Dynamic remote sensing monitoring method and system based on land utilization

Abstract

The invention belongs to the technical field of electric digital data processing, and discloses a dynamic remote sensing monitoring method and system based on land utilization. The method comprises the steps of obtaining a multi-source heterogeneous time sequence data set, and generating standardized spatio-temporal data through preprocessing; extracting space-time correlation features based on the dynamic graph convolutional network, and generating dynamic mode data through multi-dimensional feature fusion; performing classification decision and state transition analysis on the dynamic mode data, identifying a state change area and generating a dynamic analysis report; constructing an interactive visualization engine to map a space-time thermodynamic diagram and a trend graph; and in combination with the incremental learning optimization model, outputting a long-term evolution prediction result. Through the dynamic graph convolutional network and the adaptive cross-modal alignment technology, the problems that multi-source data space-time correlation modeling efficiency is low and the noise suppression capability is insufficient are solved, the dynamic mode recognition precision is remarkably improved, real-time interaction analysis and long-term prediction are supported, and the method is suitable for accurate decision making of industrial monitoring, traffic planning and other scenes.

Description

Technical Field

[0001] The present application relates to the technical field of electronic digital data processing, and particularly to a dynamic remote sensing monitoring method and system based on land use. Background Art

[0002] With the rapid development of Internet of Things, sensor network and distributed computing technologies, the scale and complexity of multi-source heterogeneous data have increased exponentially. In the fields of industrial equipment monitoring, traffic flow analysis, medical health management, etc., the real-time acquisition, fusion and analysis of dynamic spatio-temporal data have become core requirements. However, there are significant bottlenecks in the integration efficiency of multi-source heterogeneous data, the accuracy of dynamic pattern recognition and the scalability of the system in the existing technologies, making it difficult to meet the high-timeliness and high-concurrency data analysis requirements.

[0003] 1. Problems of data heterogeneity and integration efficiency. Existing technologies usually rely on a single data source or simple rules to achieve multi-modal data alignment (such as the temporal matching of sensor data and log data), lacking the ability of dynamic modeling for cross-modal relevance. For example: Data noise suppression: Traditional filtering algorithms (such as mean filtering) need to preset fixed parameters and cannot be dynamically adjusted according to the data distribution, resulting in a significant decrease in the signal-to-noise ratio in high-noise scenarios.

[0004] Cross-modal alignment: Data fusion methods based on static coordinate systems (such as rigid registration) are difficult to adapt to the spatio-temporal drift of multi-source data (such as differences in sensor sampling frequencies and spatial resolutions), resulting in the accumulation of alignment errors.

[0005] 2. Insufficient real-time performance and generalization ability of dynamic pattern recognition. Existing dynamic analysis models (such as the traditional time series prediction algorithm ARIMA) usually based on fixed time windows or static feature extraction have the following defects: Limitations in spatio-temporal correlation modeling: Discrete time series analysis methods (such as sliding window statistics) cannot capture long-range spatio-temporal dependence relationships, resulting in a high missed detection rate for complex dynamic patterns (such as cross-sensor propagation of equipment failures).

[0006] Weak model generalization ability: Models trained for specific scenarios (such as a single industrial equipment) are difficult to adapt to other fields (such as medical monitoring), and it is necessary to repeatedly design feature engineering and model architectures, resulting in high development costs. Summary of the Invention

[0007] The present application provides a dynamic remote sensing monitoring method and system based on land use to solve the problem of how to achieve cross-modal dynamic spatio-temporal correlation modeling and evolution law analysis based on multi-source heterogeneous time series data (sensor data, spatial coordinate data and time series data), and improve the real-time decision-making accuracy and system scalability in complex scenarios.

[0008] To solve the above problems, the present application provides a dynamic remote sensing monitoring method based on land use, including: S100: Obtain a multi-source heterogeneous time-series dataset, preprocess the dataset, and generate standardized spatio-temporal data; S200: Extract spatio-temporal correlation features from the standardized spatio-temporal data based on a dynamic graph convolutional network, perform multi-dimensional feature fusion using a dynamic pattern recognition model, and generate dynamic pattern data of the target object; S300: Make a classification decision on the dynamic pattern data to generate a multi-period state analysis result of the target object, including: S310: Model the probability distribution of the dynamic pattern data through a classification model to obtain classification results at each spatial location in a continuous time period; The probability distribution of the classification result is defined by the following formula: ; where, is spatio-temporal pattern data, is the inner product after feature mapping, representing the weight of the spatio-temporal feature in the classification decision function, is the prediction probability of the land use category, is the number of categories of the classification model, is the bias term; S320: Perform state transition analysis on the classification results in a continuous time period to identify the state change area of the target object, and its change rate is calculated as: ; where, represents time, is the land use type at the same location in different time periods, represents the dynamic pattern analysis rate of this location; And conduct classification statistics on the dynamic pattern analysis: ; where, represents the state of different change types mapped to spatial positions, is the indicator function; S330: Make a classification decision on the dynamic pattern data to generate a multi-period state analysis result and a dynamic analysis report of the target object; S400: Construct an interactive visualization engine to map the dynamic analysis report into spatio-temporal data visualization and trend graphs; S500: Optimize the dynamic pattern recognition model based on the incremental learning of historical data and real-time data to generate a long-term evolution prediction result of the target object.

[0009] Furthermore, the preprocessing of the multi-source heterogeneous time-series dataset includes: S110: Preprocess the multi-source heterogeneous time-series data set, including: denoising, geometric correction, and radiometric correction; S120: Perform spatial resolution matching and integration on the preprocessed data to obtain integrated data; S130: Perform radiometric correction and denoising on the integrated data to construct standardized spatio-temporal data under a unified spatio-temporal coordinate system.

[0010] Furthermore, extracting spatio-temporal correlation features based on the dynamic graph convolutional network includes: S210: Construct a spatio-temporal graph structure, where nodes represent spatial positions, edges represent spatio-temporal correlations, and the adjacency matrix is updated through a dynamic attention mechanism; S220: Input the spatio-temporal feature vector into the spatio-temporal feature extraction model for inference to obtain the first classification result output by the model; S230: Use the first classification result to obtain change pattern data by comparing multiple time periods; S240: Use the change pattern data to construct spatio-temporal pattern data for dynamic pattern analysis.

[0011] Furthermore, mapping the dynamic analysis report to spatio-temporal data visualization and trend maps includes: S410: Format the analysis results and store them as a visualization data file; the visualization data file includes: the spatial position of the changed area, the classification information of the change type, and the time series of the change; S420: Use the geographic information system platform to generate a visualization display interface according to the stored visualization data file. The visualization display interface includes: spatio-temporal data visualization of the change and a trend map.

[0012] Furthermore, generating the long-term evolution prediction result of the target object includes: S510: Perform statistical calculations based on satellite images and data from the multi-source heterogeneous data acquisition terminal to obtain land data characteristics. The land data characteristics include: the trend of land use type changing over time and derivative characteristics; S520: Perform feature fusion on the land data characteristics to obtain comprehensive feature data; S530: According to the comprehensive feature data, use the land use type prediction model to predict the change of land use type at future moments to obtain a prediction result; S540: Generate a comprehensive report for dynamic pattern analysis based on the prediction result and social and economic environment data.

[0013] A dynamic remote sensing monitoring system based on land use, applied to the method described in any one of the above, includes: A data processing module that acquires data from multi-source heterogeneous data collection terminals, preprocesses and integrates the data from multi-source heterogeneous data collection terminals to obtain image data; A land change recognition module that extracts spatio-temporal features from the image data and uses a dynamic pattern analysis model for dynamic change detection to obtain spatio-temporal pattern data of dynamic pattern analysis; A type change analysis module that analyzes the change situations in different time periods using the spatio-temporal pattern data to obtain the analysis results of type changes; A result display module that displays the classification results through a visualization interface; A report generation module that comprehensively analyzes the ground data and multi-source heterogeneous data to generate a prediction report of dynamic pattern analysis.

[0014] The beneficial effects are as follows: (1) Efficient alignment of multi-source data and noise suppression. Through cross-modal tensor decomposition and adaptive filtering algorithms, the problems of spatio-temporal drift and noise interference in the integration of multi-source heterogeneous data (such as sensor data and spatial coordinate data) by traditional methods are solved, the data alignment error is reduced, and the input quality of subsequent analysis is significantly improved.

[0015] (2) Improvement in the accuracy of dynamic spatio-temporal correlation modeling. Based on the dynamic graph convolutional network and spatio-temporal attention mechanism, the locality limitation of traditional sliding window statistics is broken through, the modeling of long-range spatio-temporal dependence relationships is realized, and the recognition accuracy of complex dynamic patterns (such as the fault propagation path of industrial equipment) is improved.

[0016] (3) Optimization of the state change detection efficiency. By performing probability distribution modeling and state transition probability calculation on the dynamic pattern data, compared with the traditional threshold method, the false alarm rate of the changed area is reduced, and the detection response time is shortened to the millisecond level.

[0017] (4) Enhancement of real-time interaction and long-term prediction capabilities. By combining incremental learning optimization and an interactive visualization engine, it supports the real-time rendering of dynamic analysis reports and the online update of models, reduces the monthly error rate of long-term evolution prediction, and allows users to customize filtering rules to improve decision-making flexibility.

[0018] Of course, it is not necessary for any product implementing this application to achieve all the above-mentioned advantages simultaneously. Description of the Drawings

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 This is a flowchart of a dynamic remote sensing monitoring method based on land use provided by an embodiment of the present application; Figure 2 This is a structural block diagram of a dynamic remote sensing monitoring system based on land use provided by an embodiment of the present application. Detailed implementation manners

[0021] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application belong to the scope protected by the present application.

[0022] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms of "a", "the" and "said" used in the embodiments of the present invention and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise.

[0023] It should be understood that the term "and / or" used herein is only an association relationship describing associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after.

[0024] Depending on the context, the word "if" as used herein can be interpreted as "when" or "while" or "in response to a determination" or "in response to a detection". Similarly, depending on the context, the phrase "if a determination" or "if a detection (stated condition or event)" can be interpreted as "when a determination is made" or "in response to a determination" or "when a detection (stated condition or event) is made" or "in response to a detection (stated condition or event)".

[0025] Currently, there are some technologies that have not been able to provide a comprehensive, real-time, and accurate land use dynamic monitoring system, making it difficult to provide effective decision-making support in land resource management and policy formulation.

[0026] In view of this, the present application provides a new idea. Figure 1 This is a flowchart of a dynamic remote sensing monitoring method based on land use provided by an embodiment of the present application. As Figure 1 shown in, the method may include the following steps: S100: Obtain a multi-source heterogeneous time-series data set, preprocess the data set, and generate standardized spatio-temporal data; S200: Extract spatiotemporal correlation features from the normalized spatiotemporal data based on a dynamic graph convolutional network, and perform multi-dimensional feature fusion using a dynamic pattern recognition model to generate dynamic pattern data of the target object; S300: Perform classification decision on the dynamic pattern data to generate multi-period state analysis results of the target object, including: S400: Construct an interactive visualization engine to map the dynamic analysis report into spatiotemporal data visualization and trend graphs; S500: Optimize the dynamic pattern recognition model based on the incremental learning of historical data and real-time data to generate long-term evolution prediction results of the target object.

[0027] As can be seen from the above process, this application significantly improves data quality and consistency by obtaining satellite images and multi-source heterogeneous data acquisition terminal data and preprocessing and integrating them, laying a foundation for subsequent analysis. Using spatiotemporal feature extraction and deep learning models for land use dynamic change detection can accurately capture complex dynamic pattern analysis trends, overcome the limitations of traditional methods in dynamic monitoring, and achieve high-precision dynamic monitoring. Through refined classification and analysis of dynamic patterns at different time periods, the generated results can not only accurately reveal the change rules but also provide data support for resource management and policy making. Further combined with visualization technology, the dynamic pattern analysis is intuitively displayed through the visualization of changing spatiotemporal data and trend graphs, enabling users to quickly understand the changing areas and their trends and facilitating scientific decision making.

[0028] The following describes in detail each step in the above process and the further effects that can be generated in combination with embodiments. It should be noted that the "first", "second", etc. limitations involved in this disclosure do not have limitations in terms of size, order, quantity, etc., and are only used to distinguish in name.

[0029] First, in combination with an embodiment, the above step S100, that is, "obtain satellite images and multi-source heterogeneous data acquisition terminal data, preprocess and integrate the satellite images and the multi-source heterogeneous data acquisition terminal data to obtain remote sensing image data", is described in detail.

[0030] Satellite image data usually comes from multiple remote sensing satellite platforms, such as the Landsat series, the Sentinel series, or high-resolution commercial satellites (such as WorldView, QuickBird, etc.). For example, public remote sensing data platforms can be accessed through international or regional remote sensing data centers to select the target area, time range, and band type to download image data; or for specific areas and monitoring needs, an image acquisition request can be issued through the satellite mission scheduling system to obtain real-time images with high spatio-temporal resolution; high-resolution or specific-band image data can also be purchased through commercial remote sensing data providers to meet more refined monitoring needs.

[0031] Images of multi-source heterogeneous data acquisition terminals can collect environmental data (such as temperature, humidity, soil humidity, etc.) for a long time through devices such as weather stations and soil monitoring stations. These data are usually automatically aggregated to the data center through the Internet of Things system; multiple sensor nodes (such as spectral sensors, temperature and humidity sensors) can also be deployed in the target area to monitor local environmental information in real time and transmit data through wireless communication technology; or drones or vehicles can be used to carry sensors to collect high-precision ground data in specific areas to make up for the lack of spatial coverage of fixed sensor networks.

[0032] Satellite image data has the characteristics of wide coverage and rich spectral information, and can provide regional land use information, but may be affected by factors such as atmospheric conditions and sensor performance, resulting in noise or geometric distortion. The data of multi-source heterogeneous data acquisition terminals contains local environmental factors, with limited spatial coverage, but can provide supplementary information for satellite images. By preprocessing the two types of data, including denoising, geometric correction, and radiometric correction, interference factors are eliminated, and the authenticity and consistency of the data are improved. Subsequently, through spatial resolution matching and integration operations, the data of multi-source heterogeneous data acquisition terminals is combined with satellite image data into unified remote sensing image data, thus providing accurate and comprehensive basic data for dynamic monitoring and refined analysis of dynamic pattern analysis.

[0033] As an implementable method, step S100 can be realized by steps S110 - S130: S110: Preprocess the multi-source heterogeneous time series dataset, including: denoising, geometric correction, and radiometric correction.

[0034] Satellite image data includes images at different time points and different spatial resolutions, and can provide the spectral characteristics of the target area (such as bands of red, green, blue, near-infrared, etc.). The spatial resolution of these satellite image data is usually 10 meters to 30 meters, and the time frequency is monthly or quarterly. The preprocessing of satellite image data includes: denoising, geometric correction, and radiometric correction.

[0035] Among them, denoising can be achieved through spatial filtering algorithms (such as Gaussian filters or median filters). The Gaussian filter uses the following formula: (1) Where is the pixel value of the original satellite image at the coordinate position, is the pixel value of the denoised satellite image, is the Gaussian kernel function, defined as: (2) This formula reduces the impact of noise by weighted averaging the values of surrounding pixels. is the filter window size, is the standard deviation, which controls the intensity of denoising.

[0036] For geometric correction, in order to correct remote sensing image data to a geographic coordinate system, ground control points (GCPs) are used to calibrate the geometric distortion of the satellite image. Assume that the position of any pixel point in the satellite image is , and its corresponding geographic coordinate is . By solving the spatial transformation model using the least squares method, the image coordinate system is matched with the geographic coordinate system. The transformation matrix of geometric correction can be expressed by the following formula: (3) Where is the transformation matrix calculated from the ground control points.

[0037] For radiometric correction, an atmospheric correction model (such as the 6S model) or an inversion method based on ground measured data can be used to perform radiometric correction on the image. Assume that the original radiance value of the image is , and it is converted into surface reflectance (4) Where is the impact of atmospheric scattering, is the atmospheric transmittance, and its value depends on atmospheric conditions, sensor altitude, etc.

[0038] Through the above steps, high-quality satellite image data after denoising, geometric correction, and radiometric correction is obtained.

[0039] S120: Perform spatial resolution matching and integration on the preprocessed data to obtain integrated data.

[0040] Obtain ground monitoring data within a certain period through a multi-source heterogeneous data acquisition terminal network or acquisition platform. This data usually includes environmental factors related to dynamic pattern analysis, such as soil moisture, air temperature, vegetation coverage, etc.

[0041] Fuse the data from the multi-source heterogeneous data acquisition terminal with satellite image data to obtain remote sensing image data. To ensure spatial consistency, it is necessary to match the spatial resolution of the multi-source heterogeneous data acquisition terminal data with that of the satellite image. This can be achieved through spatial interpolation or by using ground sampling points with the same resolution as the image. Common interpolation methods include Kriging interpolation or inverse distance weighted interpolation (IDW).

[0042] (5) Among them, For integrating data, is the data integration function, and are the satellite image data and the data from the multi-source heterogeneous data acquisition terminal respectively.

[0043] During the data integration process, there may be some inconsistencies or missing values, and the missing data needs to be filled through data imputation or other methods (such as multiple imputation methods).

[0044] Through this step, the integrated data will contain data from satellite images and multi-source heterogeneous data acquisition terminals, and be prepared for subsequent radiometric correction and denoising processing.

[0045] S130: Perform radiometric correction and denoising on the integrated data to construct standardized spatio-temporal data under a unified spatio-temporal coordinate system.

[0046] Perform radiometric correction on the integrated data The radiometric correction adjusts the radiometric values of the image data by considering factors such as atmospheric effects, sensor characteristics, and imaging angles to obtain standardized data which can represent more real ground information.

[0047] The general process of radiometric correction can be expressed as: (6) Among them, is the data after radiometric correction, is the radiometric correction function, is the integrated data, is the atmospheric parameter, is the sensor characteristic.

[0048] After radiation correction, noise may still exist in the data, especially the noise introduced by factors such as sensor hardware limitations and external environmental changes. To improve the data quality, the integrated data needs to be denoised. Denoising operations can employ some common image processing techniques, such as median filtering, mean filtering, bilateral filtering, etc. The aim is to remove the high-frequency noise in the data and retain the meaningful signal information.

[0049] The denoising process can use filtering methods, and the mathematical expression is: (7) where is the data after denoising, is the denoising filter function.

[0050] After denoising and radiation correction, it is necessary to evaluate the quality of the data to ensure that it meets the requirements of subsequent analysis. The data quality assessment can be carried out by comparing with existing ground measured data, or by error metrics (such as root mean square error RMSE, correlation coefficient, etc.) to quantify the accuracy and reliability of the data.

[0051] The data quality assessment formula is: (8) where is the data quality assessment value, is the number of data points, and are the values of the denoised data and the ground measured data at the th position respectively.

[0052] Through the above steps, the obtained is high-quality remote sensing data after radiation correction and denoising, which is suitable for subsequent spatio-temporal pattern analysis and dynamic pattern analysis detection.

[0053] The following describes in detail the above step S200, that is, "extract spatio-temporal features from the remote sensing image data, use the dynamic pattern analysis model to detect land use dynamic changes, and obtain the spatio-temporal pattern data of dynamic pattern analysis" in combination with embodiments.

[0054] This step describes a dynamic change detection method based on remote sensing image data. By extracting spatio-temporal features and combining with a dynamic pattern analysis model, it can accurately capture the change patterns of land use. Specifically, spatio-temporal feature extraction analyzes the temporal change trends and spatial distribution features from the multi-temporal data of remote sensing images. First, using time series analysis methods (such as short-time Fourier transform or sliding window method), temporal features are extracted from images at different times to identify the periodicity, suddenness, or long-term trends of dynamic pattern analysis. Second, through spatial analysis techniques (such as gray-level co-occurrence matrix or texture feature extraction), the spatial distribution information of different plots in the image is captured, including the boundary changes and regional differences of land cover types. The extracted spatio-temporal features are input into the dynamic pattern analysis model for dynamic change detection. This model usually uses deep learning methods (such as convolutional neural network CNN or spatio-temporal convolutional network ST-CNN). By learning the spatio-temporal associations of a large amount of remote sensing image data, it predicts and classifies the types of land use changes. For example, when detecting the process of agricultural land being gradually occupied by urban construction in a certain urban area, the model can identify the conversion of land use types at specific locations according to the change intensity at different times and generate visualizations and trend charts of the change spatio-temporal data.

[0055] As an implementable way, step S200 can be realized by steps S210 - S230: S210: Construct a spatio-temporal graph structure, where nodes represent spatial positions, edges represent spatio-temporal correlations, and the adjacency matrix is updated through a dynamic attention mechanism.

[0056] By extracting spatio-temporal features from remote sensing image data, the time series change patterns in the image are identified. First, select remote sensing image data at multiple time points (such as quarterly or annually) as the input data set, and the input data set includes , where is the time at the coordinate of the corrected remote sensing image data.

[0057] Further, use the sliding window method to extract temporal features from continuous remote sensing image data. Assume the size of the sliding window is , where the starting time point of the window is , and the ending time point is , then the spatio-temporal change sequence can be extracted from the multi-temporal remote sensing images, and this sequence can reflect the change trends of the image in space and time.

[0058] For the time series data , adopt time feature analysis methods, such as short-time Fourier transform or wavelet transform, to extract the time domain and frequency domain features of the image. The formula is as follows: (9) Among them, is the spatio-temporal spectrum extracted by short-time Fourier transform, is the frequency, is the imaginary unit, is the length of the time series. This spatio-temporal spectrum can effectively capture the change characteristics of remote sensing images in the time domain and frequency domain, providing input data for subsequent deep learning modeling.

[0059] While extracting spatio-temporal features, spatial features also need to be extracted from remote sensing images. Specifically, traditional remote sensing image feature extraction methods (such as gray-level co-occurrence matrix, texture features, etc.) are used to extract the spatial information of the images. For example, calculate the gray-level co-occurrence matrix of the image , and its calculation formula is as follows: (10) Among them, is the gray-level co-occurrence matrix is the pixel gray value, is the pixel spacing, is the angle, is the Dirac function. The extracted spatial features can reflect the spatial distribution laws such as land use types and vegetation changes.

[0060] Combining the time-domain features and spatial features, a spatio-temporal feature vector is obtained through a feature fusion method (such as weighted average, principal component analysis, etc.). Assuming the spatio-temporal feature vector is , then the time-domain features and spatial features are fused in the following way: (11) Among them, and are weight coefficients, controlling the fusion ratio of the time-domain features and spatial features. The finally obtained spatio-temporal feature vector can be used as the input data of the deep learning model.

[0061] S220: Input the spatio-temporal feature vector into the spatio-temporal feature extraction model for inference, and obtain the first classification result output by the model.

[0062] Select a convolutional neural network (CNN) as the spatio-temporal feature extraction model, and extract spatio-temporal features by constructing convolutional layers, pooling layers and fully connected layers. Assuming the spatio-temporal feature vector is , the input of the model is the stacked data of multiple spatio-temporal features. The convolutional layer of the model extracts spatio-temporal features through the following convolutional operation: (12) Among them, is the convolution kernel, is the output of the convolution operation, representing the local response to features in space and time.

[0063] After the convolution operation, the model will reduce the dimensionality through a pooling layer to reduce computational complexity. Assume the output of the pooling layer is , then the final classification or regression is performed through a fully connected layer to obtain the prediction result of the dynamic change of land use. Specifically, use function to predict the category: (13) where, is the number of land use categories, is the predicted probability of the land use category at a given location and time point.

[0064] Based on the output of the spatio-temporal feature extraction model, extract the spatio-temporal patterns of dynamic pattern analysis. Assume the first classification result of the model output is .

[0065] S230: Use the first classification result to obtain the change pattern data by comparing multiple time periods.

[0066] By comparing multiple time periods of the first classification result, identify the change pattern of land use and obtain the change pattern data of land use . The spatio-temporal change pattern data can be extracted in the following way: (14) where, represents the change amount of the land use category at time . If is positive, it means that the land use has changed; if it is zero, it means that the land use has not changed.

[0067] S240: Use the change pattern data to construct the spatio-temporal pattern data of the dynamic pattern analysis. The spatio-temporal pattern data will include: the dynamic pattern analysis situation at multiple time points, and the change trend of the spatial position.

[0068] Based on the above spatio-temporal change pattern data , construct a spatio-temporal pattern dataset of the dynamic change of land use. These spatio-temporal pattern data will include the dynamic pattern analysis situation at multiple time points, and the change trend of the spatial position. These data will provide a basis for the long-term monitoring and trend prediction of land use.

[0069] At the same time, save the extracted spatio-temporal pattern data in a formatted data storage to ensure data visualization and subsequent analysis.

[0070] The following describes in detail step S300 above, that is, "analyze the changes in land use in different time periods using spatio-temporal pattern data to obtain the analysis results of land use type changes", in combination with the embodiments.

[0071] This step generates detailed analysis results of regional land use type changes by analyzing the land use classification information in the spatio-temporal pattern data and combining with the dynamic changes in land use in different time periods. Compare the land use classification results in the spatio-temporal pattern data according to time periods to identify the changes in land use types at the same spatial location in different time periods. Specifically, the classification results in the spatio-temporal pattern data can be compared and analyzed to determine the areas where land use types have changed, and mark the change types (such as from forest land to agricultural land). Statistical analysis is performed on the changed areas to evaluate the change intensity (such as the area and proportion of the reduction of agricultural land in a certain place) of different land use types and their time trends (such as decreasing year by year or changing rapidly).

[0072] As an implementable way, step S300 can be implemented by steps S310 - S330: S310: Model the probability distribution of the dynamic pattern data through a classification model to obtain the classification results of each spatial location in consecutive time periods.

[0073] Classify the spatio-temporal pattern data to obtain the land use type data including the second classification results of each spatial location in different time periods; obtain the land use type data for each time period according to the second classification results.

[0074] Specifically, use the spatio-temporal pattern data of land use dynamic changes , combined with spatio-temporal features and historical land use data, to construct a classification model.

[0075] First, obtain historical land use classification data , and use this data to train the classification model. Preferably, the classification model adopted is a support vector machine , then the training objective of the model is to minimize the following objective function: (15) Among them, is the weight of the classifier, is the bias term, is the penalty coefficient, is the slack variable, is the number of samples. This classification model aims to obtain a decision boundary that can classify new spatio-temporal pattern data through learning historical land use data.

[0076] Put the spatio-temporal pattern data Input it into the trained SVM model for classification. Based on the learning of historical data, the model will output the probability distribution of different land use types , which is the second classification result. Specifically, for each location and time , the model will calculate the predicted probability of the land use type at this location: (16) where is the inner product after feature mapping, representing the weight of spatio-temporal features in the classification decision function, is the predicted probability of the land use category, is the number of categories of the classification model.

[0077] Generate land use type data for each time period according to the second classification result output by the classification model , which contains the land use classification results of each spatial location in different time periods. This data will provide a basis for dynamic pattern analysis. Specifically, according to the time tags in the second classification result, the classification results within each time period can be extracted separately. Divide the land use types output by the classification model according to the time series to generate the land use type distribution corresponding to each time point. For example, for the classification data of a certain area, the classification results in 2020 show that area A is "agricultural land" and area B is "construction land"; the classification results in 2025 show that area A has become "construction land" while area B remains unchanged.

[0078] S320: Conduct state transition analysis on the classification results of consecutive time periods to identify the state change areas of the target object.

[0079] After obtaining the land use type data, further analyze the changes in land use types. Specifically, by comparing the land use types at the same location in different time periods , identify the change areas. For example, calculate the dynamic pattern analysis between time and : (17) If is not zero, it indicates that the land use at this location has changed, and the type of change is indicated by . Through this calculation, the spatial distribution and change trend of dynamic pattern analysis can be identified.

[0080] Furthermore, conduct classification statistics on the dynamic pattern analysis. Assume that different types of changes include the conversion of agricultural land to urban land, forest cover changes, etc. Count the number of various changes and map them to spatial locations. The statistical formula is as follows: (18) Among them, is an indicator function. If belongs to a specific change type , otherwise it is zero. This statistical result provides data support for the specific type analysis of dynamic pattern analysis.

[0081] S330: Statistically analyze the spatio-temporal evolution law of the state change region, and generate a dynamic analysis report including change intensity, relevance, and trend prediction.

[0082] Based on the above change statistics, further analyze the trends and patterns of dynamic pattern analysis. Regression analysis or time series analysis methods can be used to identify the laws of dynamic pattern analysis. For example, by fitting a linear regression model, analyze the trend of dynamic pattern analysis in a certain area: (19) Among them, is the predicted land use type, and are the intercept and slope of the regression model respectively, is the time variable. Through the regression coefficient , the growth or decline trend of dynamic pattern analysis can be judged, thus providing a basis for the prediction of future land use.

[0083] The following describes in detail step S400 above, that is, "display the land use classification result through a visual interface", in combination with an embodiment.

[0084] This step presents the complex land use classification results and change analysis data to the user in an intuitive and interactive form, improving the comprehensibility and practicality of the data. This technology combines the classification results with a Geographic Information System (GIS) by designing a user-friendly interface, and displays the spatial distribution and temporal change trend of land use in a graphical way. Specifically, the visual interface can include the following functions: Classification result display: Display the land use classification results of each time period on the geospatial map in the form of color coding or a legend; Change trend analysis: Display the classification results of different time periods through a time axis control, allowing the user to dynamically view the dynamic pattern analysis process in a certain area; Change hot spot area marking: Use change spatio-temporal data visualization technology to highlight the hot spot areas where dynamic pattern analysis is frequent, such as urban expansion areas or ecologically sensitive areas; Interactive data query: The user can click on a certain area on the map to view the detailed classification information and change records of that area, etc.

[0085] As an implementable way, step S400 can be implemented by steps S410 - S420: S410: Format the analysis results and store them as a visualization data file; the visualization data file includes: the spatial location of the changed area, the classification information of the change type, and the time series of the change.

[0086] Format the analysis results of land use type changes and the results of trend analysis, and store them as a visualization data file. The formatted data includes the spatial location of the changed area, the classification information of the change type, and the time series of the change. Specifically, and the trend of dynamic pattern analysis are stored in a standard data format for subsequent visualization display and spatial analysis.

[0087] S420: Use a geographic information system platform to generate a visualization display interface according to the stored visualization data file; the visualization display interface includes: visualization of spatio-temporal change data and trend charts.

[0088] Use a GIS (Geographic Information System) platform to visually display the stored data. For example, display the spatial distribution of dynamic pattern analysis by color coding, and draw a trend chart to intuitively present the situation of dynamic pattern analysis. Specifically, use spatio-temporal data visualization or hierarchical display of the changed area and the spatial location of the change type to generate an intuitive analysis result image.

[0089] Based on the above analysis results, generate a comprehensive analysis report on dynamic pattern analysis. The report content includes the spatial distribution, change type, change trend, etc. of dynamic pattern analysis, aiming to provide decision-making support for land management and planning. The report will include statistical charts, change analysis results, and trend predictions, etc., to help policymakers understand the dynamic situation of dynamic pattern analysis.

[0090] Based on the analysis results of land use type changes and trend analysis results , design a user-friendly visualization interface. This interface will display the land use classification information for different time periods, as well as the spatial distribution of dynamic pattern analysis. Understandably, use map overlay technology to code and display land use categories with different colors or symbols, enabling users to intuitively understand the land use types and their changes in each plot. This visualization interface supports multi-level user needs. For example, policymakers can view the global change trend, while local managers can view the detailed information of local changes.

[0091] To ensure the efficient display of the visualization interface, the spatio-temporal data and dynamic pattern analysis data Perform formatting and convert it into a format suitable for visualization tools to read. Specifically, the data is converted into a format recognizable by a Geographic Information System (GIS) for display on a visualization platform.

[0092] (20) Among them, is the data in visualization format, is a data conversion function that converts the classification result into a suitable display format.

[0093] Design user interaction functions at different levels in the visualization interface. Users at different levels can select different geographical regions and time periods for query and analysis according to their needs. For example, local managers can select the changes in a specific city or village, while national policymakers can select a larger range for dynamic pattern analysis to analyze trends.

[0094] For the visualization of spatio-temporal change data, data can be analyzed based on dynamic patterns , and spatio-temporal change data visualization is generated to show the intensity of dynamic pattern analysis in different regions. Spatio-temporal change data visualization is displayed by aggregating the spatial distribution of dynamic pattern analysis and applying a color scale mapping. First, based on the dynamic pattern analysis within the time , calculate the change intensity at each location: (21) Among them, represents the change intensity of location within the time interval, is an indicator function. When is not zero, it means that dynamic pattern analysis has occurred at this location during time , and the value is 1; otherwise, it is 0. By aggregating these change intensities, spatio-temporal change data visualization is obtained.

[0095] Input the calculated change intensity data into the visualization platform and display the spatio-temporal data visualization through color scale mapping. Specifically, map the change intensity to a certain color range. Regions with larger intensities are represented by red, and regions with smaller intensities are represented by green or blue. Spatio-temporal data visualization can clearly show the hotspots of dynamic pattern analysis, helping users quickly identify the regions with the most significant changes. Users can dynamically adjust the time range and space range of spatio-temporal data visualization through the interaction functions of the visualization interface. For example, users can select different time periods to view the spatio-temporal data visualization of different stages of dynamic pattern analysis, thereby discovering the laws and trends of dynamic pattern analysis.

[0096] For the trend chart, the trend analysis results can be visualized as a trend chart, and users can select different regions for trend analysis. The trend chart shows the overall trend of the dynamic pattern analysis. For example, whether agricultural land in a certain area is gradually transformed into urban land. The trend chart usually uses line charts or bar charts to show the dynamic pattern analysis of each region at different time periods, helping users understand the land use development trend of a certain area.

[0097] Furthermore, users can use the interactive function to select different land use types and time periods to dynamically view the trend of the dynamic pattern analysis in different regions. For example, select the change trend of agricultural land in a specific city, or select the urban expansion trend in a specific time period for more detailed analysis.

[0098] The following describes in detail the above-mentioned step S500, that is, "combining ground data and remote sensing data for comprehensive analysis to generate a prediction report of dynamic pattern analysis" in conjunction with embodiments.

[0099] Based on the prediction results, combined with social, economic, and environmental data (such as population density, ecological sensitivity area division), analyze the spatial distribution, type conversion, and possible impacts of dynamic pattern analysis under different change scenarios. For example, analyze the impact of policy adjustments on the reduction of agricultural land or the expansion of urban land, as well as the environmental costs of different development strategies.

[0100] The prediction report is presented in the form of visualization and statistical analysis, including the time series trend of dynamic pattern analysis, spatial distribution map, visualization of spatio-temporal data of change intensity, and evaluation suggestions for future scenarios. The report will also highlight key regions (such as urban expansion hotspots) and high-risk regions (such as potential damage to ecological protection areas).

[0101] As an implementable method, step S500 can be implemented by steps S510 - S520: S510: Perform statistical calculations based on satellite images and multi-source heterogeneous data acquisition terminal data to obtain land data characteristics, and the land data characteristics include: the trend and derivative characteristics of land use types changing over time.

[0102] Assume the spatial accuracy is , and adjust the original remote sensing data by spatial resampling. Specifically, the remote sensing image data after resampling can be expressed as: (22) Among them, F(t) is the original remote sensing image data, represents the remote sensing image data after resampling.

[0103] Next, extract features from the original remote sensing data. In terms of spatial feature extraction, the spatial distribution and its changes of land use types are mainly concerned. Specifically, the changes in land use types can be measured by calculating the differences between the images at the current moment and the previous moment, and the spatial features are expressed as: (23) where represents the remote sensing data at time , is the total number of pixels in the image. In this way, the spatial aggregation and change trends of land use types can be quantified.

[0104] In terms of temporal feature extraction, the trend of land use type changes over time is concerned. By weighting the time series data of land use types, the corresponding temporal features are obtained: (24) where represents the land use type in area at time , is the weight coefficient of this area, is the number of land use type categories.

[0105] In addition, other derivative features need to be extracted, such as ecological benefits, resource utilization efficiency, etc. Assuming these features are , and its calculation formula is: (25) where represents the area of each pixel point, is the total area. Through these features, the spatial distribution and temporal changes of land use can be understood more comprehensively.

[0106] S520: Perform feature fusion on the land data features to obtain comprehensive feature data.

[0107] Fuse these land data features. The specific method is to use techniques such as weighted average or principal component analysis (PCA) to fuse different types of features. The fused comprehensive feature data is expressed as: (26) where is the th type of feature, is the corresponding weight coefficient, is the total number of features. Through this weighted fusion, the importance of each feature in the comprehensive analysis is fully reflected.

[0108] S530: Based on the comprehensive feature data, use the land use type prediction model to predict the change of land use type at a future time, and obtain the prediction result.

[0109] Based on the comprehensive feature data, use the prediction model to predict the future dynamic pattern analysis. First, input the comprehensive feature data and the historical dynamic pattern analysis data , where the latter includes information such as the dynamic pattern analysis rate and change type. Based on these data, establish a prediction model to predict the change of land use type at a future time.

[0110] In the selection of the prediction model, the support vector machine regression (SVR) model can be used for prediction. The prediction process is expressed as: (27) where represents the predicted future land use type, is the support vector regression model. The input of the model is the historical data of land use and the extracted feature data, and after training, it can predict the change of future land use.

[0111] The parameter optimization of the model is carried out by methods such as cross-validation to find the best model parameters. The optimization process is as follows: (28) where is the actual dynamic pattern analysis data, is the predicted value, is the hyperparameter of the model.

[0112] Finally, use the trained prediction model to predict the future land use type, and the formula is: (29) where is the predicted land use type.

[0113] S540: Generate a comprehensive report on dynamic pattern analysis based on the prediction result and social and economic environment data.

[0114] Based on the land use prediction result and social and economic environment data, conduct multi-dimensional comprehensive analysis, and finally provide policy support and decision-making basis for dynamic pattern analysis for decision-makers.

[0115] This step can be achieved in various ways. For example, a dynamic feedback system for dynamic pattern analysis and the socio-economic environment can be constructed, and system dynamics models (such as Vensim or AnyLogic) can be used to simulate the interactions between land use and the economy and environment, and analyze the long-term impacts under different policy scenarios. For example, by adjusting parameters (such as the economic growth rate and the intensity of policy intervention), the effects of policies on dynamic pattern analysis can be evaluated.

[0116] As a preferred implementation method, the prediction results and the socio-economic environment data can be input into a comprehensive decision-making model, and a comprehensive report on dynamic pattern analysis can be generated according to the output results of the comprehensive decision-making model; among them, the comprehensive decision-making model is pre-trained, and the model parameters are optimized according to the optimization objective during the model training process; the optimization objective is constructed based on the impact of dynamic pattern analysis on the socio-economy and the risks brought by land use types.

[0117] Specifically, first, input the land use prediction results , and relevant data from the socio-economy and environment, such as population changes , policy adjustments and ecological environment indices .

[0118] During the construction of the comprehensive decision-making model, the multiple impacts of dynamic pattern analysis on factors such as the socio-economy and environmental protection are considered. Assuming a multi-objective optimization model is used, comprehensively considering dynamic pattern analysis predictions, socio-economic data, and environmental data, the model formula is as follows: (30) Among them, is the impact of dynamic pattern analysis on the socio-economy, is the risk assessment brought by land use types, is the risk weight, is the weight of each factor.

[0119] Finally, based on the comprehensive analysis results, a comprehensive report on dynamic pattern analysis is generated. The report content includes: predictions of future dynamic pattern analysis trends, the impact of policy adjustments on dynamic pattern analysis, suggestions for ecological protection and resource management, etc. The report can be presented through a visualization interface, such as the visualization of dynamic pattern analysis spatio-temporal data, trend charts, etc., providing intuitive decision-making basis for policymakers.

[0120] The above description is of specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0121] Embodiment 2: Figure 2 The structural block diagram of a dynamic remote sensing monitoring system based on land use according to an embodiment of the present invention is shown. As Figure 2 shown, the system may include: A data processing module 201, which acquires data from a multi-source heterogeneous data acquisition terminal, preprocesses and integrates the data from the multi-source heterogeneous data acquisition terminal to obtain image data; A land change recognition module 202, which extracts spatio-temporal features from the image data, and uses dynamic pattern analysis and an analysis model to perform dynamic change detection to obtain spatio-temporal pattern data of the dynamic pattern analysis.

[0122] A type change analysis module 203, which analyzes the change situations in different time periods using the spatio-temporal pattern data to obtain an analysis result of the type change.

[0123] A result display module 204, which displays the classification result through a visualization interface.

[0124] A report generation module 205, which comprehensively analyzes by combining ground data and multi-source heterogeneous data to generate a prediction report of the dynamic pattern analysis.

[0125] As an implementable manner, the data processing module 201 may be configured to: preprocess the data, and the preprocessing includes: denoising, geometric correction, and radiometric correction; perform spatial resolution matching and integration on the preprocessed data and the data from the multi-source heterogeneous data acquisition terminal to obtain integrated data; perform radiometric correction and denoising on the integrated data to obtain remote sensing data.

[0126] As an implementable manner, the land change recognition module 202 may be configured to: identify a time series change pattern from the image data through a sliding window method to obtain a spatio-temporal feature vector; input the spatio-temporal feature vector into a spatio-temporal feature extraction model for inference to obtain a first classification result output by the model; use the first classification result to obtain change pattern data by comparing multiple time periods; use the change pattern data to construct the spatio-temporal pattern data of the dynamic pattern analysis. The spatio-temporal pattern data will include: the dynamic pattern analysis situations at multiple time points, and the change trend of the spatial position.

[0127] As an implementable manner, the type change analysis module 203 can be configured to: classify the spatio-temporal pattern data to obtain the type data including the second classification results of each spatial position in different time periods; obtain the type data of each time period according to the second classification results; identify the change regions by comparing the type data of each spatial position in different time periods; count the change trends of the change regions, and generate the analysis results of type changes.

[0128] As an implementable manner, the result display module 204 can be configured to: format the analysis results and store them as visual data files; the visual data files include: the spatial positions of the change regions, the classification information of the change types, and the time series of the changes; use the geographic information system platform to generate a visual display interface according to the stored visual data files, and the visual display interface includes: spatio-temporal change data visualization and trend charts.

[0129] As an implementable manner, the report generation module 205 can be configured to: perform statistical calculations according to satellite images and multi-source heterogeneous data acquisition terminal data to obtain land data characteristics, and the land data characteristics include: the trends and derivative characteristics of land use types changing over time; perform feature fusion on the land data characteristics to obtain comprehensive feature data; according to the comprehensive feature data, use the land use type prediction model to predict the changes of land use types at future moments to obtain prediction results; generate a comprehensive report on dynamic pattern analysis according to the prediction results and social and economic environment data.

[0130] As an implementable manner, when the report generation module 205 generates a comprehensive report on dynamic pattern analysis according to the prediction results and social and economic environment data, it can be configured to: input the prediction results and the social and economic environment data into a comprehensive decision-making model, and generate a comprehensive report on dynamic pattern analysis according to the output results of the comprehensive decision-making model; wherein, the comprehensive decision-making model is pre-trained, and the model parameters are optimized according to the optimization objective during the model training process; the optimization objective is constructed according to the impact of dynamic pattern analysis on the social economy and the risks brought by land use types.

[0131] The above has introduced the technical solutions provided by this application in detail. Specific examples are used in this article to elaborate on the principles and implementation manners of this application. The descriptions of the above embodiments are only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A dynamic remote sensing monitoring method based on land use, characterized in that The method includes: S100: Obtain a multi-source heterogeneous time-series data set, preprocess the data set, and generate standardized spatio-temporal data; S200: Extract spatio-temporal correlation features from the standardized spatio-temporal data based on a dynamic graph convolutional network, perform multi-dimensional feature fusion using a dynamic pattern recognition model, and generate dynamic pattern data of the target object; S300: Perform classification decision on the dynamic pattern data to generate multi-period state analysis results of the target object, including: S310: Perform probability distribution modeling on the dynamic pattern data through a classification model to obtain classification results at each spatial position in consecutive time periods; The probability distribution of the classification result is defined by the following formula: ; Among them, is spatio-temporal pattern data, is the inner product after feature mapping, representing the weight of spatio-temporal features in the classification decision function, is the predicted probability of land use categories, is the number of categories of the classification model, is the bias term; S320: Perform state transition analysis on the classification results of consecutive time periods to identify the state change regions of the target object, and its change rate is calculated as: ; Among them, represents time, is the land use type at the same location in different time periods, represents the dynamic pattern analysis rate of this location; And classify and count the dynamic mode analysis: ; Among them, represents the state where different change types are mapped to spatial positions, is an indicator function; S330: Perform classification decision on the dynamic pattern data to generate multi-period state analysis results and a dynamic analysis report of the target object; S400: Construct an interactive visualization engine to map the dynamic analysis report into spatio-temporal data visualization and trend graphs; S500: Optimize the dynamic pattern recognition model based on incremental learning of historical data and real-time data to generate long-term evolution prediction results of the target object.

2. The dynamic remote sensing monitoring method based on land use according to claim 1, characterized in that Preprocessing the multi-source heterogeneous time-series data set includes: S110: Preprocess the multi-source heterogeneous time-series data set, including: denoising, geometric correction, and radiometric correction; S120: Perform spatial resolution matching and integration on the preprocessed data to obtain integrated data; S130: Perform radiometric correction and denoising on the integrated data to construct standardized spatio-temporal data under a unified spatio-temporal coordinate system.

3. The method according to claim 1, characterized in that, Extracting spatio-temporal correlation features based on a dynamic graph convolutional network includes: S210: Identify time-series change patterns from remote sensing image data through a sliding window method to obtain spatio-temporal feature vectors; S220: Input the spatio-temporal feature vectors into a spatio-temporal feature extraction model for inference to obtain the first classification result output by the model; S230: Use the first classification result to obtain change pattern data by comparing multiple time periods; S240: Use the change pattern data to construct spatio-temporal pattern data for dynamic pattern analysis.

4. The method according to claim 1, characterized in that, Mapping the dynamic analysis report into spatio-temporal data visualization and trend graphs includes: S410: Format the analysis results and store them as a visualization data file; the visualization data file includes: the spatial positions of the changed areas, the classification information of the change types, and the time series of the changes; S420: Use a geographic information system platform to generate a visualization display interface based on the stored visualization data file, and the visualization display interface includes: spatio-temporal data visualization of the changes and trend graphs.

5. The method according to claim 1, characterized in that, Generating long-term evolution prediction results of the target object includes: S510: Perform statistical calculations based on satellite images and data from multi-source heterogeneous data acquisition terminals to obtain land data characteristics, and the land data characteristics include: the trend of land use type changing over time and derivative characteristics; S520: Perform feature fusion on the land data characteristics to obtain comprehensive feature data; S530: According to the comprehensive feature data, use a land use type prediction model to predict the change of land use type at a future moment to obtain a prediction result; S540: Generate a comprehensive report for dynamic pattern analysis based on the prediction result and socio-economic environment data.

6. A dynamic remote sensing monitoring system based on land use, applied to the method described in any one of claims 1-5, comprising: A data processing module, which acquires data from multi-source heterogeneous data acquisition terminals, preprocesses and integrates the data from multi-source heterogeneous data acquisition terminals to obtain image data; A land change recognition module, which extracts spatio-temporal features from the image data and uses a dynamic pattern analysis model for dynamic change detection to obtain spatio-temporal pattern data of dynamic pattern analysis; A type change analysis module, which analyzes the change situations in different time periods by using the spatio-temporal pattern data to obtain the analysis results of type changes; A result display module, which displays the classification results through a visualization interface; A report generation module, which conducts comprehensive analysis by combining ground data and multi-source heterogeneous data to generate a prediction report of dynamic pattern analysis.

Images