Land degradation trend analysis method and system based on remote sensing

By performing spectral analysis and time series analysis on multi-time phase remote sensing data, combining variable point detection and trend slope, a land cover prediction model is generated, which solves the problem of insensitive land degradation monitoring in the existing technology and achieves a more accurate land degradation trend Analyze and forecast to support more effective land management decisions.

CN119007023AInactive Publication Date: 2024-11-22DALIAN WANLAI CONSTR ENG CO LTD
View PDF 0 Cites 7 Cited by

Patent Information

Application Number
CN202411488480.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-24
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology does not interpret long-term change trends in processing multi-phase data, and lacks the ability to capture subtle changes, which leads to insensitive monitoring of land degradation and the inability to timely identify early signals of degradation, affecting decision-making in agriculture and urban planning.

Method used

Using remote sensing-based land degradation trend analysis method, the spectral reflectance changes of pixel points are calculated by spectral analysis of multi-time phase remote sensing data, time series analysis and variable point detection are applied, and land cover prediction is generated based on trend slope and seasonal parameters. Model, predict future changes in land cover types.

Benefits of technology

It significantly improves the accuracy of land degradation monitoring, detects signs of land degradation early, reduces random fluctuations and improves the stability of results, helps to quickly adjust land management strategies, and optimizes the accuracy of resource allocation and planning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007023A_ABST
    Figure CN119007023A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of image analysis, in particular to a land degradation trend analysis method and system based on remote sensing, and the method comprises the following steps: based on multi-temporal remote sensing data of a target area, carrying out the spectral analysis of each pixel point, and calculating the spectral reflectivity change of the pixel point at a plurality of time points; according to the method, the precision of land degradation monitoring is improved by performing detailed spectral analysis and time sequence analysis on the multi-temporal remote sensing data, the time change of the spectral reflectivity data is accurately calculated to help to find land degradation signs as early as possible, the main trend is analyzed by using the moving average value, the interference of random fluctuation is effectively reduced, and the accuracy of land degradation monitoring is improved. According to the method, the stability of a result is improved, key change points can be marked through change point detection, understanding of land degradation dynamics is improved, a land management strategy can be rapidly adjusted, and through further spectral difference data classification and feature extraction, future land coverage type changes can be predicted, and the accuracy of resource allocation and planning can be optimized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a land degradation trend analysis method and system based on remote sensing. Background Art

[0002] The field of image analysis technology covers a wide range of techniques from basic image processing to advanced image recognition, detection, and classification. These techniques aim to extract useful information from digital images and transform them into structured data for understanding the image content. Image analysis methods include, but are not limited to, edge detection, feature extraction, pattern recognition, and the application of machine learning algorithms, which are used to enhance, interpret, and quantify image data. Technologies in this field are widely used in medical imaging, satellite remote sensing, industrial inspection, security monitoring, and autonomous driving systems, providing a powerful tool to support decision-making and analysis.

[0003] Among them, the land degradation trend analysis method refers to the use of image analysis technology to monitor and evaluate changes in land cover and land use over time, especially focusing on land degradation. This method usually uses remote sensing images to compare surface images at different time points to analyze changes in land cover and identify degradation trends. The uses of this technology include environmental monitoring, agricultural management, urban planning, and natural resource management, supporting land management decisions and environmental protection measures by providing timely data.

[0004] Although existing technologies support basic image analysis, common problems when processing multi-temporal data are insufficient interpretation of long-term trends and lack of ability to capture subtle changes. The shortcomings of this method in spectral data analysis lead to insensitive land degradation monitoring and failure to identify early signs of degradation in a timely manner. For example, in agricultural management, failure to identify land degradation in a timely manner may lead to reduced crop yields, affecting food security and agricultural sustainability. In urban planning and resource management, the shortcomings of this technology may lead to planning errors and increase the risk of damage to ecosystems. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a land degradation trend analysis method and system based on remote sensing.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a land degradation trend analysis method based on remote sensing, comprising the following steps: S1: Based on the multi-temporal remote sensing data of the target area, perform spectral analysis on each pixel point, calculate the spectral reflectance changes of the pixel point at multiple time points, record the corresponding time series data, and generate initial time series difference data; S2: applying time series analysis to the initial time series difference data, calculating the moving average of each pixel point, extracting the trend component of each data point, and generating a trend analysis result; S3: Using the trend analysis result to perform change point detection, identify the change point in the time series of each pixel point, record the time position of each change point, and compare the data difference before and after the change point to generate a change point detection result; S4: Using the change point detection results, sorting out the spectral difference data before and after the change point, combining the trend slope and seasonal parameters, performing pixel classification and feature extraction, predicting future land cover type changes, and generating a land cover prediction model.

[0007] As a further solution of the present invention, the step of acquiring the initial time series difference data is specifically as follows: S111: Collect multi-temporal remote sensing data in the target area, perform spectrometer measurement, obtain spectral reflectance data of each pixel at multiple time points, and generate a multi-temporal spectral data set; S112: Analyze the multi-phase spectral data set, calculate the spectral change between any two consecutive time points, including the spectral reflectance difference of each pixel point, and record the spectral change value; S113: Integrate the spectral change values ​​using the formula: ; The time series spectrum difference of each pixel point is calculated to generate initial time series difference data; in, Represents time series difference data, Representing time point The spectral reflectance data, represents the absolute value difference, is the number of time points involved in the calculation, means taking square roots to smooth the differences.

[0008] As a further solution of the present invention, the step of obtaining the trend analysis result is specifically: S211: Select each pixel point data extracted from the initial time series difference data, apply the time series analysis method, perform moving average calculation, and generate moving average data by weighted smoothing of the data points; S212: From the moving average data, by calculating the long-term trend at each time point, extracting key trend changes, and generating main trend data; S213: Based on the main trend data, the formula is adopted: ; Estimate trends in future time periods and generate trend analysis results; in, Indicates the trend analysis results. Before The main trend data at a point in time, Represents the weighting factor, ensuring that the farther the time is from the current time, the smaller the weight. is the total number of weights, Used to enhance the impact of long-term data.

[0009] As a further solution of the present invention, the step of obtaining the change point detection result is specifically: S311: Apply the trend analysis results, deploy a change point detection algorithm to identify changes in the time series, calibrate potential change points by comparing data fluctuations at consecutive time points, and generate a preliminary change point list; S312: Determine the time position of each change point from the preliminary change point list, record the data difference before and after each change point, plot the time and associated impact of the change point, and generate a change point time position record; S313: Integrate the change point time position records and calculate the significance of each change point using the formula: ; Generate change point detection results; in, Indicates time point The data value of For weight, highlight the time point and The importance of the difference between is a constant used to balance the denominator, Represents the change point significance score to ensure that the score reflects the importance of the change point.

[0010] As a further solution of the present invention, the steps of obtaining the land cover prediction model are specifically as follows: S411: sorting the change point detection results, extracting spectral difference data of each pixel point, including the spectral reflectance difference before and after the change point, and generating spectral difference data; S412: Based on the spectral difference data, a time series decomposition algorithm is used to analyze trend slope and seasonal parameters, and pixels are classified, each class representing the trend and periodic behavior of a land cover type, and classification and feature data are generated; S413: Process the classification and feature data to predict the change of land cover type in the future time period, using the formula: ; Generate land cover prediction models; in, represents the trend slope data, represents the seasonal parameter, Represents the characteristics of neighboring pixels and enhances the model's ability to predict future changes. represents the output of the prediction model, , , Adjust the weights of multiple factors in the prediction.

[0011] The time series decomposition algorithm decomposes the spectral difference data into three parts: long-term trend, seasonal cycle and noise, according to the formula: ; Calculate trend slopes and seasonal parameters for land cover types; in, Representing time point The spectral data of Representing time point The long-term trend Representing time point The seasonal components, Representing time point The noise, It is a coefficient for adjusting the degree of noise influence and is used to scale the noise. Is with the cycle The associated adjustment factor is used to control the strength of the seasonal cycle, is the noise smoothing coefficient, which is used to reduce the effect of noise when its influence is small. Represents the length of the seasonal cycle.

[0012] A land degradation trend analysis system based on remote sensing, the land degradation trend analysis system based on remote sensing is used to execute the land degradation trend analysis method based on remote sensing, the system comprises: The multi-temporal data processing module obtains multi-temporal image data of the target area from the remote sensing satellite, reads the spectral reflectance data of each pixel in the image, records the changes in spectral reflectance at multiple time points, calculates the initial time series difference data, and generates time series spectral difference records; The trend analysis module performs moving average calculation based on the time series spectrum difference record, performs main trend extraction on each data point, applies linear regression analysis to the time series data, and obtains the main trend analysis result; The change point detection module uses the main trend analysis results to perform change point detection, identifies significant change points in the time series of each pixel point through statistical tests, records the time position of each change point, compares the data before and after the change point, analyzes the intensity and direction of data fluctuations, and generates change point time records and change intensity analysis; The land cover prediction module utilizes the change point time record and change intensity analysis, integrates land seasonal change parameters, classifies pixels through cluster analysis, extracts feature data, establishes a land cover type prediction model, and generates land cover prediction results.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the accuracy of land degradation monitoring is significantly improved by performing detailed spectral analysis and time series analysis on multi-temporal remote sensing data. By accurately calculating the temporal changes in spectral reflectance data, it helps to detect signs of land degradation early. The use of moving averages to analyze the main trend effectively reduces the interference of random fluctuations and improves the stability of the results. Change point detection can mark key change points, improve the understanding of land degradation dynamics, and help to quickly adjust land management strategies. Through further spectral difference data classification and feature extraction, the solution can predict future changes in land cover types and optimize the accuracy of resource allocation and planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2 This is a flow chart of the steps for obtaining the initial timing difference data of the present invention; Figure 3 A flow chart of steps for obtaining trend analysis results of the present invention; Figure 4 Flow chart of the steps for obtaining the change point detection results of the present invention; Figure 5 The figure is a flow chart of the steps for obtaining the land cover prediction model of the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0016] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0017] Example 1: Please refer to Figure 1 The present invention provides a technical solution: a land degradation trend analysis method based on remote sensing, comprising the following steps: S1: Based on the multi-temporal remote sensing data of the target area, perform spectral analysis on each pixel point, calculate the spectral reflectance changes of the pixel point at multiple time points, record the corresponding time series data, and generate initial time series difference data; S2: Apply time series analysis to the initial time series difference data, calculate the moving average of each pixel, extract the trend component of each data point, and generate trend analysis results; S3: Use the trend analysis results to perform change point detection, identify the change point in the time series of each pixel point, record the time position of each change point, and compare the data difference before and after the change point to generate the change point detection result; S4: Using the change point detection results, the spectral difference data before and after the change point are sorted out, and combined with the trend slope and seasonal parameters, pixel classification and feature extraction are performed to predict future land cover type changes and generate a land cover prediction model.

[0018] The initial time series difference data include time point marks and spectral reflectance difference records. The trend analysis results include moving average and main trend. The change point detection results specifically refer to the time position of the change point and data difference comparison. The land cover prediction model includes pixel classification, feature extraction, trend slope and seasonal parameters.

[0019] See also Figure 2 , the specific steps for obtaining the initial time series difference data are: S111: Collect multi-temporal remote sensing data in the target area, perform spectrometer measurement, obtain spectral reflectance data of each pixel at multiple time points, and generate a multi-temporal spectral data set; After selecting the target area and collecting multi-temporal remote sensing data, spectrometer measurements were performed to obtain the spectral reflectance data of each pixel at multiple time points, which was obtained through real-time monitoring. The multi-temporal spectral data set includes the original data obtained from the remote sensing satellite. Spectral analysis requires time series comparison of the data. By comparing the spectral resolution, the changes in material composition at multiple time points can be identified. The spectral data of each pixel includes its spectral intensity and wavelength information, which is directly measured by the spectrometer, including visible light to near-infrared bands, which lays the foundation for the next step of spectral difference data calculation.

[0020] S112: Analyze the multi-phase spectral data set, calculate the spectral change between any two consecutive time points, including the spectral reflectance difference of each pixel point, and record the spectral change value; When analyzing multi-temporal spectral data sets, the spectral reflectance data of each pixel between two consecutive time points is first extracted and recorded. This step ensures that the data at each time point can be captured. Then, by comparing the spectral differences between consecutive time points, we focus on the amount of change in the pixel points and evaluate the direction and magnitude of the change. This method reveals potential trends and abnormal changes in the data. The recording of spectral changes is crucial to subsequent data analysis and provides the necessary basis for identifying and resolving pattern changes in spectral data. Finally, the recorded spectral change values ​​will be used for iterative analysis and model building to provide a basis for predicting future spectral behavior.

[0021] S113: Integrate the spectral change value using the formula: ; The time series spectrum difference of each pixel point is calculated to generate initial time series difference data; in, Represents time series difference data, Representing time point The spectral reflectance data, represents the absolute value difference, is the number of time points involved in the calculation, means taking square roots to smooth the differences.

[0022] formula: ; The benefit of this formula is that it smoothes out extreme variations in spectral data by taking the root of the sum of squares, reducing the impact of outliers on the overall difference assessment.

[0023] Detailed explanation of the formula and the process of formula calculation and derivation: Set at a certain target pixel point, time point Spectral reflectance data is 150, and at time point Spectral reflectance data is 160, then the difference is: ; If there are 5 such consecutive time points, the sum of the squares of the differences is: ; So the timing difference data: ; The results show that during the survey period, the average change in spectral reflectance data between each time point was 4.47 units. The numerical method assisted in the analysis of the average intensity of spectral differences in the time series, which can be used to monitor environmental changes or other key surface characteristics.

[0024] See also Figure 3 , the steps to obtain the trend analysis results are as follows: S211: Select each pixel point data extracted from the initial time series difference data, apply the time series analysis method, perform moving average calculation, and generate moving average data by weighted smoothing of the data points; Select the initial time series difference data, execute the moving average calculation method, and smooth the data points through the weighted method to reduce the impact of random fluctuations. During the calculation process, the size of the reference time window is crucial to the result. If the time window is too small, the data cannot be effectively smoothed, and if it is too large, key trend changes will be obscured. The most matching window size is determined through the frequency distribution and volatility analysis of the actual data to ensure that the smooth trend of each pixel point can be obtained through this weighted moving average method. This trend will serve as the basis for analyzing long-term trends.

[0025] S212: extracting key trend changes from the moving average data by calculating the long-term trend at each time point to generate main trend data; Based on the moving average data, the main trend is extracted, and the key trend changes are extracted by calculating the long-term trend components to determine the dominant trend of the data. By detecting the seasonal and cyclical components in the actual data, time series analysis methods including ARIMA or seasonal decomposition methods are used to model the components, and then the main trend of each data point is extracted and quantified. This main trend reflects the overall movement direction of the data and provides a quantitative basis for subsequent trend analysis.

[0026] S213: Comprehensive main trend data, using the formula: ; Estimate trends in future time periods and generate trend analysis results; in, Indicates the trend analysis results. Before The main trend data at a point in time, Represents the weighting factor, ensuring that the farther the time is from the current time, the smaller the weight. is the total number of weights, Used to enhance the impact of long-term data.

[0027] formula: ; The benefit of the formula is that by introducing square terms and adjustment coefficients, it increases sensitivity and responsiveness to data trends, can more carefully reflect the impact of data fluctuations on trends, and enhance the accuracy of predictions.

[0028] Detailed explanation of the formula and the process of formula calculation and derivation: set up , , , , , , first calculate the denominator of the weight adjustment: ; Then calculate the numerator of the weighted trend value: ; so, ; The results show that based on the current and past data points, with reference to the weights between the data points and the data volatility, the trend analysis result is 89, which means that in the current time series analysis, the trend value is 89, which is a comprehensive trend indicator based on past data and adjustment coefficients. It reflects the main trend of the data points in the time series and provides a quantitative basis for data analysis and decision-making.

[0029] See also Figure 4 , the specific steps for obtaining the change point detection results are: S311: Apply the trend analysis results, deploy the change point detection algorithm to identify changes in the time series, calibrate potential change points by comparing the data fluctuations at consecutive time points, and generate a preliminary change point list; The change point detection algorithm is deployed to identify changes in time series. By comparing the data fluctuations between consecutive time points, the change points are calibrated according to the difference greater than the preset sensitivity threshold. The threshold is determined based on the average value of the data change in the previous consecutive periods and matched with the reference environmental noise or potential data errors, thereby ensuring the sensitivity and accuracy of the algorithm to the real change points. In this way, the generated preliminary change point list not only reflects the mutation of the time series, but also excludes the impact caused by random fluctuations or occasional errors in data collection.

[0030] S312: from the preliminary change point list, determine the time position of each change point, record the data difference before and after each change point, plot the time and associated impact of the change point, and generate a change point time position record; Determine the time position of each change point from the generated preliminary change point list, record the data difference before and after the change point, record the time and data changes of each change point, including comparative analysis of the data before and after the change point, and use quantitative data comparison, including the magnitude and rate of change, to ensure that the importance and scope of influence of each change point can be accurately identified. This will screen out the truly critical change points and provide accurate input for subsequent data processing and analysis.

[0031] S313: Integrate the time position records of the change points and calculate the significance of each change point using the formula: ; Generate change point detection results; in, Indicates time point The data value of For weight, highlight the time point and The importance of the difference between is a constant used to balance the denominator, Represents the change point significance score to ensure that the score reflects the importance of the change point.

[0032] formula: ; The benefit of the formula is that it can quantify the significance of the change point through weighted differences, refer to the local characteristics of time series data, enhance the sensitivity and reaction speed of the model to local mutations, and at the same time, through the parameter Adjust to enhance the adaptability and stability of the model in differentiated data environments.

[0033] Detailed explanation of the formula and the process of formula calculation and derivation: For the change point detection example, let’s set a time series in arrive There is a significant change during the period. The data difference before and after the change point is calculated according to |120-100|, |130-80|, |140-70|. All are 1, .

[0034] ; The results show that in arrive During the time period, the score of significant change is 0.9333, indicating that this is a highly significant change point, which has key reference value for subsequent data analysis and decision-making.

[0035] See also Figure 5 , the specific steps for obtaining the land cover prediction model are: S411: sorting the change point detection results, extracting the spectral difference data of each pixel point, including the spectral reflectance difference before and after the change point, and generating spectral difference data; In the process of collating the change point detection results, we first need to extract the spectral difference data of each pixel. The key to this step is to determine which changes are significant and which are just noise by comparing the changes in the spectral reflectance data before and after the change point. Next, record the data to provide the necessary input information for subsequent pixel classification and feature extraction. By monitoring the changes in real time, the real-time and relevance of the data are ensured, so that the subsequent steps can be carried out according to the actual observed spectral changes when processing the data.

[0036] S412: Based on the spectral difference data, a time series decomposition algorithm is used to analyze the trend slope and seasonal parameters, classify the pixels, and each class represents the trend and periodic behavior of a land cover type, and generate classification and feature data; Based on the spectral difference data, analyzing the trend slope and seasonal parameters of each pixel is the core task of this section. This analysis includes combining the trend slope with the seasonal change data through a mathematical model. The classification algorithm divides the pixels into differentiated land cover types based on the comprehensive data. Each type represents a target land use method. During the classification process, special attention is paid to the accurate entry and correct processing of the data to ensure the reliability and accuracy of the classification results, thereby providing a stable foundation for the land cover prediction model.

[0037] S413: Process the classification and feature data to predict the change of land cover type in the future time period, using the formula: ; Generate land cover prediction models; in, represents the trend slope data, represents the seasonal parameter, Represents the characteristics of neighboring pixels and enhances the model's ability to predict future changes. represents the output of the prediction model, , , Adjust the weights of multiple factors in the prediction.

[0038] formula: ; The usefulness of the formula is that by taking the trend slope ( ), seasonal parameters ( ) and the neighboring pixel data ( ) can comprehensively refer to the immediate changes and long-term periodic characteristics of the time series, thereby more accurately predicting land cover changes in future time periods.

[0039] Detailed explanation of the formula and the process of formula calculation and derivation: set up , ,for , , weight coefficient , , , .

[0040] ; ; ; The results show that with the given weight and parameter settings, the model predicts a land cover change index of 0.77, which means that based on the current trend slope, seasonal parameters, and data from neighboring pixels, the land cover type has a high probability of changing in the future.

[0041] The time series decomposition algorithm decomposes the spectral difference data into three parts: long-term trend, seasonal cycle and noise, according to the formula: ; Calculate trend slopes and seasonal parameters for land cover types; in, Representing time point The spectral data of Representing time point The long-term trend Representing time point The seasonal components, Representing time point The noise, It is a coefficient for adjusting the degree of noise influence and is used to scale the noise. Is with the cycle The associated adjustment factor is used to control the strength of the seasonal cycle, is the noise smoothing coefficient, which is used to reduce the effect of noise when its influence is small. Represents the length of the seasonal cycle.

[0042] formula: ; Detailed explanation of the formula and the process of formula calculation and derivation: Parameter acquisition method and value: Indicates time point The spectral data is obtained through spectral analysis of remote sensing images, taking a pixel point in a certain area at a certain time point Spectral reflectance value, assuming that time point The spectral data is (unit is reflectivity); Indicates time point The long-term trend can be obtained by moving the average of long-term historical spectral data, using a 5-year moving average to obtain ; Indicates time point The seasonal component is obtained by periodically decomposing the spectral data and using Fourier transform to decompose the data. Assuming that the Fourier transform is obtained at time point The seasonal component of ; Indicates time point The noise term can be determined by residual analysis, using the residual after subtracting the spectral difference data from the trend and seasonal components as the noise, assuming that the noise is ; To adjust the coefficient of noise impact, it is set according to the stability of land cover type and noise level. , this value is set according to historical experience data. When the noise is large, this value needs to be increased to reduce its impact; is the adjustment coefficient related to the seasonal cycle, which is used to control the cycle impact. , this coefficient is set based on the importance of seasonal effects compared to the overall trend; is the length of the seasonal cycle, which is obtained by analyzing the periodic variation characteristics in the data and determining that the cycle is one year, that is, ; is the noise smoothing coefficient, which is used to reduce the influence of small noise. , based on the analysis of the impact of noise on the overall calculation.

[0043] The calculation and derivation process of the formula: First, put all known parameters into the formula: ; Substitute the known values: ; Perform step-by-step calculations: First calculate the square root term: ; Then calculate the denominator: ; Calculate the components of the noise: ; Calculate the final : ; Result analysis: The results show that the time point Spectral data The value is 0.6377, which includes the influence of long-term trend, seasonal component and noise. Through calculation, it can be found that the influence of noise on spectral data is properly adjusted and smooth, and the long-term trend and seasonal component are reasonably separated and weighted integrated. This value can be used to further analyze the changing characteristics of land cover types, which is helpful for accurate classification and modeling of land cover types.

[0044] A land degradation trend analysis system based on remote sensing, the land degradation trend analysis system based on remote sensing is used to execute the above land degradation trend analysis method based on remote sensing, the system comprises: The multi-temporal data processing module obtains multi-temporal image data of the target area from the remote sensing satellite, reads the spectral reflectance data of each pixel in the image, records the changes in spectral reflectance at multiple time points, calculates the initial time series difference data, and generates time series spectral difference records; The trend analysis module performs moving average calculation based on the time series spectrum difference records, performs main trend extraction for each data point, applies linear regression analysis to the time series data, and obtains the main trend analysis results; The change point detection module uses the main trend analysis results to perform change point detection. It identifies the significant change points in the time series of each pixel point through statistical tests, records the time position of each change point, compares the data before and after the change point, analyzes the intensity and direction of data fluctuations, and generates change point time records and change intensity analysis; The land cover prediction module uses change point time records and change intensity analysis to integrate land seasonal change parameters, classifies pixels through cluster analysis, extracts feature data, establishes a prediction model for land cover types, and generates land cover prediction results.

[0045] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A land degradation trend analysis method based on remote sensing, characterized in that: The following steps are involved: Based on the multi-temporal remote sensing data of the target area, spectral analysis is performed on each pixel point, the spectral reflectance changes of the pixel point at multiple time points are calculated, the corresponding time series data are recorded, and the initial time series difference data is generated; Applying time series analysis to the initial time series difference data, calculating the moving average of each pixel point, extracting the trend component of each data point, and generating a trend analysis result; Using the trend analysis results to perform change point detection, identify the change point in the time series of each pixel point, record the time position of each change point, and compare the data difference before and after the change point to generate a change point detection result; The change point detection results are used to sort out the spectral difference data before and after the change point, and combined with the trend slope and seasonal parameters, pixel classification and feature extraction are performed to predict future land cover type changes and generate a land cover prediction model.

2. The land degradation trend analysis method based on remote sensing according to claim 1, characterized in that: The initial time series difference data includes time point marks and spectral reflectance difference records. The trend analysis results include moving averages and main trends. The change point detection results specifically refer to the change point time positions and data difference comparisons. The land cover prediction model includes pixel classification, feature extraction, trend slopes, and seasonal parameters.

3. The land degradation trend analysis method based on remote sensing according to claim 1, characterized in that: The steps for obtaining the initial time series difference data are specifically as follows: Collect multi-temporal remote sensing data in the target area, perform spectrometer measurements, obtain spectral reflectance data for each pixel at multiple time points, and generate a multi-temporal spectral dataset; Analyze the multi-phase spectral data set, calculate the spectral change between any two consecutive time points, including the spectral reflectance difference of each pixel point, and record the spectral change value; Integrate the spectral change values ​​using the formula: ; The time series spectrum difference of each pixel point is calculated to generate initial time series difference data; in, Represents time series difference data, Representing time point The spectral reflectance data, represents the absolute value difference, is the number of time points involved in the calculation, means taking square roots to smooth the differences.

4. The land degradation trend analysis method based on remote sensing according to claim 3 is characterized in that: The steps for obtaining the trend analysis results are specifically as follows: Selecting each pixel point data extracted from the initial time series difference data, applying a time series analysis method, performing a moving average calculation, and generating moving average data by weighted smoothing of the data points; From the moving average data, by calculating the long-term trend at each time point, key trend changes are extracted to generate main trend data; Based on the main trend data, the formula is adopted: ; Estimate trends in future time periods and generate trend analysis results; in, Indicates the trend analysis results. Before The main trend data at a point in time, Represents the weighting factor, ensuring that the farther the time is from the current time, the smaller the weight. is the total number of weights, Used to enhance the impact of long-term data.

5. The land degradation trend analysis method based on remote sensing according to claim 4 is characterized in that: The steps for obtaining the change point detection result are specifically as follows: Applying the trend analysis results, deploying a change point detection algorithm to identify changes in the time series, calibrating potential change points by comparing data fluctuations at consecutive time points, and generating a preliminary change point list; From the preliminary change point list, determine the time position of each change point, record the data difference before and after each change point, plot the time and associated impact of the change point, and generate a change point time position record; Integrate the change point time position records and calculate the significance of each change point using the formula: ; Generate change point detection results; in, Indicates time point The data value of For weight, highlight the time point and The importance of the difference between Indicates the calculation time point The number of surrounding time points considered when evaluating the significance of a change point. is a constant used to balance the denominator, Represents the change point significance score to ensure that the score reflects the importance of the change point.

6. The land degradation trend analysis method based on remote sensing according to claim 5, characterized in that: The steps for obtaining the land cover prediction model are specifically as follows: Arranging the change point detection results, extracting spectral difference data of each pixel point, including the spectral reflectance difference before and after the change point, and generating spectral difference data; Based on the spectral difference data, a time series decomposition algorithm is used to analyze the trend slope and seasonal parameters, and the pixels are classified, each class representing the trend and periodic behavior of a land cover type, and classification and feature data are generated; The classification and characteristic data are processed to predict the change of land cover type in the future time period, using the formula: ; Generate land cover prediction models; in, represents the trend slope data, represents the seasonal parameter, Represents the characteristics of neighboring pixels and enhances the model's ability to predict future changes. represents the output of the prediction model, , , Adjust the weights of multiple factors in the prediction.

7. The land degradation trend analysis method based on remote sensing according to claim 6, characterized in that: The time series decomposition algorithm decomposes the spectral difference data into three parts: long-term trend, seasonal cycle and noise, according to the formula: ; Calculate trend slopes and seasonal parameters for land cover types; in, Representing time point The spectral data of Representing time point The long-term trend Representing time point The seasonal components, Representing time point The noise, It is a coefficient for adjusting the degree of noise influence and is used to scale the noise. Is with the cycle The associated adjustment factor is used to control the strength of the seasonal cycle, is the noise smoothing coefficient, which is used to reduce the effect of noise when its influence is small. Represents the length of the seasonal cycle.

8. A land degradation trend analysis system based on remote sensing, characterized in that: According to the remote sensing-based land degradation trend analysis method according to any one of claims 1 to 5, the system comprises: The multi-temporal data processing module obtains multi-temporal image data of the target area from the remote sensing satellite, reads the spectral reflectance data of each pixel in the image, records the changes in spectral reflectance at multiple time points, calculates the initial time series difference data, and generates time series spectral difference records; The trend analysis module performs moving average calculation based on the time series spectrum difference record, performs main trend extraction on each data point, applies linear regression analysis to the time series data, and obtains the main trend analysis result; The change point detection module uses the main trend analysis results to perform change point detection, identifies significant change points in the time series of each pixel point through statistical tests, records the time position of each change point, compares the data before and after the change point, analyzes the intensity and direction of data fluctuations, and generates change point time records and change intensity analysis; The land cover prediction module utilizes the change point time record and change intensity analysis, integrates land seasonal change parameters, classifies pixels through cluster analysis, extracts feature data, establishes a land cover type prediction model, and generates land cover prediction results.

Citation Information

Cited By

  • Engineering surveying and mapping method and system based on unmanned aerial vehicle remote sensing

    CN119374566A

  • Land coverage change monitoring method, model training method and related device

    CN119380204A

  • Land cover change monitoring method, model training method and related device

    CN119380204B

  • Urban green land invasion and degradation identification system and method based on computer visual identification

    CN119785211A

  • Terrain recognition and three-dimensional reconstruction method based on remote sensing surveying and mapping image

    CN120071163A