Method and system for identifying urban mixed-use areas based on multi-source remote sensing geographic data

Abstract

This invention discloses a method and system for identifying mixed-function urban areas based on multi-source remote sensing geographic data, belonging to the field of remote sensing image recognition technology. The method includes: acquiring remote sensing image data and road network data of the study area, performing preprocessing and data overlay to obtain basic street block units of the study area; acquiring POI data of the study area, performing preprocessing, and extracting a first dataset from the basic street block units using a POI distribution information entropy-guided sample extraction method; acquiring AOI data of the study area, performing preprocessing, and extracting a third dataset from the basic street block units using a multi-view difference information learning method; merging the first dataset and the third dataset to obtain a comprehensive dataset; and training a neural network model using the comprehensive dataset for urban functional area classification. This invention improves the accuracy of single-function area identification and can accurately capture the true layout of various mixed-function areas in complex urban environments.

Description

Technical Field

[0001] This invention belongs to the field of remote sensing image processing and recognition technology, specifically relating to a method and system for identifying mixed functional zones in cities based on multi-source remote sensing geographic data. Background Technology

[0002] With the rapid development of urbanization in my country, a thorough understanding of the current state of urban development is crucial for promoting economic growth, improving people's quality of life, and driving comprehensive and balanced high-quality urban development. Unlike urban zoning such as city centers and suburbs, urban functional zones emphasize the concentration and enhancement of the main functions within a region, such as industrial zones, residential areas, and commercial zones. As an important component of urban planning, urban functional zones not only help urban planners and managers rationally plan urban construction and promote balanced development across different areas, but also coordinate the functional layout of mixed-use areas, thereby achieving healthy urban development. Furthermore, identifying urban functional zones helps in understanding their spatial distribution characteristics, intensively and economically utilizing land resources, improving urban land utilization, optimizing resource allocation, and playing a vital role in the scientific planning of cities and the optimization of the living environment.

[0003] Urban functional zones are not always single-purpose; mixed-use development is quite common. These areas encompass multiple functions, such as mixed residential and commercial types, and communities combining public services and commercial functions. The advantages of mixed-use zones are significant: they can provide diverse living and working environments, enhancing urban flexibility; the coexistence of multiple functions can promote community vitality and improve residents' quality of life; rational layout can optimize urban resource allocation and improve land use efficiency; more importantly, mixed-use zones are better able to adapt to changing urban development needs, enhancing the city's sustainable development capabilities. Therefore, accurately identifying mixed-use urban zones is crucial.

[0004] In recent years, remote sensing imagery has received widespread attention due to its ability to objectively reflect the characteristics of ground objects. It contains rich information and possesses strong identification capabilities. Particularly noteworthy is the development of deep learning technology, which has provided a powerful tool for urban functional zone identification. This technology not only improves the accuracy of identification but also handles large-scale datasets. On the other hand, AOI (Area of ​​Interest) data from crowdsourced geographic information provides detailed geographic boundaries and rich category labels, supplementing the shortcomings of remote sensing imagery in terms of ground feature classification details. The combination of these two technologies significantly enhances the accuracy and detail of urban functional zone identification.

[0005] Currently, urban functional zone identification efforts primarily focus on single-type areas, with relatively insufficient attention paid to mixed-function zones. Traditional urban functional zone identification methods are mostly based on supervised classification, relying heavily on manually labeled data. This not only increases workload but also limits identification efficiency and breadth. Some urban functional zone identification methods combine the advantages of POI data and remote sensing imagery, achieving urban functional zone identification. However, due to the significant differences between different POI categories, relying solely on POI density is insufficient to effectively reveal the main categories of urban functional zones. For example, the number of commercial POIs often far exceeds that of other categories, potentially leading to some areas being misclassified as commercial zones.

[0006] Therefore, the accuracy of existing technologies in identifying mixed functional areas needs to be improved, especially in complex urban environments where it is difficult to accurately capture the true layout of each functional area. Summary of the Invention

[0007] In view of this, the present invention proposes a method and system for identifying mixed urban functional zones based on multi-source remote sensing geographic data, which is used to solve the problem of difficulty in accurately identifying mixed urban functional zones in complex urban environments.

[0008] In a first aspect, this invention discloses a method for identifying mixed functional zones in cities based on multi-source remote sensing geographic data, the method comprising:

[0009] Remote sensing image data and road network data of the study area were acquired and preprocessed. The preprocessed remote sensing image data and road network data were overlaid to obtain the basic street block units of the study area.

[0010] POI data of the study area is acquired and preprocessed. The mixing degree of the functional category to which the basic block unit belongs is calculated based on the preprocessed POI data. A first dataset is extracted from the basic block unit based on the mixing degree. The first dataset includes a first single functional area sample and a first mixed functional area sample.

[0011] Acquire AOI data of the study area and preprocess it, then extract the second single functional area sample based on the preprocessed AOI data.

[0012] The first single-function area sample and the second single-function area sample are merged to obtain the second dataset;

[0013] Based on the second dataset, multi-view difference information is learned, and voting prediction is performed on the basic block units to obtain the third dataset;

[0014] The first dataset and the third dataset are merged to obtain a comprehensive dataset;

[0015] A neural network model is trained using the comprehensive dataset to obtain an urban functional zone identification model, which is then used to classify the study area into urban functional zones.

[0016] Based on the above technical solutions, preferably, the calculation of the mixing degree of the functional category to which the basic street block unit belongs specifically includes:

[0017] Construct a Voronoi diagram based on POI data, and use the Voronoi diagram to calculate the average area of ​​each type of POI data;

[0018] The area weighting ratio of each POI data category is calculated by the average area of ​​each POI data category and the number of POI data categories in each category.

[0019] Based on the area weighting ratio, the mixing degree of the functional category to which any basic block unit in the study area belongs is calculated using the POI distribution information entropy.

[0020] Based on the above technical solutions, preferably, the formula for calculating the area weighting ratio of each type of POI data is as follows:

[0021]

[0022] Where r represents any region within the study area, AP r (k) is the area-weighted proportion of the k-th type of POI in region r, T k The total number of POI types in the k-th class is represented by A(i,k), where k is the type number of a single class, k = 1,...,5; A(i,k) is the average area of ​​the i-th POI type in the k-th class; nr(i,k) is the number of POI types in the k-th class in region r; i = 1,2,...,T k .

[0023] Based on the above technical solutions, preferably, the formula for calculating the mixing degree of the functional category of any basic block unit within the study area using POI distribution information entropy is as follows:

[0024]

[0025] Among them, MD r AP represents the mixing degree of region r corresponding to any basic block unit. r (k) is the area-weighted proportion of the k-th type of POI in region r.

[0026] Based on the above technical solutions, preferably, the step of extracting the first dataset from the basic block unit according to the degree of mixing specifically includes:

[0027] Set a mixing degree threshold. If the mixing degree of region r corresponding to a basic block unit in the study area is less than the mixing degree threshold, then region r is a single functional area, and the category with the largest area weighting ratio in region r is taken as the category label of region r.

[0028] If the mixing degree of a basic block unit within the study area is greater than or equal to the mixing degree threshold, then the area r is a mixed functional area, and the category with an area weighting ratio greater than a preset threshold within the area r is used as the category label of the area r.

[0029] The first dataset is constructed based on each basic street block unit and its corresponding category label.

[0030] Based on the above technical solutions, preferably, the third dataset is obtained by learning multi-view difference information based on the second dataset and performing voting prediction on the basic street blocks, specifically including:

[0031] The second dataset is divided into N data subsets, and N neural network models are trained using each of the N data subsets.

[0032] The third dataset is obtained by using N trained neural network models to predict the basic street block units through voting.

[0033] Based on the above technical solutions, preferably, the step of using N trained neural network models to perform voting predictions on the basic street block units to obtain the third dataset specifically includes:

[0034] The N trained neural network models are used to predict the voting results for each basic street block, and the final label for each image is determined by a majority voting mechanism.

[0035] For cases where the prediction results show mixed labels, the prediction results of the N trained neural network models are statistically analyzed, and the label combination supported by the majority of models is selected as the final mixed label for the corresponding image.

[0036] In a second aspect, the present invention discloses an urban mixed-functional zone identification system based on multi-source remote sensing geographic data, the system comprising:

[0037] Data overlay module: used to acquire remote sensing image data and road network data of the study area and preprocess them, and overlay the preprocessed remote sensing image data with the road network data to obtain the basic street block units of the study area;

[0038] First Dataset Module: Used to acquire POI data of the study area and preprocess it, calculate the mixing degree of the functional category to which the basic block unit belongs based on the preprocessed POI data, and extract the first dataset from the basic block unit based on the mixing degree; the first dataset includes a first single functional area sample and a first mixed functional area sample;

[0039] The second dataset module is used to acquire and preprocess AOI data of the study area, extract second single functional area samples based on the preprocessed AOI data, and merge the first single functional area samples with the second single functional area samples to obtain the second dataset.

[0040] The third dataset module is used to learn multi-view difference information based on the second dataset and to perform voting prediction on the basic block units to obtain the third dataset.

[0041] Functional area classification module: used to merge the first dataset and the third dataset to obtain a comprehensive dataset; train a neural network model using the comprehensive dataset to obtain an urban functional area identification model; and classify the study area into urban functional areas using the urban functional area identification model.

[0042] A third aspect of the present invention discloses an electronic device comprising: at least one processor, at least one memory, a communication interface, and a bus;

[0043] The processor, memory, and communication interface communicate with each other through the bus.

[0044] The memory stores program instructions that can be executed by the processor, which invokes the program instructions to implement the method as described in the first aspect of the present invention.

[0045] In a fourth aspect, the present invention discloses a computer-readable storage medium storing computer instructions that cause a computer to perform the method described in the first aspect of the present invention.

[0046] The present invention has the following advantages over the prior art:

[0047] 1) This invention combines deep learning, remote sensing imagery, and AOI and POI data to form multi-source geographic data. Based on the single-type functional area labels provided by the multi-source geographic data, a sample extraction method based on the mixing degree of regional functional categories is proposed. Furthermore, the single-functional area samples from the multi-source data are combined to learn multi-view difference information, effectively and automatically filtering reliable samples containing single and mixed functional labels. Thus, the identification of mixed urban functional areas is carried out using remote sensing imagery, further improving the ability to identify urban functional areas.

[0048] 2) This invention proposes a sample extraction method guided by POI distribution information entropy. This method uses area-weighted ratios to measure the land use ratio of different functional zones and introduces the concept of POI distribution information entropy to quantify the mixing degree of functional categories within a region. The mixing degree of a region reflects the distribution of POIs of different functional categories within that region. Based on the mixing degree value, the dominant type and its mixed components of each functional zone can be evaluated and identified, thereby more accurately guiding the extraction of urban functional zone samples. This method can not only accurately reflect the actual use and complexity of each functional zone, but also provide a data foundation for subsequent analysis.

[0049] 3) This invention uses a multi-view difference information learning method, which uses single functional area samples extracted from multi-source data to train multiple models, and uses a multi-model voting mechanism to predict the labels of basic street units. This not only fills in potential data gaps, but also increases the number of samples in mixed functional areas.

[0050] 4) This invention combines the sample extraction method guided by POI distribution information entropy with the multi-view difference information learning method, which not only improves the accuracy of single functional area identification, but also ensures the accurate capture of the real layout of each mixed functional area in a complex urban environment, thus providing strong support for urban planning and management. By analyzing a large amount of spatiotemporal data, the spatial distribution characteristics and changing trends of urban functional areas can be revealed. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart of a method for identifying mixed-function urban areas based on multi-source remote sensing geographic data, provided in an embodiment of the present invention. Detailed Implementation

[0053] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0054] This invention aims to identify mixed-use urban areas, building upon the existing methods for identifying single-function urban areas. It combines deep learning, remote sensing imagery, AOI (Area of ​​Interest) data, and POI (Point of Interest) data. AOI data provides detailed geographic boundaries and rich category labels to supplement the lack of detail in land cover classification in remote sensing imagery. Furthermore, it employs a POI area weighting ratio to more accurately reveal the functional characteristics of urban areas, thereby enhancing the ability to identify mixed-use urban areas.

[0055] It is understood that before using the method of this invention, the classification system of urban functional zones should be determined first. This invention divides urban functional zones into 5 single-functional zone categories and 2 mixed-functional zone categories, namely, residential land, commercial service facility land, industrial land, public management and public service land, green space and square land, mixed commercial and residential land, and mixed residential and public land, for a total of 7 categories.

[0056] Table 1 Functional Area Classification System

[0057]

[0058]

[0059] Please see Figure 1 This invention discloses a method for identifying mixed functional zones in cities based on multi-source remote sensing geographic data, the method comprising:

[0060] S1. Acquire remote sensing image data and road network data of the study area and preprocess them. Overlay the preprocessed remote sensing image data and road network data to obtain the basic street block units of the study area.

[0061] In this embodiment of the invention, remote sensing data acquisition and processing are performed on the Google Earth Engine (GEE) platform. Sentinel-2 10m and 20m resolution band data are selected, and all images are resampled to 10m. Multiple images are stitched together using ENVI to cover the entire study area.

[0062] In this embodiment of the invention, the road network data originates from an Open Street Map (OSM). First, the road network data is precisely cropped according to the scope of the study area. Then, internal and invalid roads are removed, and a topological check of the roads is performed to ensure that all block boundaries are closed and to prevent overlap between plots. Finally, the processed road network layer is precisely overlaid with remote sensing imagery to form the basic block unit sample for functional zoning mapping. Block_ori .

[0063] S2. Obtain POI data of the study area and preprocess it. Calculate the mixing degree of the functional category to which the basic block unit belongs based on the preprocessed POI data. Extract the first dataset from the basic block unit based on the mixing degree.

[0064] Considering that some functional areas in a city are composed of a mixture of multiple categories, in order to quantify this diversity, this invention proposes a method for extracting urban functional area samples guided by POI distribution information entropy. This method extracts a first dataset from the basic street block unit, specifically including the following steps:

[0065] S21. POI Data Acquisition and Preprocessing.

[0066] In this embodiment of the invention, the original POI data comes from existing map software. The GCJ-02 coordinates of the original POI data are converted to WGS-84 coordinates. Then, the POI data is filtered to remove data with missing, duplicate, erroneous, or inconsistent formats, and some points with low public awareness, such as public toilets, are removed. Next, the POI data is reclassified into five major categories: residential land, commercial service facilities land, industrial land, public management and public service land, and green space and plaza land. Subcategories of each point are retained, ultimately obtaining POI data containing four attributes: name, major category, subcategory, and latitude and longitude.

[0067] S22. Construct a Voronoi diagram based on the preprocessed POI data, and use the Voronoi diagram to calculate the average area of ​​each type of POI data.

[0068] Specifically, by constructing a Voronoi diagram, the average area of ​​each type of POI data is obtained from the POI location information, so that each POI has an area attribute.

[0069] S23. Calculate the area weighting ratio of each type of POI data by using the average area of ​​each type of POI data and the number of POI data in each type.

[0070] In embodiments of the present invention, for any defined region, the area weighting ratio of each type of POI is determined by considering the total number of all types of POIs in the region, the average area of ​​each POI, and the number of a specific type of POI in the region.

[0071] In embodiments of the present invention, for any defined region r, the area-weighted ratio APr(k) of the k-th type POI can be calculated using the following formula:

[0072]

[0073] Wherein: T kA represents the total number of POI types of the kth class; A(i,k) is the average area of ​​the kth class and the i-th POI type; this area can be converted from the POI location information by constructing a Voronoi diagram so that each POI has an area attribute; nr(i,k) is the number of POI types of the kth class and the i-th type in region r.

[0074] The above calculation formulas can better reflect the actual influence of different types of POIs in urban spatial layout, thereby helping to more accurately identify and define the functional areas of the city.

[0075] S24. Based on the area weighting ratio, calculate the degree of mixing of functional categories of any basic block unit within the study area using the POI distribution information entropy.

[0076] Some functional zones in a city are not composed of a single category, but exhibit diverse characteristics due to the mixed existence of multiple categories. To quantify this diversity, the concept of POI distribution information entropy is introduced to quantify the degree of mixing of functional categories within a region, i.e., Mixed Diversity (MD). r Information entropy is a statistical concept used to describe the degree of disorder in a system or the uncertainty of information.

[0077] In an embodiment of the present invention, the formula for calculating the degree of mixing of the functional categories to which any basic block unit within the study area belongs is:

[0078]

[0079] Among them, MD r AP represents the mixing degree of region r corresponding to any basic block unit. r (k) is the area weighted proportion of the k-th type of POI in region r, where k is the type number of a single category, k = 1,...,5.

[0080] MD r The value range is [0, 2.32], when the area weighted ratio AP of all categories r (k) are equal (when both are 1 / 5),

[0081] S25. Extract the first dataset from the basic block unit based on the degree of mixing.

[0082] According to the degree of mixing MD r The value of MD can be used to determine the functional category characteristics of region r: when MD r When the value is close to zero, it indicates that the functional categories within region r are relatively simple and the uncertainty is low. At this time, the region mainly belongs to AP. rThe category with the largest (k) value; when MD r A higher value indicates that the functional categories within the region are more diverse and the uncertainty is higher. In this case, region r can be considered to belong to a mixed category.

[0083] In embodiments of the present invention, a first dataset is extracted by setting a mixing degree threshold to distinguish between single-functional areas and mixed-functional areas within the basic street block unit. Specifically, this is achieved by analyzing the sample... Block_ori The data mixing degree was statistically analyzed at intervals of 0.1 to provide a detailed understanding of the distribution of functional areas within different mixing degree ranges. Further analysis revealed a significant point of change when the mixing degree reached a certain value. A mixing degree threshold was set based on this threshold, and the samples were then weighted according to this threshold and the area weighting ratio of each category. Block_ori The data is divided into single-function area samples. Block_single and Mixed Functional Area Sample Block_mixed .

[0084] In embodiments of the present invention, extracting the first dataset from the basic block unit based on the degree of mixing specifically includes:

[0085] If the mixing degree of region r corresponding to a basic block unit within the study area is less than the mixing degree threshold, then region r is a single functional area, and the category with the largest area weighting ratio in region r is used as the category label of region r; if the mixing degree of region r corresponding to a basic block unit within the study area is greater than or equal to the mixing degree threshold, then region r is a mixed functional area, and the two categories with an area weighting ratio greater than a preset threshold in region r are used as the mixing label of region r; the first dataset is constructed based on each basic block unit and its corresponding category label.

[0086] S3. Obtain AOI data of the study area and preprocess it. Extract the second single functional area sample based on the preprocessed AOI data.

[0087] The initial samples of urban functional areas in this invention are derived from AOI data samples provided by the Open Street Map (OSM). AOI_ori AOI provides detailed geographic information and rich category labels, which typically cover multiple single-purpose categories.

[0088] Based on the study area, samples of single functional areas of AOI within the study area were selected. AOI_ori Based on the research needs of this invention, the samples are reclassified into 5 categories: residential land, commercial service facilities land, industrial land, public management and public service land, and green space and square land.

[0089] S4. Merge the first single functional area sample with the second single functional area sample to obtain the second dataset. Based on the second dataset, perform multi-view difference information learning and vote prediction on the basic block unit to obtain the third dataset.

[0090] Considering that there is no POI data distribution in some basic street blocks, in order to make up for the data omissions and increase the number of samples in mixed functional areas, this invention uses a multi-view difference information learning method to extract a third dataset using samples from a single functional area.

[0091] Specifically, the second dataset is divided into N subsets, each of which is used to independently train a neural network model. N neural network models are trained using each of the N subsets; these trained models are then used to perform voting predictions on the basic street blocks to obtain the third dataset. In this embodiment, the neural network model used is the SE-ResNet50 model.

[0092] In embodiments of the present invention, the step of obtaining a third dataset by voting prediction on the basic street block units using N trained neural network models specifically includes:

[0093] The N trained neural network models are used to predict the voting results for each basic street block, and the final label for each image is determined by a majority voting mechanism.

[0094] For cases where the prediction results show mixed labels, the prediction results of the N trained neural network models are statistically analyzed, and the label combination supported by the majority of models is selected as the final mixed label for the corresponding image.

[0095] S5. Merge the first dataset and the third dataset to obtain a comprehensive dataset. Train a neural network model using the comprehensive dataset to obtain an urban functional area identification model. Use the urban functional area identification model to classify the study area into urban functional areas.

[0096] The first dataset and the third dataset are merged to form a comprehensive dataset, which includes samples of 7 functional areas. These reclassified data are then input into SE-ResNet50 for training to obtain the final urban functional area identification model, so as to achieve the classification of urban functional areas.

[0097] Based on the above method embodiments, the present invention also proposes an urban mixed functional area identification system based on multi-source remote sensing geographic data, the system comprising:

[0098] Data overlay module: used to acquire remote sensing image data and road network data of the study area and preprocess them, and overlay the preprocessed remote sensing image data with the road network data to obtain the basic street block units of the study area;

[0099] First Dataset Module: Used to acquire POI data of the study area and preprocess it, calculate the mixing degree of the functional category to which the basic block unit belongs based on the preprocessed POI data, and extract the first dataset from the basic block unit based on the mixing degree; the first dataset includes a first single functional area sample and a first mixed functional area sample;

[0100] The second dataset module is used to acquire and preprocess AOI data of the study area, extract second single functional area samples based on the preprocessed AOI data, and merge the first single functional area samples with the second single functional area samples to obtain the second dataset.

[0101] The third dataset module is used to learn multi-view difference information based on the second dataset and to perform voting prediction on the basic block units to obtain the third dataset.

[0102] Functional area classification module: used to merge the first dataset and the third dataset to obtain a comprehensive dataset; train a neural network model using the comprehensive dataset to obtain an urban functional area identification model; and classify the study area into urban functional areas using the urban functional area identification model.

[0103] The above system embodiments and method embodiments are one-to-one correspondences. For a brief description of the system embodiments, please refer to the method embodiments.

[0104] The present invention also discloses an electronic device, comprising: at least one processor, at least one memory, a communication interface, and a bus; wherein the processor, memory, and communication interface communicate with each other through the bus; the memory stores program instructions executable by the processor, and the processor calls the program instructions to implement the aforementioned method of the present invention.

[0105] The present invention also discloses a computer-readable storage medium that stores computer instructions, which cause the computer to implement all or part of the steps of the method described in the embodiments of the present invention. The storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0106] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, meaning they can be distributed across multiple network units. Those skilled in the art can select some or all of the modules to achieve the purpose of this embodiment without any inventive effort, based on actual needs.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for identifying mixed-function urban areas based on multi-source remote sensing geographic data, characterized in that, The method includes: Remote sensing image data and road network data of the study area were acquired and preprocessed. The preprocessed remote sensing image data and road network data were overlaid to obtain the basic street block units of the study area. POI data of the study area is acquired and preprocessed. The mixing degree of the functional category to which the basic block unit belongs is calculated based on the preprocessed POI data. A first dataset is extracted from the basic block unit based on the mixing degree. The first dataset includes a first single functional area sample and a first mixed functional area sample. Acquire AOI data of the study area and preprocess it, then extract the second single functional area sample based on the preprocessed AOI data. The first single-function area sample and the second single-function area sample are merged to obtain the second dataset; Based on the second dataset, multi-view difference information is learned, and voting prediction is performed on the basic block units to obtain the third dataset; The first dataset and the third dataset are merged to obtain a comprehensive dataset; A neural network model is trained using the comprehensive dataset to obtain an urban functional zone identification model, which is then used to classify the study area into urban functional zones. 2.The method of claim 1, wherein, The calculation of the mixing degree of the functional category to which the basic block unit belongs specifically includes: Construct a Voronoi diagram based on POI data, and use the Voronoi diagram to calculate the average area of ​​each type of POI data; The area weighting ratio of each POI data category is calculated by the average area of ​​each POI data category and the number of POI data categories in each category. Based on the area weighting ratio, the mixing degree of the functional category to which any basic block unit in the study area belongs is calculated using the POI distribution information entropy.

3. The method for identifying mixed-function urban areas based on multi-source remote sensing geographic data according to claim 2, characterized in that, The formula for calculating the area weighting ratio of each type of POI data is as follows: Where r represents any region within the study area, AP r (k) is the area-weighted proportion of the k-th type of POI in region r, T k The r represents the total number of POI types in the k-th class, where k is the type number of a single class, k = 1, ..., 5; A(i,k) is the average area of ​​the i-th POI type in the k-th class; nr(i,k) is the number of POI types in the k-th class in region r; i = 1, 2, ..., T k . 4.The method of claim 3, wherein, The formula for calculating the mixing degree of the functional category of any basic block unit within the study area using POI distribution information entropy is as follows: Among them, MD r AP represents the mixing degree of region r corresponding to any basic block unit. r (k) is the area-weighted proportion of the k-th type of POI in region r. 5.The method of claim 1, wherein, The step of extracting the first dataset from the basic block unit based on the mixing degree specifically includes: Set a mixing degree threshold. If the mixing degree of region r corresponding to a basic block unit in the study area is less than the mixing degree threshold, then region r is a single functional area, and the category with the largest area weighting ratio in region r is taken as the category label of region r. If the mixing degree of a basic block unit within the study area is greater than or equal to the mixing degree threshold, then the area r is a mixed functional area, and the category with an area weighting ratio greater than a preset threshold within the area r is used as the category label of the area r. The first dataset is constructed based on each basic street block unit and its corresponding category label. 6.The method of claim 1, wherein, Based on the second dataset, multi-view difference information is learned, and voting prediction is performed on the basic block units to obtain the third dataset, which specifically includes: The second dataset is divided into N data subsets, and N neural network models are trained using each of the N data subsets. The third dataset is obtained by using N trained neural network models to predict the basic street block units through voting.

7. The method for identifying mixed-function urban areas based on multi-source remote sensing geographic data according to claim 6, characterized in that, The process of using N trained neural network models to predict the basic street blocks through voting to obtain the third dataset specifically includes: The N trained neural network models are used to predict the voting results for each basic street block, and the final label for each image is determined by a majority voting mechanism. For cases where the prediction results show mixed labels, the prediction results of the N trained neural network models are statistically analyzed, and the label combination supported by the majority of models is selected as the final mixed label for the corresponding image.

8. A multi-source remote sensing geographic data-based urban mixed function area identification system, characterized in that, The system includes: Data overlay module: used to acquire remote sensing image data and road network data of the study area and preprocess them, and overlay the preprocessed remote sensing image data with the road network data to obtain the basic street block units of the study area; First Dataset Module: Used to acquire POI data of the study area and preprocess it, calculate the mixing degree of the functional category to which the basic block unit belongs based on the preprocessed POI data, and extract the first dataset from the basic block unit based on the mixing degree; the first dataset includes a first single functional area sample and a first mixed functional area sample; The second dataset module is used to acquire and preprocess AOI data of the study area, extract second single functional area samples based on the preprocessed AOI data, and merge the first single functional area samples with the second single functional area samples to obtain the second dataset. The third dataset module is used to learn multi-view difference information based on the second dataset and to perform voting prediction on the basic block units to obtain the third dataset. Functional area classification module: used to merge the first dataset and the third dataset to obtain a comprehensive dataset; train a neural network model using the comprehensive dataset to obtain an urban functional area identification model; and classify the study area into urban functional areas using the urban functional area identification model.

9. An electronic device, characterized in that, include: At least one processor, at least one memory, a communication interface, and a bus; The processor, memory, and communication interface communicate with each other through the bus. The memory stores program instructions that can be executed by the processor, which invokes the program instructions to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause the computer to perform the method as described in any one of claims 1 to 7.

Images