the reflective characteristics of the ground film, and the method for obtaining the light intensity sensitivity index is:

[0017] The spectral reflectance of the ground film in different bands is measured by a multispectral camera, and the reflectance value R under each light intensity condition is recorded. i and the corresponding light intensity I i , where i represents different light intensity levels, and calculates the change in ground film reflectivity ΔR under adjacent light intensity conditions i , the expression is: ΔR i =R i+1 -R i ; Among them, R i+1 and R i are the reflectances under two adjacent light intensity levels; and calculate the light intensity change ΔI under adjacent light intensity conditions i , the expression is: ΔI i =I i+1 -I i ; Among them, I i+1 and I i are the light intensities of two adjacent light intensity levels. For each light intensity change interval, the light intensity sensitivity index ISI is calculated. i , the expression is: ISI i Indicates that when the light intensity changes from I i To I i+1 The sensitivity of the ground film reflectivity during the change of light intensity is calculated by integrating all light intensity intervals to calculate the overall light intensity sensitivity index ISI. avg , the expression is: Where n is the number of light intensity variation intervals.

[0018] Preferably, in S3, the peak-to-trough displacement fluctuation index and the light intensity sensitivity index are normalized, and the spectral confusion degree anomaly coefficient of the ground film and the soil or plant residues is calculated by the normalized peak-to-trough displacement fluctuation index and the light intensity sensitivity index.

[0019] Preferably, the obtained spectral confusion degree anomaly coefficient between the ground film and the soil or plant residue is compared with a preset spectral confusion degree anomaly coefficient reference threshold;

[0020] If the spectral confusion degree anomaly coefficient between the ground film and soil or plant residues is greater than or equal to the spectral confusion degree anomaly coefficient reference threshold, it means that the spectral confusion degree between the ground film and soil or plant residues is high, and the accuracy of remote sensing data in distinguishing ground film and soil or plant residues is low; if the spectral confusion degree anomaly coefficient between the ground film and soil or plant residues is less than the spectral confusion degree anomaly coefficient reference threshold, it means that the spectral confusion degree between the ground film and soil or plant residues is low, and the accuracy of remote sensing data in distinguishing ground film and soil or plant residues is high.

[0021] Preferably, in S4, if the distinction accuracy is low, auxiliary judgment is performed through farmland environmental data to enhance the distinction accuracy, specifically:

[0022] If the spectral confusion degree anomaly coefficient between the mulch film and the soil or plant residues is less than the spectral confusion degree anomaly coefficient reference threshold, the remote sensing data has low accuracy in distinguishing between the mulch film and the soil or plant residues. observed , that is, the currently observed spectral reflectance, and the farmland temperature T, farmland humidity H and light intensity L when the spectral data were collected;

[0023] Calculate the temperature correction factor C T , that is, the effect of temperature on spectral reflectance, expressed as: C T =α T ·(T-T0); where α T is the temperature sensitivity coefficient, T0 is the reference temperature, and the humidity correction factor C is calculated. H , that is, the effect of humidity on spectral reflectance, expressed as: C H =α H ·(H-H0); where α H is the humidity sensitivity coefficient, H0 is the reference humidity; the light intensity correction factor C L , that is, the effect of light intensity on spectral reflectance, expressed as: C L =α L ·(L-L0); where α L is the light sensitivity coefficient, L0 is the reference light intensity; calculate the comprehensive environmental correction factor C env, the expression is: C env =w T ·C T +w H ·C H +w L ·C L ; Among them, w T , w H and w L is the weight of the temperature, humidity and light intensity correction factors, using the comprehensive environmental correction factor C env For the observed spectral data R observed Correction is performed to obtain the corrected spectral reflectance R corrected , the expression is: R corrected =R observed -C env .

[0024] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0025] 1. The present invention systematically monitors the use and aging of farmland mulch by combining multispectral and hyperspectral remote sensing technologies with farmland environmental data. Using multiple technical means such as support vector machine classification, peak-to-trough displacement fluctuation index, and light intensity sensitivity index, the spectral characteristics of the mulch can be accurately identified, and the accuracy of remote sensing data differentiation can be further enhanced through environmental data correction algorithms. This method effectively overcomes the problem of confusion between the spectral characteristics of the mulch and the spectra of soil or plant residues, and improves the accuracy and reliability of remote sensing monitoring.

[0026] 2. The present invention not only optimizes the management of farmland mulch film, but also uses more accurate remote sensing data to determine the state and service life of the mulch film, effectively extending the life of the mulch film, reducing the replacement frequency and resource waste. At the same time, by combining environmental correction factors, a reliable correction method is provided when the accuracy of the distinction is low, ensuring the scientific nature of the decision-making basis, and ultimately achieving the goal of improving farmland management efficiency and saving costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0028] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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.

[0030] For examples, see Figure 1 As shown, the remote sensing monitoring method of the film laying period of farmland described in this embodiment includes the following steps:

[0031] S1: Use multispectral and hyperspectral remote sensing equipment to continuously collect spectral data of mulch, soil and plant residues in farmland, and obtain farmland environmental data at different time periods, wherein the farmland environmental data includes temperature, humidity and light intensity;

[0032] S2: Classify the collected spectral data, establish spectral databases of mulch, soil and plant residues respectively, compare the spectral features of mulch in the current remote sensing data with the mulch features in the preliminary spectral database, and divide them into normal spectral features and abnormal spectral features;

[0033] S3: Analyze the abnormal spectral characteristics and determine the accuracy of remote sensing data in distinguishing between mulch and soil or plant residues according to the degree of spectral confusion between mulch and soil or plant residues at different aging stages;

[0034] S4: If the accuracy of distinction is high, remote sensing data will be directly used for farmland management decisions. If the accuracy of distinction is low, farmland environmental data will be used for auxiliary judgment to enhance the accuracy of distinction.

[0035] In S1, select appropriate multispectral and hyperspectral remote sensing equipment, determine its working band and resolution, and ensure that the spectral characteristics of the mulch, soil and plant residues in the farmland can be captured. Before data collection, perform radiation calibration and geometric calibration on the remote sensing equipment to ensure the accuracy and consistency of the equipment under various lighting conditions. Determine the farmland area that needs to be monitored, and develop a flight or shooting route based on factors such as farmland area, terrain characteristics, and crop type. Set an appropriate frequency of remote sensing data collection based on the aging rate of the mulch and farmland management needs. For example, monitoring can be carried out once a week in the early stage, and the collection frequency can be increased during the critical growing season.

[0036] Use drones or satellite platforms to carry multispectral and hyperspectral sensors, fly at predetermined times and paths, and collect spectral data. During each flight or shooting process, obtain spectral reflectance data of mulch, soil, and plant residues in the field. Ensure that the equipment collects spectral information in different bands to capture subtle differences in target features.

[0037] Select high-precision temperature, humidity, and light intensity sensors to ensure that they can work long-term in the farmland environment and have automatic data recording capabilities. Deploy sensors in key areas within the farmland, such as plots with different soil types or areas with different mulch cover densities. Sensors should cover representative areas of the entire farmland to capture changes in the overall environment. Set the collection frequency of environmental sensors. Depending on the sensitivity of weather changes and mulch aging, collect environmental data once an hour or a day. Configure sensors to automatically record temperature, humidity, and light intensity data, and store the data in a local or cloud database for subsequent analysis and processing.

[0038] In this application, the use of multispectral and hyperspectral remote sensing equipment to continuously collect spectral data of mulch film, soil and plant residues in farmland helps to accurately identify and monitor the aging and damage of mulch film and avoid confusion with the spectral characteristics of soil or plant residues. By continuously collecting data, changes in the state of mulch film can be dynamically tracked, the accuracy and efficiency of farmland management can be improved, the service life of mulch film can be extended and resource input can be optimized.

[0039] In S2, the collected raw spectral data is cleaned to remove noise and outliers to ensure the quality and accuracy of the data. Radiometric correction is performed to standardize the spectral data to eliminate spectral differences caused by sensor or environmental changes. The geometric deformation of remote sensing images is corrected to ensure the consistency of data in spatial position.

[0040] According to the marks and known information at the time of collection, the spectral data were initially classified by source into three categories: mulch, soil and plant residues. Feature extraction was performed on each type of spectral data, and its spectral reflectance characteristics in different bands were analyzed, and the spectral curve of each sample was recorded.

[0041] A spectral database is designed, which includes three main categories: mulch, soil and plant residues. Each category contains the spectral characteristics, collection time, location and other related information of different samples. The classified spectral data and extracted features are entered into the corresponding database to form a preliminary spectral feature library.

[0042] The spectral characteristics of the current remote sensing data were standardized with those of the ground film in the preliminary spectral database to ensure that the spectral characteristics were compared on the same basis.

[0043] Select a suitable feature matching algorithm, such as Euclidean distance, correlation coefficient or support vector machine, to compare the similarity of spectral features.

[0044] For example, by comparing the similarity of spectral features through support vector machines, spectral data can be classified into normal features and abnormal features, specifically:

[0045] Collect spectral data of mulch, soil, and plant residues in farmland. These data should cover the spectral characteristics of mulch at different aging stages, various soil types, and plant residues. Standardize all spectral data to make them within the same scale range to facilitate subsequent comparison and classification. Commonly used methods include normalization and mean variance standardization.

[0046] The preprocessed spectral data are divided into two categories: normal spectral features (such as mulch that meets the standards) and abnormal spectral features (such as spectra of aged or damaged mulch, soil or plant residues).

[0047] The data set is divided into a training set and a test set, usually in a ratio of 70%-30% or 80%-20%. The training set is used to train the SVM model, and the test set is used to verify the model performance.

[0048] Feature vectors are extracted from each spectral sample. These feature vectors may include reflectance values ​​or other relevant features (such as peak position, bandwidth, etc.) of each band of the spectral data.

[0049] According to the characteristics of the spectral data, select the appropriate SVM kernel function. Commonly used kernel functions include: Linear kernel function: suitable for linearly separable feature data. Gaussian kernel function (RBF kernel): suitable for nonlinear data, widely used in the classification of spectral data. Polynomial kernel function: suitable for complex data in multidimensional feature space.

[0050] The training set data is used to train the SVM model. The SVM will find a hyperplane based on the input spectral feature vector to maximize the distinction between normal spectral features and abnormal spectral features.

[0051] Use cross-validation or grid search to adjust the key parameters of the SVM model (such as C and gamma values) to improve the classification accuracy of the model.

[0052] The trained SVM model is applied to the test set to evaluate its classification accuracy. The main indicators include precision, recall and F1 value.

[0053] Use the confusion matrix to analyze the specific performance of the model classification results, including the number of correctly classified and misclassified samples.

[0054] The spectral feature vector currently collected by remote sensing is input into the trained SVM model.

[0055] The SVM model outputs the classification results of each spectral feature and gives the confidence of the classification. The confidence reflects the similarity between the current spectral feature and the known categories in the model.

[0056] According to the classification results output by SVM, the current spectral feature is classified as a normal feature or an abnormal feature. If the confidence is greater than or equal to the preset confidence standard threshold, it is classified as a normal feature and is determined to be a normal spectrum; if the confidence is less than the preset confidence standard threshold, it is classified as an abnormal feature and is determined to be an abnormal spectrum. It may indicate that the ground film is aged, damaged, or confused with other substances.

[0057] The classification results, i.e. normal spectral features and abnormal spectral features, are stored in a new database for subsequent analysis and decision support. The classification results are associated with environmental data (such as temperature, humidity, light, etc.) to analyze the impact of environmental factors on spectral feature changes.

[0058] In S3, data marked as abnormal spectral features are extracted from the support vector machine (SVM) classification results. These data usually include spectral features of mulch that may be confused with soil or plant residue spectra. According to the aging stage of the mulch, the abnormal spectral features are further classified, for example, into mild aging, moderate aging, and severe aging categories.

[0059] The aging of mulch will cause changes in the location of peaks and troughs in its spectral curve. By analyzing these changes, especially in the near-infrared and short-wave infrared bands, the possibility of spectral confusion can be determined. For example, the displacement or disappearance of peaks and troughs may mean that the spectral features overlap with soil or plant residues.

[0060] The peak-to-valley displacement fluctuation index is generated according to the change of the peak and trough positions of the spectral curve. The method for obtaining the peak-to-valley displacement fluctuation index is as follows:

[0061] The spectral data of ground films at different aging stages were standardized to eliminate the influence of acquisition conditions and ensure that the data were compared within the same scale range.

[0062] On the spectrum curve, the peak position is determined by identifying the local maximum point, and the peak position of the i-th band is λ p,i The trough position is determined by identifying the local minimum point, and the trough position of the i-th band is set to λ v,i .

[0063] For the spectral curve of the aged mulch film, the displacement of the peak of each band relative to the peak position of the standard (non-aged or slightly aged) mulch film is calculated, and the expression is: Δλ p,i =|λ p,i,老化 -λ p,i,标准 |; Calculate the displacement of the trough of each band relative to the trough position of the standard ground film, the expression is: Δλ v,i =|λ v,i,老化 -λ v,i,标准|; Calculate the standard deviation of the peak displacement of all bands and the standard deviation of the trough displacement of all bands respectively, and calculate the peak-trough displacement fluctuation index PVSFI after comprehensive analysis. The expression is: In the formula, σ p and σ v is the standard deviation of the peak and trough displacements, indicating the degree of displacement fluctuation.

[0064] The larger the peak-to-valley displacement fluctuation index (PVSFI), the greater the fluctuation of the spectral characteristics of the mulch at different aging stages, and the more significant changes in the positions of the peaks and troughs. Such significant changes may cause the spectral characteristics of the mulch to be more similar to the spectra of soil or plant residues, increasing the possibility of spectral confusion.

[0065] As the peak-to-trough displacement fluctuation index increases, the accuracy of remote sensing data in distinguishing mulch from soil or plant residues may decrease. Due to the large fluctuation amplitude of spectral features, the classification model may find it more difficult to accurately distinguish mulch from other substances, resulting in an increased risk of misclassification, which in turn affects the accuracy of remote sensing monitoring.

[0066] Light intensity affects the reflective characteristics of the ground film, especially in the short wavelength (such as visible light and near infrared) band. Insufficient or excessive light intensity will cause abnormal spectral reflectance, thereby increasing the possibility of spectral confusion. The spectrum of the same material may be different under different lighting conditions.

[0067] The light intensity sensitivity index is generated according to the influence of different light intensities on the reflective characteristics of the ground film. The method for obtaining the light intensity sensitivity index is:

[0068] The spectral reflectance of the ground film in different bands (such as visible light, near infrared, etc.) is measured by a multispectral camera, and the reflectance value R under each light intensity condition is recorded. i and the corresponding light intensity I i , where i represents different light intensity levels, and calculates the change in ground film reflectivity ΔR under adjacent light intensity conditions i , the expression is: ΔR i =R i+1 -R i ; Among them, R i+1 and R i are the reflectances under two adjacent light intensity levels; and calculate the light intensity change ΔI under adjacent light intensity conditions i , the expression is: ΔI i =I i+1 -I i ; Among them, I i+1 and I iare the light intensities of two adjacent light intensity levels. For each light intensity change interval, the light intensity sensitivity index ISI is calculated. i , the expression is: ISI i Indicates that when the light intensity changes from I i To I i+1 In the process of changes, the sensitivity of the ground film reflectivity is integrated with all light intensity intervals to calculate the overall light intensity sensitivity index ISI avg , the expression is: Where n is the number of light intensity variation intervals.

[0069] The larger the light intensity sensitivity index, the more sensitive the spectral reflectance of the mulch is to changes in light intensity. This means that the spectral characteristics of the mulch at different aging stages are more susceptible to changes in light intensity, resulting in a closer resemblance to the spectral characteristics of soil or plant residues, increasing the possibility of spectral confusion.

[0070] As the light intensity sensitivity index increases, the accuracy of remote sensing data in distinguishing mulch from soil or plant residues may decrease. Since the spectral characteristics change significantly under different light conditions, it is difficult for the classification model to maintain consistent differentiation, increasing the risk of misclassification, which in turn affects the accuracy and reliability of remote sensing monitoring.

[0071] The peak-to-trough displacement fluctuation index and the light intensity sensitivity index are normalized, and the spectral confusion degree anomaly coefficient between the mulch film and the soil or plant residues is calculated based on the normalized peak-to-trough displacement fluctuation index and the light intensity sensitivity index.

[0072] For example, the present invention can use the following formula to calculate the anomaly coefficient of the spectral confusion degree between the ground film and the soil or plant residues, and the calculation expression is: Where, LK is the anomaly coefficient of the spectral confusion between the mulch and the soil or plant residues, ISI is avg is the light intensity sensitivity index, PVSFI is the peak-to-trough displacement fluctuation index, a1 and a2 are the proportional coefficients of the peak-to-trough displacement fluctuation index and the light intensity sensitivity index, and a2>a1>0.

[0073] The obtained spectral confusion degree anomaly coefficient between the mulch film and the soil or plant residue is compared with a preset spectral confusion degree anomaly coefficient reference threshold;

[0074] If the spectral confusion degree anomaly coefficient between mulch and soil or plant residues is greater than or equal to the spectral confusion degree anomaly coefficient reference threshold, this indicates that the spectral characteristics of mulch and soil or plant residues are highly similar under current conditions, resulting in a high degree of spectral confusion between the two. In this case, remote sensing data cannot effectively distinguish between mulch and other substances, and misclassification is prone to occur, so the distinction accuracy is low.

[0075] If the spectral confusion degree anomaly coefficient between mulch film and soil or plant residues is less than the spectral confusion degree anomaly coefficient reference threshold, it means that the difference between the spectral characteristics of mulch film and soil or plant residues is more obvious, and the spectral confusion degree is low. Due to the large difference in spectral characteristics between the two, remote sensing data can more accurately distinguish mulch film from other substances, thereby improving the accuracy of classification.

[0076] S4: If the accuracy of distinction is high, remote sensing data will be directly used for farmland management decisions. If the accuracy of distinction is low, farmland environmental data will be used for auxiliary judgment to enhance the accuracy of distinction.

[0077] When the remote sensing data has a high degree of discrimination accuracy, it means that the spectral characteristics of the mulch film and soil or plant residues can be clearly distinguished, and the risk of misclassification is low. In this case, these accurate data can be directly used for farmland management decisions to assess the status of the mulch film and determine its replacement or maintenance needs, thereby optimizing farmland management efficiency and reducing resource waste.

[0078] If the spectral confusion degree anomaly coefficient between the mulch film and the soil or plant residues is less than the spectral confusion degree anomaly coefficient reference threshold, the remote sensing data has low accuracy in distinguishing between the mulch film and the soil or plant residues. observed , that is, the currently observed spectral reflectance, and the farmland temperature T, farmland humidity H and light intensity L when the spectral data were collected.

[0079] Calculate the temperature correction factor C T , that is, the effect of temperature on spectral reflectance, expressed as: C T =α T ·(T-T0); where α T is the temperature sensitivity coefficient, T0 is the reference temperature (usually the temperature when collecting standard spectrum data). Calculate the humidity correction factor C H , that is, the effect of humidity on spectral reflectance, expressed as: C H =α H ·(H-H0); where α H is the humidity sensitivity coefficient, H0 is the reference humidity. Light intensity correction factor C L , that is, the effect of light intensity on spectral reflectance, expressed as: CL =α L ·(L-L0); where α L is the light sensitivity coefficient, L0 is the reference light intensity; calculate the comprehensive environmental correction factor C env , the expression is: C env =w T ·C T +w H ·C H +w L ·C L ; Among them, w T , w H and w L It is the weight of the temperature, humidity and light intensity correction factor, which is determined according to the importance of environmental data or historical data.

[0080] Use the comprehensive environmental correction factor C env For the observed spectral data R observed Correction is performed to obtain the corrected spectral reflectance R corrected , the expression is: R corrected =R observed -C env ; Output corrected spectral reflectance R corrected , and used for subsequent differentiation and analysis of mulch, soil and plant residues. The spectral confusion degree is recalculated with the corrected data to improve the accuracy of the classification model.

[0081] In this embodiment, multispectral and hyperspectral remote sensing equipment continuously collects spectral data of mulch, soil and plant residues in farmland, and analyzes them in combination with environmental data such as temperature, humidity and light intensity. First, the spectral data is classified and a spectral database is established to distinguish normal and abnormal spectral features. Subsequently, the abnormal spectral features are analyzed to evaluate the degree of spectral confusion between the mulch and soil or plant residues at different aging stages, and determine the accuracy of the distinction. If the accuracy is high, it is directly used for farmland management decisions; if the accuracy is low, the environmental data is used to assist in judgment and enhance the spectral distinction ability.

[0082] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0083] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0084] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0085] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0086] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be covered by the protection scope of the present application.

Claims

1. A remote sensing monitoring method for the years of film laying on farmland, characterized in that: The steps include: S1: Use multispectral and hyperspectral remote sensing equipment to continuously collect spectral data of mulch, soil and plant residues in farmland, and obtain farmland environmental data at different time periods, wherein the farmland environmental data includes temperature, humidity and light intensity; S2: Classify the collected spectral data, establish spectral databases of mulch, soil and plant residues respectively, compare the spectral features of mulch in the current remote sensing data with the mulch features in the preliminary spectral database, and divide them into normal spectral features and abnormal spectral features; S3: Analyze the abnormal spectral characteristics and determine the accuracy of remote sensing data in distinguishing between mulch and soil or plant residues according to the degree of spectral confusion between mulch and soil or plant residues at different aging stages; S4: If the accuracy of distinction is high, remote sensing data will be directly used for farmland management decisions. If the accuracy of distinction is low, farmland environmental data will be used for auxiliary judgment to enhance the accuracy of distinction.

2. The remote sensing monitoring method of the film laying period of farmland according to claim 1 is characterized by: In S2, the similarity of spectral features is compared by support vector machine, and the spectral data is classified into normal features and abnormal features, specifically: Collect spectral data of mulch, soil and plant residues in farmland, and divide the pre-processed spectral data into normal spectral features and abnormal spectral features; The data set is divided into a training set and a test set; feature vectors are extracted from each spectral sample, and the SVM kernel function is selected according to the characteristics of the spectral data; the SVM model is trained using the training set data, and the trained SVM model is applied to the test set to evaluate its classification accuracy. The confusion matrix is ​​used to analyze the specific performance of the model classification results, including the number of correctly classified and misclassified samples, and the spectral feature vector currently collected by remote sensing is input into the trained SVM model; the SVM model outputs the classification result of each spectral feature and gives the confidence of the classification.

3. A remote sensing monitoring method for the years of film laying in farmland according to claim 2, characterized in that: According to the classification results output by SVM, the current spectral features are classified as normal features or abnormal features, specifically: If the confidence is greater than or equal to the preset confidence standard threshold, it is classified as a normal feature and determined as a normal spectrum; if the confidence is less than the preset confidence standard threshold, it is classified as an abnormal feature and determined as an abnormal spectrum.

4. The remote sensing monitoring method of the film laying period of farmland according to claim 1 is characterized in that: In S3, the peak-to-valley displacement fluctuation index is generated according to the change of the peak and trough positions of the spectrum curve, and the method for obtaining the peak-to-valley displacement fluctuation index is: The spectral data of ground film at different aging stages are standardized. On the spectral curve, the peak position is determined by identifying the local maximum point, and the peak position of the i-th band is λ p,i , the position of the trough is determined by identifying the local minimum point, and the trough position of the i-th band is set to λ v,i ; For the spectral curve of the aged mulch film, calculate the displacement of the peak of each band relative to the peak position of the standard mulch film, the expression is: Δλ p,i =|λ p,i,老化 -λ p,i,标准 |; Calculate the displacement of the trough of each band relative to the trough position of the standard ground film, the expression is: Δλ v,i =|λ v,i,老化 -λ v,i,标准 |; Calculate the standard deviation of the peak displacement of all bands and the standard deviation of the trough displacement of all bands respectively, and calculate the peak-trough displacement fluctuation index PVSFI after comprehensive analysis. The expression is: In the formula, σ p and σ v is the standard deviation of the peak and trough displacements.

5. A remote sensing monitoring method for the years of film laying in farmland according to claim 4, characterized in that: In S3, the light intensity sensitivity index is generated according to the influence of different light intensities on the reflection characteristics of the ground film. The method for obtaining the light intensity sensitivity index is: The spectral reflectance of the ground film in different bands is measured by a multispectral camera, and the reflectance value R under each light intensity condition is recorded. i and the corresponding light intensity I i , where i represents different light intensity levels, and calculates the change in ground film reflectivity ΔR under adjacent light intensity conditions i , the expression is: ΔR i =R i+1 -R i ; Among them, R i+1 and R i are the reflectances under two adjacent light intensity levels; and calculate the light intensity change ΔI under adjacent light intensity conditions i , the expression is: ΔI i =I i+1 -I i ; Among them, I i+1 and I i are the light intensities of two adjacent light intensity levels. For each light intensity change interval, the light intensity sensitivity index ISI is calculated. i , the expression is: ISI i Indicates that when the light intensity changes from I i To I i+1 The sensitivity of the ground film reflectivity during the change of light intensity is calculated by integrating all light intensity intervals to calculate the overall light intensity sensitivity index ISI. avg , the expression is: Where n is the number of light intensity variation intervals.

6. A remote sensing monitoring method for the years of film laying in farmland according to claim 5, characterized in that: In S3, the peak-to-trough displacement fluctuation index and the light intensity sensitivity index are normalized, and the spectral confusion degree anomaly coefficient between the ground film and the soil or plant residue is calculated by the normalized peak-to-trough displacement fluctuation index and the light intensity sensitivity index.

7. A remote sensing monitoring method for the years of film laying in farmland according to claim 6, characterized in that: The obtained spectral confusion degree anomaly coefficient between the mulch film and the soil or plant residue is compared with a preset spectral confusion degree anomaly coefficient reference threshold; If the spectral confusion degree anomaly coefficient between the ground film and soil or plant residues is greater than or equal to the spectral confusion degree anomaly coefficient reference threshold, it means that the spectral confusion degree between the ground film and soil or plant residues is high, and the accuracy of remote sensing data in distinguishing ground film and soil or plant residues is low; if the spectral confusion degree anomaly coefficient between the ground film and soil or plant residues is less than the spectral confusion degree anomaly coefficient reference threshold, it means that the spectral confusion degree between the ground film and soil or plant residues is low, and the accuracy of remote sensing data in distinguishing ground film and soil or plant residues is high.

8. The remote sensing monitoring method of the agricultural film laying years according to claim 1 is characterized by: In S4, if the distinction accuracy is low, the farmland environmental data is used for auxiliary judgment to enhance the distinction accuracy, specifically: If the spectral confusion degree anomaly coefficient between the mulch film and the soil or plant residues is less than the spectral confusion degree anomaly coefficient reference threshold, the remote sensing data has low accuracy in distinguishing between the mulch film and the soil or plant residues. observed , that is, the currently observed spectral reflectance, and the farmland temperature T, farmland humidity H and light intensity L when the spectral data were collected; Calculate the temperature correction factor C T , that is, the effect of temperature on spectral reflectance, expressed as: C T =α T ·(T-T0); where α T is the temperature sensitivity coefficient, T0 is the reference temperature, and the humidity correction factor C is calculated. H , that is, the effect of humidity on spectral reflectance, expressed as: C H =α H ·(H-H0); where α H is the humidity sensitivity coefficient, H0 is the reference humidity; the light intensity correction factor C L , that is, the effect of light intensity on spectral reflectance, expressed as: C L =α L ·(L-L0); where α L is the light sensitivity coefficient, L0 is the reference light intensity; calculate the comprehensive environmental correction factor C env , the expression is: C env =w T ·C T +w H ·C H +w L ·C L ; Among them, w T , w H and w L is the weight of the temperature, humidity and light intensity correction factors, using the comprehensive environmental correction factor C env For the observed spectral data R observed Correction is performed to obtain the corrected spectral reflectance R corrected , the expression is: R corrected =R observed -C env .

Images