Geographic scene area modeling method based on space-time big data analysis

Abstract

The invention provides a geographic scene area modeling method based on space-time big data analysis. The method comprises the following steps: obtaining space-time characteristics of each space data point in a geographic scene area; screening out key spatial data points based on the correlation values of the spatial-temporal characteristics among different spatial data points; dividing the geographic scene area into a plurality of sub-areas based on the space-time pattern of each key space data point; calculating a region feature value and an attribute feature value of each sub-region, and calculating a region evolution index of each sub-region; and according to the attribute feature value and the regional evolution index of each sub-region, calculating the feature distance between different sub-regions, and clustering the plurality of sub-regions to form different regional clusters. According to the method, through feature correlation analysis and spatio-temporal pattern mining, the regional features and spatio-temporal evolution features thereof are systematically described, and based on regional dynamic evolution analysis, the accuracy of regional feature similarity calculation can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of modeling design, and more specifically, to a geographical scene area modeling method based on spatio-temporal big data analysis. Background Art

[0002] With the development of big data technology, spatio-temporal big data analysis has gradually occupied an important position in geographical information processing and area modeling. In geographical scene area modeling, how to accurately extract and analyze area features and identify evolution patterns in a complex and changing spatio-temporal environment has become a research hotspot and difficulty. The existing technologies mainly focus on the following aspects:

[0003] Insufficient area feature extraction methods: Existing methods mostly use direct statistics and simple weighting methods for feature extraction, and fail to fully consider the correlation and non-linear mapping characteristics between features, resulting in insufficient area feature expression ability.

[0004] Lack of dynamics in area attribute modeling: Most studies lack a comprehensive description of dynamic evolution features when modeling area attributes, and cannot accurately describe the trend of area features changing over time, resulting in low modeling accuracy.

[0005] Clustering analysis fails to fully reflect evolution features: Existing clustering methods mostly calculate based on static feature values, and ignore the impact of area evolution indices on clustering accuracy and effectiveness, making it difficult to achieve dynamic classification of areas with similar features. Summary of the Invention

[0006] In view of the technical problems existing in the prior art, the present invention provides a geographical scene area modeling method based on spatio-temporal big data analysis, which solves the problems of insufficient area feature extraction and lack of dynamics in the prior art.

[0007] The present invention provides a geographical scene area modeling method based on spatio-temporal big data analysis, including:

[0008] Obtain a spatio-temporal big data set within a geographical scene area, where the spatio-temporal big data set includes a plurality of spatial data points;

[0009] Obtain the spatio-temporal features of each of the spatial data points;

[0010] Based on the spatio-temporal features, calculate the correlation value between each of the spatial data points and other spatial data points, and screen out key spatial data points from the spatio-temporal big data set based on the correlation value;

[0011] Obtain the spatio-temporal patterns of each of the key spatial data points, and divide the geographical scene area into multiple sub-areas based on the spatio-temporal patterns of each of the key spatial data points;

[0012] Calculate the regional eigenvalue of each sub-region based on the spatio-temporal pattern contained in each sub-region, and calculate the attribute eigenvalue of each sub-region based on the regional eigenvalue of each sub-region;

[0013] Calculate the regional evolution index of each sub-region based on the attribute eigenvalue of each sub-region;

[0014] Calculate the feature distance between different sub-regions according to the attribute eigenvalue and the regional evolution index of each sub-region, and cluster multiple sub-regions based on the feature distance to form different regional clusters.

[0015] A geographical scene regional modeling method based on spatio-temporal big data analysis provided by the present invention systematically depicts regional features and their spatio-temporal evolution characteristics by performing feature correlation analysis and spatio-temporal pattern mining. This method effectively solves the problems of insufficient regional feature extraction and lack of dynamics in the prior art through non-linear feature mapping and comprehensive feature extraction. At the same time, based on regional dynamic evolution analysis, it can significantly improve the accuracy of regional feature similarity calculation. Description of the Drawings

[0016] Figure 1 It is a flowchart of a geographical scene regional modeling method based on spatio-temporal big data analysis provided by an embodiment of the present invention;

[0017] Figure 2 It is a schematic structural diagram of a geographical scene regional modeling system based on spatio-temporal big data analysis provided by an embodiment of the present invention;

[0018] Figure 3 It is a schematic hardware structure diagram of a possible electronic device provided by the present invention;

[0019] Figure 4 It is a schematic hardware structure diagram of a possible computer-readable storage medium provided by the present invention. Detailed Embodiments

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. In addition, the technical features in each embodiment or individual embodiment provided by the present invention can be combined with each other arbitrarily to form a feasible technical solution. Such combination is not restricted by the order of steps and / or the mode of structural composition, but must be based on the fact that those of ordinary skill in the art can implement it. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0021] Figure 1 The following is a flowchart of a geographic scene area modeling method based on spatio-temporal big data analysis provided by an embodiment of the present invention. As Figure 1 shown, the method includes:

[0022] Step 1: Obtain a spatio-temporal big data set within a geographic scene area, where the spatio-temporal big data set includes a plurality of spatial data points.

[0023] Among them, the spatio-temporal big data set within the geographic scene area can be expressed as:

[0024] ;

[0025] Among them, is the geographic coordinate of the i-th spatial data point, is the timestamp, is the observed variable value, including temperature, humidity, or pedestrian flow, etc.

[0026] Step 2: Obtain the spatio-temporal features of each of the spatial data points.

[0027] It can be understood that since spatio-temporal big data has complex temporal and spatial features, the spatio-temporal coupling and heterogeneity should be concerned when extracting the spatio-temporal features of spatial data points. This step is the basis of the entire modeling process, and the extracted spatio-temporal features will be used as the input for subsequent modeling steps, determining the accuracy and stability of the modeling effect.

[0028] In an embodiment of the present invention, the obtaining the spatio-temporal features of each of the spatial data points includes:

[0029] Step 21: Calculate the spatial density feature, time evolution feature, and context feature of each spatial data point respectively.

[0030] Among them, the spatial density feature, the time evolution feature, and the context feature of each spatial data point are calculated respectively, including:

[0031] ;

[0032] Among them, represents the number of all spatial data points in the geographical scene area, that is, the total number of spatial data points involved in the spatio-temporal feature extraction within the geographical scene area range, is the spatial data point and the spatial data point is the spatial distance between them, is the spatial scale parameter, is the spatial density feature of the spatial data point .

[0033] ;

[0034] Among them, is the time evolution feature, representing the change rate in the time dimension, is the change amount of the observed variable value, is the time interval.

[0035] For the context feature of each spatial data point, among them, the acquisition of the context feature mainly depends on neighborhood analysis, statistical calculation, and time series analysis. First, through neighborhood analysis, several data points around a certain area can be selected, and their average value or weighted average value can be calculated to obtain the local feature of the area. Second, statistical calculation can be used to measure the degree of change within the area, such as calculating the standard deviation, coefficient of variation, or dispersion degree of the neighborhood data to determine the consistency or diversity of the area feature. In addition, time series analysis can be used to obtain dynamic context features, that is, to analyze the change trend of a certain area in different time periods.

[0036] Step 22, according to the spatial density feature, the time evolution feature, and the context feature of each spatial data point, calculate the spatio-temporal feature of each spatial data point.

[0037] Among them, the spatio-temporal feature of each spatial data point is:

[0038] ;

[0039] Among them, is the spatio-temporal feature of the spatial data point , is the context feature of the spatial data point , , and are feature weight coefficients.

[0040] The spatio-temporal features of each spatial data point extracted in this step will be used as the input for subsequent modeling steps, providing a basic feature representation for regional modeling.

[0041] Step 3: Based on the spatio-temporal features, calculate the correlation value between each spatial data point and other spatial data points, and screen out key spatial data points from the spatio-temporal big data set based on the correlation value.

[0042] It can be understood that since the number of spatial data points in the geographical scene area is relatively large, some spatial data points have weak correlation with other spatial data points and are not very useful for subsequent scene area modeling. To reduce the complexity of subsequent modeling, key spatial data points are screened out from the spatial data points extracted from the geographical scene area, and then the geographical scene area is modeled based on the key spatial data points.

[0043] In the embodiment of the present invention, key spatial data points are screened out based on the correlation between different spatial data points. In one embodiment of the present invention, based on the spatio-temporal features, calculating the correlation value between each spatial data point and other spatial data points includes:

[0044] ;

[0045] wherein is the spatio-temporal feature of spatial data point and the spatio-temporal feature of spatial data point ; is and the covariance between; , are respectively and the variances of.

[0046] The screening out of key spatial data points from the spatio-temporal big data set based on the correlation value includes:

[0047] According to the calculated correlation value between the spatio-temporal features of every two spatial data points , a feature correlation matrix is formed;

[0048] Performing eigenvalue decomposition on the feature correlation matrix : ;

[0049] wherein is the eigenvector matrix, ​​is an eigenvalue diagonal matrix, representing the distribution of spatio-temporal features. Among them, includes multiple element values, and each element value characterizes the distribution of spatio-temporal features of the corresponding spatial data point.

[0050] According to the element values in the eigenvalue diagonal matrix the key spatial data points are screened out. Among them, when the element value is greater than the preset threshold, the spatial data point corresponding to the element value is a key spatial data point; otherwise, the spatial data point corresponding to the element value is not a key spatial data point.

[0051] According to the correlation degrees of different spatial data points, the key spatial data points in the geographical scene area are screened out. This step aims to reduce feature redundancy and ensure that the joint effects of key features can be captured when constructing the model.

[0052] Step 4, obtain the spatio-temporal patterns of each of the key spatial data points, and based on the spatio-temporal patterns of each of the key spatial data points, divide the geographical scene area into multiple sub-regions.

[0053] It can be understood that after determining the key spatial data points, spatio-temporal pattern mining is entered, and the spatio-temporal evolution characteristics existing in the region are identified by extracting typical spatio-temporal patterns.

[0054] In an embodiment of the present invention, the obtaining the spatio-temporal patterns of each of the key spatial data points, and based on the spatio-temporal patterns of each of the key spatial data points, dividing the geographical scene area into multiple sub-regions includes:

[0055] Obtain the spatio-temporal patterns of each of the key spatial data points;

[0056] According to the spatio-temporal patterns of each of the key spatial data points, obtain the key spatial data points included in each type of spatio-temporal pattern;

[0057] According to the key spatial data points included in each type of spatio-temporal pattern, calculate the pattern eigenvalue of each type of spatio-temporal pattern:

[0058] ;

[0059] Among them, is the pattern eigenvalue of the th type of spatio-temporal pattern. Using the form of weighted summation can ensure the interpretability of the calculation result, so that as the numerical expression of the spatio-temporal pattern can clearly establish a connection with each feature variable; is the number of key spatial data points included in the th type of spatio-temporal pattern; is the feature weight of the th key spatial data point, For key spatial data points of spatio-temporal characteristics;

[0060] Based on a density clustering algorithm (such as DBSCAN), cluster the pattern feature values of multiple spatio-temporal patterns to form spatio-temporal pattern clusters, and obtain the key spatial data points included in each of the spatio-temporal pattern clusters.

[0061] Among them, the clustering algorithm is expressed as:

[0062]

[0063] In the formula, is the class pattern region, is the similarity measure between the h-th spatio-temporal pattern and the g-th spatio-temporal pattern, used to measure the similarity degree between two patterns. The Euclidean distance, cosine similarity or other measurement methods can be used to calculate the pattern similarity between different patterns, is the pattern discrimination threshold.

[0064] Traverse all patterns , find all patterns that satisfy , and classify them into the same category . If a certain pattern does not satisfy any existing category, create a new category , and iterate until all pattern points are classified.

[0065] Through pattern mining, typical spatio-temporal characteristics inside the region can be extracted, providing a basis for region division in subsequent modeling.

[0066] According to the key spatial data points included in each of the spatio-temporal pattern clusters, divide the geographical scene region to obtain multiple sub-regions, where one spatio-temporal pattern cluster corresponds to one sub-region.

[0067] Step 5, based on the spatio-temporal patterns included in each sub-region, calculate the regional feature value of each sub-region, and based on the regional feature value of each sub-region, calculate the attribute feature value of each sub-region.

[0068] It can be understood that after step 4 divides the geographical scene region into multiple sub-regions according to the spatio-temporal patterns, this step analyzes the characteristics of each sub-region.

[0069] In an embodiment of the present invention, calculating the regional feature value of each sub-region based on the spatio-temporal patterns included in each sub-region includes:

[0070]

[0071] Among them, is the regional eigenvalue of the m-th sub-region, represents the total number of spatio-temporal pattern categories included in the m-th sub-region, is the spatio-temporal pattern coefficient of the class, indicating the influence intensity of the spatio-temporal pattern in the sub-region.

[0072] Then, an attribute feature model is constructed for each sub-region to quantitatively describe the feature performance of different sub-regions in different scenarios.

[0073] In an embodiment of the present invention, calculating the attribute eigenvalue of each sub-region based on the regional eigenvalue of each sub-region includes:

[0074]

[0075] wherein, is the attribute eigenvalue of the m-th sub-region;

[0076] W is the number of regional eigenvalues within the sub-region;

[0077] is the weight coefficient of the w-th regional eigenvalue, reflecting the contribution degree of the regional eigenvalue to the regional attribute;

[0078] is the regional eigenvalue mapping function, wherein:

[0079] ;

[0080] wherein, is the mapping coefficient, adjusting the shape of the eigenvalue curve.

[0081] Among them, when analyzing the attribute features of each sub-region, according to the sub-region division result, the regional eigenvalue is calculated for each sub-region . Through the mapping function the regional eigenvalue is non-linearly normalized to enhance the feature sensitivity. Using the feature weight calculate the regional attribute eigenvalue . Identify the sub-region priority and feature significance through the sorting of the regional attribute eigenvalues.

[0082] Step 6, calculate the regional evolution index of each sub-region based on the attribute eigenvalue of each sub-region.

[0083] It can be understood that after obtaining the attribute features of each sub-region, analyze the dynamic change trend of the attribute features of the sub-region over time and space to identify the evolution characteristics of the sub-region attribute features in the spatio-temporal dimension.

[0084] In one embodiment of the present invention, calculating the regional evolution index of each sub-region based on the attribute characteristic values of each sub-region includes:

[0085] Step 61: Calculate the attribute change rate of each sub-region according to the attribute characteristic values of each sub-region at different times.

[0086] Among them, calculating the attribute change rate of each sub-region according to the attribute characteristic values of each sub-region at different times includes:

[0087] ;

[0088] Among them, is the attribute change rate of the m-th sub-region at time t, is the time interval, is the attribute characteristic value of the m-th sub-region at time t, is the m-th sub-region at the attribute characteristic value at time.

[0089] Step 62: Calculate the attribute change acceleration of each sub-region according to the attribute change rates of each sub-region at different times.

[0090] Among them, calculating the attribute change acceleration of each sub-region according to the attribute change rates of each sub-region at different times:

[0091] ;

[0092] Among them, represents the attribute change acceleration of the m-th sub-region at time t, represents the m-th sub-region at the attribute change rate at time, represents the attribute change rate of the m-th sub-region at time t.

[0093] Step 63: Establish a regional dynamic evolution function according to the attribute change rate and attribute change acceleration of each sub-region, and calculate the regional evolution index of each sub-region.

[0094] Establishing a regional dynamic evolution function according to the attribute change rate and attribute change acceleration of each sub-region, and calculating the regional evolution index of each sub-region:

[0095] ;

[0096] Among them, represents the regional evolution index of the m-th sub-region, reflecting the intensity of sub-region feature changes; 、 It is a weight parameter that controls the contribution ratio of the rate of change and acceleration.

[0097] Calculate the regional evolution index of each sub-region To identify the intensity of change in the sub-region.

[0098] By analyzing the spatial distribution of the regional evolution index, it is possible to identify whether the sub-region is a drastically changing region or a relatively stable region. Specifically, when the regional evolution index of each sub-region is greater than a preset threshold, the mth sub-region is a drastically changing sub-region. When the regional evolution index of each sub-region is less than or equal to the preset threshold, the mth sub-region is a relatively stable sub-region.

[0099] Step 7: Calculate the characteristic distance between different sub-regions according to the attribute characteristic value and the regional evolution index of each sub-region, and cluster the multiple sub-regions based on the characteristic distance to form different regional clusters.

[0100] It is understandable that after completing the dynamic change analysis of regional attributes, cluster analysis is performed on the features of different sub-regions to identify a set of regions with similar features.

[0101] Based on the regional attribute feature values ​​obtained in step 5 and the regional evolution index obtained in step 6 , using clustering methods to identify regions with similar characteristics.

[0102] Among them, the attribute characteristic value of each sub-area is and regional evolution index Normalize to get the normalized attribute feature value and the normalized regional evolution index ;

[0103] Normalize the attribute feature value and the normalized regional evolution index Combine into feature vector ;

[0104] Calculate the feature distances of different sub-regions based on the feature vectors:

[0105] ;

[0106] in, represents the characteristic distance between the mth sub-region and the wth sub-region, is the normalized regional evolution index of the w-th sub-region, is the normalized attribute feature value of the w-th sub-region.

[0107] Use K - means clustering on the feature distance matrix to calculate the distance between regions and cluster centers, thereby determining the cluster to which each sub - region belongs, and extracting region clusters with similar features. The feature distance matrix reflects to a certain extent the comprehensive differences in the evolution characteristics and attribute features between regions. This comprehensive feature provides a scientific and reasonable basis for similarity evaluation in K - means clustering, ensuring the accuracy and rationality of clustering.

[0108] After completing the clustering analysis, optimize and verify the entire geographical scene area modeling method. The optimization and verification include:

[0109] Error evaluation: Calculate the sum of squared errors of clustering. If the sum of squared errors of clustering is large, it indicates that the distribution differences of regional features are large and the similarity of features within the cluster is low. At this time, the cluster centers and parameters need to be readjusted.

[0110] Model verification: Adopt the cross - validation method, randomly select some data for clustering verification. Randomly select some regional data as the validation set, and the remaining data as the training set. After clustering and modeling the training set, use the validation set to calculate the clustering error and compare the model effects. Use the average error rate and standard deviation as model verification indicators to evaluate the performance fluctuations of the model under different data distributions. If the model shows large fluctuations in different validation sets, it indicates that the model is highly sensitive to data distribution and the feature selection or optimization algorithm parameters need to be adjusted.

[0111] Parameter tuning: Based on the model error and verification effect, update the model parameters and adjust the feature weight coefficients and the number of clusters, thereby adjusting the contribution ratio of features to the clustering effect. On the basis of adjusting the feature weights and the number of clusters, use gradient descent or genetic algorithms for automatic tuning to avoid biases caused by manual adjustment.

[0112] See Figure 2 , which provides a geographical scene area modeling system based on spatio - temporal big data analysis according to an embodiment of the present invention. The system includes:

[0113] An acquisition module 201, configured to acquire a spatio - temporal big data set within a geographical scene area, where the spatio - temporal big data set includes multiple spatial data points; and acquire the spatio - temporal features of each of the spatial data points;

[0114] A first calculation module 202, configured to calculate the association value between each of the spatial data points and other spatial data points based on the spatio - temporal features, and screen out key spatial data points from the spatio - temporal big data set based on the association value;

[0115] A partitioning module 203, configured to obtain the spatio-temporal patterns of each of the key spatio-temporal data points, and divide the geographical scene area into multiple sub-areas based on the spatio-temporal patterns of each of the key spatio-temporal data points;

[0116] A second calculation module 204, configured to calculate the regional feature values of each sub-area based on the spatio-temporal patterns included in each sub-area, and calculate the attribute feature values of each sub-area based on the regional feature values of each sub-area; and further configured to calculate the regional evolution index of each sub-area based on the attribute feature values of each sub-area;

[0117] A clustering module 205, configured to calculate the feature distances between different sub-areas according to the attribute feature values and the regional evolution indices of each sub-area, and cluster the multiple sub-areas based on the feature distances to form different regional clusters.

[0118] It can be understood that a geographical scene area modeling system based on spatio-temporal big data analysis provided by the present invention corresponds to the geographical scene area modeling method based on spatio-temporal big data analysis provided in the foregoing embodiments. The related technical features of the geographical scene area modeling system based on spatio-temporal big data analysis can refer to the related technical features of the geographical scene area modeling method based on spatio-temporal big data analysis, which will not be elaborated herein.

[0119] Please refer to Figure 3 , Figure 3 which is a schematic diagram of an embodiment of an electronic device provided by an embodiment of the present invention. As Figure 3 shown, an embodiment of the present invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored on the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, the steps of the geographical scene area modeling method based on spatio-temporal big data analysis are implemented.

[0120] Please refer to Figure 4 , Figure 4 which is a schematic diagram of an embodiment of a computer-readable storage medium provided by the present invention. As Figure 4 shown, this embodiment provides a computer-readable storage medium 400, on which a computer program 411 is stored. When the computer program 411 is executed by a processor, the steps of the geographical scene area modeling method based on spatio-temporal big data analysis are implemented.

[0121] A method and system for geographical scene area modeling based on spatio-temporal big data analysis provided by an embodiment of the present invention systematically depicts regional characteristics and their spatio-temporal evolution characteristics by performing feature correlation analysis and spatio-temporal pattern mining. This method effectively solves the problems of insufficient regional feature extraction and lack of dynamics in the prior art through non-linear feature mapping and comprehensive feature extraction. At the same time, based on regional dynamic evolution analysis, the accuracy of regional feature similarity calculation can be significantly improved. In the clustering analysis stage, through the K-means clustering algorithm combined with the feature distance matrix, accurate classification of regional patterns is achieved, effectively improving the ability to extract feature-similar regions in complex geographical scenes.

[0122] It should be noted that in the above embodiments, the descriptions of each embodiment have their own focuses. For parts not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0123] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0124] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more flows and / or Figure 1 blocks or multiple blocks.

[0125] These computer program instructions can also be stored in a computer-readable memory capable of guiding a computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one or more flows and / or Figure 1 blocks or multiple blocks.

[0126] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus, causing a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 one block or a plurality of blocks.

[0127] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to cover the preferred embodiments as well as all changes and modifications falling within the scope of the present invention.

[0128] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A method for modeling geographical scene regions based on spatio-temporal big data analysis, characterized in that, Including: Obtain a spatio-temporal big data set within a geographical scene area, where the spatio-temporal big data set includes a plurality of spatial data points; Obtain the spatio-temporal characteristics of each of the spatial data points; Based on the spatio-temporal characteristics, calculate the correlation value between each of the spatial data points and other spatial data points, and screen out key spatial data points from the spatio-temporal big data set based on the correlation value; Obtain the spatio-temporal pattern of each of the key spatial data points, and divide the geographical scene area into multiple sub-areas based on the spatio-temporal pattern of each of the key spatial data points; Based on the spatio-temporal patterns included in each sub-area, calculate the regional characteristic value of each sub-area, and calculate the attribute characteristic value of each sub-area based on the regional characteristic value of each sub-area; Based on the attribute characteristic value of each sub-area, calculate the regional evolution index of each sub-area; According to the attribute characteristic value and the regional evolution index of each sub-area, calculate the characteristic distance between different sub-areas, and cluster multiple sub-areas based on the characteristic distance to form different regional clusters.

2. The method for modeling geographical scene regions according to claim 1, characterized in that, The obtaining of the spatio-temporal big data set within the geographical scene area includes: ； wherein, is the geographic coordinate of the i-th spatial data point, is the timestamp, is the observed variable value, and the observed variable value includes temperature, humidity or the number of people; The obtaining of the spatio-temporal characteristics of each of the spatial data points includes: Calculate the spatial density characteristic, the time evolution characteristic and the context characteristic of each spatial data point respectively; According to the spatial density characteristic, the time evolution characteristic and the context characteristic of each spatial data point, calculate the spatio-temporal characteristic of each spatial data point.

3. The method for modeling geographical scene regions according to claim 2, characterized in that, The calculating of the spatial density characteristic, the time evolution characteristic and the context characteristic of each spatial data point respectively includes: ； Among them, represents the number of all spatial data points within the geographical scene area, is the spatial data point and the spatial data point is the spatial distance therebetween, is the spatial scale parameter, is the spatial density feature of the spatial data point ; ； Among them, is the time evolution feature, representing the change rate in the time dimension, is the change amount of the observed variable value, is the time interval; The spatio-temporal characteristic of each spatial data point is: ； Among them, is the spatio-temporal feature of the spatial data point , is the context feature of the spatial data point , , and are feature weight coefficients.

4. The method for modeling geographical scene regions according to claim 1, characterized in that, The calculating of the correlation value between each of the spatial data points and other spatial data points based on the spatio-temporal characteristics includes: ； Among them, is the spatio-temporal feature of the spatial data point and the spatio-temporal feature of the spatial data point and the correlation value therebetween; is the covariance between and; , are respectively the variances of and; The screening out of key spatial data points from the spatio-temporal big data set based on the correlation value includes: According to the correlation values between the spatio-temporal features of every two spatial data points calculated , a feature correlation matrix is formed ; Perform eigenvalue decomposition on the feature correlation matrix : ; Among them, is the eigenvector matrix, is the eigenvalue diagonal matrix, representing the distribution of spatio-temporal features; According to the magnitude of the element values in the eigenvalue diagonal matrix key spatial data points are screened out. Among them, when the element value is greater than a preset threshold, the spatial data point corresponding to the element value is a key spatial data point; otherwise, the spatial data point corresponding to the element value is not a key spatial data point.

5. The method for modeling geographical scene regions according to claim 1, characterized in that, The obtaining of the spatio-temporal pattern of each of the key spatial data points and dividing the geographical scene area into multiple sub-areas based on the spatio-temporal pattern of each of the key spatial data points includes: Obtain the spatio-temporal pattern of each of the key spatial data points; According to the spatio-temporal pattern of each of the key spatial data points, obtain the key spatial data points included in each type of spatio-temporal pattern; According to the key spatial data points included in each type of spatio-temporal pattern, calculate the pattern characteristic value of each type of spatio-temporal pattern: ； Among them, is the mode eigenvalue of the th type of spatio-temporal mode; is the number of key spatial data points included in the th type of spatio-temporal mode; is the characteristic weight of the th key spatial data point, is the spatio-temporal characteristic of the key spatial data point . Cluster the pattern characteristic values of multiple types of spatio-temporal patterns based on the density clustering algorithm to form spatio-temporal pattern clusters, and obtain the key spatial data points included in each of the spatio-temporal pattern clusters; Divide the geographical scene area according to the key spatial data points included in each of the spatio-temporal pattern clusters to obtain multiple sub-areas, where one spatio-temporal pattern cluster corresponds to one sub-area.

6. The geographical scene area modeling method according to claim 1, wherein, The calculating of the regional characteristic value of each sub-area based on the spatio-temporal patterns included in each sub-area includes:

7. Among them, is the regional feature value of the m-th sub-region, represents the total number of spatio-temporal pattern categories included in the m-th sub-region, is the spatio-temporal pattern coefficient of the n-th class; The calculating of the attribute characteristic value of each sub-area based on the regional characteristic value of each sub-area includes:

8. Among them, is the attribute feature value of the m-th sub-region; W is the number of regional characteristic values within the sub-area; is the weight coefficient of the w-th regional eigenvalue, reflecting the contribution degree of the regional eigenvalue to the regional attribute; is the regional feature value mapping function, where: ； Among them, is the mapping coefficient.

9. The geographical scene area modeling method according to claim 1, wherein, The calculating of the regional evolution index of each sub-region based on the attribute characteristic value of each sub-region includes: According to the attribute characteristic values ​​of each sub-region at different times, the attribute change rate of each sub-region is calculated; According to the attribute change rate of each sub-area at different times, the attribute change acceleration of each sub-area is calculated; According to the attribute change rate and attribute change acceleration of each sub-region, a regional dynamic evolution function is established to calculate the regional evolution index of each sub-region.

10. The geographical scene area modeling method according to claim 7, wherein, The calculating the attribute change rate of each sub-region according to the attribute characteristic value of each sub-region at different times includes: ； Among them, is the attribute change rate of the m-th sub-region at time t, is the time interval, is the attribute characteristic value of the m-th sub-region at time t, is the attribute characteristic value of the m-th sub-region at time; The attribute change acceleration of each sub-region is calculated according to the attribute change rate of each sub-region at different times: ； Among them, represents the acceleration of attribute change of the m-th sub-region at time t, represents the rate of attribute change of the m-th sub-region at time, represents the rate of attribute change of the m-th sub-region at time t; According to the attribute change rate and attribute change acceleration of each sub-region, a regional dynamic evolution function is established to calculate the regional evolution index of each sub-region: ； Among them, represents the regional evolution index of the m-th sub-region, reflecting the intensity of the change in sub-region characteristics; , are weight parameters that control the contribution ratio of the change rate and the acceleration.

11. The geographical scene area modeling method according to claim 1, wherein, The method further comprises calculating the regional evolution index of each sub-region based on the attribute characteristic value of each sub-region, and then: When the regional evolution index of each sub-region is greater than the preset threshold, the mth sub-region is a sub-region with drastic changes; When the regional evolution index of each sub-region is less than or equal to a preset threshold, the mth sub-region is a relatively stable sub-region.

12. The geographical scene area modeling method according to claim 1, wherein, The calculating of the characteristic distance between different sub-regions according to the attribute characteristic value and the regional evolution index of each sub-region includes: The attribute feature values of each sub-region are respectively and the regional evolution index are normalized to obtain the normalized attribute feature value and the normalized regional evolution index ; Combine the normalized attribute eigenvalue and the normalized region evolution index into a feature vector ; Calculate the feature distances of different sub-regions based on the feature vectors: ； Among them, represents the characteristic distance between the m-th sub-region and the w-th sub-region, is the normalized regional evolution index of the w-th sub-region, is the normalized attribute eigenvalue of the w-th sub-region.

Images