An informal urban public space identification method based on multi-source big data

Abstract

The application discloses a kind of informal urban public space identification method based on multi-source big data, comprising: obtaining the multi-source data of the research area where the informal urban public space to be identified is located, and the type information of the informal urban space to be identified;The first basic feature corresponding to the undemarcated property characteristics and the basic characteristics is obtained by processing multi-source data, and the second basic feature data corresponding to the public nature of the informal urban space to be identified, the face domain data of the informal urban space to be identified is obtained based on the first basic feature;The informal urban space to be identified is converted into an equal-distance node graph structure virtual road network;The second basic feature is distributed to the virtual road network node using the shortest path method, and the equal-distance node graph structure data containing the public open feature is obtained;Each node is analyzed for accessibility, and the analysis result is input into the pre-trained extreme gradient boosting algorithm model to obtain the point set data of the informal urban space to be identified.The application can accurately and efficiently analyze the location and boundary of informal urban public space.

Description

Technical Field

[0001] This invention relates to the interdisciplinary fields of digital landscape, smart city design and artificial intelligence, and in particular to a method for identifying informal urban public spaces based on multi-source big data. Background Technology

[0002] Urban public spaces, such as squares, parks, and streets, are vital carriers of urban social, cultural, and ecological functions. Their planning directly impacts residents' quality of life, sustainable urban development, and social equity. Accurate analysis of the scope of urban public spaces helps optimize spatial resource allocation, improve space utilization efficiency, enhance social cohesion, and provide a scientific basis for urban renewal, ecological restoration, and public policy formulation. Furthermore, analyzing the distribution characteristics and current usage of public spaces can support improvements in spatial accessibility, equity, and inclusiveness.

[0003] "Informal urban public spaces" refer to public spaces that are not formally defined by the government or planning departments and lack legal status, but are formed by spontaneous or temporary use by citizens. Their boundaries are often unclear due to characteristics such as ambiguous ownership, time dependence, and mixed functions. Types include mixed-use residential blocks, waterfront spaces, and corner green spaces, which are difficult to capture by traditional maps or planning documents. Therefore, it is necessary to combine analysis with multi-source big data that is more sensitive to time to reveal the real use needs of the city and improve the inclusiveness of urban spaces.

[0004] By applying deep learning algorithms, integrating multi-source big data and high-resolution satellite map information, and conducting computational analysis on a street-based network to assess urban publicness and openness, it becomes possible to automatically identify the location and boundaries of informal public spaces. This approach not only significantly improves work efficiency and reduces human error, but also reveals deeper information through big data analysis, such as usage patterns and trends, providing more accurate and dynamic decision-making support for urban planning and management. Summary of the Invention

[0005] The purpose of this invention is to provide a method for identifying informal urban public spaces based on multi-source big data. This method can accurately and efficiently analyze the location and boundaries of informal urban public spaces based on multi-source data such as buildings, road networks, and water systems. The technical solution adopted by this invention is as follows.

[0006] On the one hand, this invention provides a method for identifying informal urban public spaces based on multi-source big data, including:

[0007] Obtain multi-source data on the study area where the informal urban public spaces to be identified are located, as well as information on the types of informal urban spaces to be identified;

[0008] The multi-source data is processed to obtain the first basic feature and the second basic feature data of the informal urban space to be identified. The first basic feature corresponds to the basic characteristics of the undefined ownership characteristics and the type of informal urban space to be identified, and the second basic feature corresponds to the public nature of the informal urban space to be identified.

[0009] Based on the first basic feature, obtain the area data of the informal urban space to be identified;

[0010] Based on the area data, the informal urban space to be identified is converted into an equidistant node graph structure virtual road network;

[0011] Based on the second basic feature and the equidistant node graph structure virtual road network, the second basic feature is assigned to the associated nodes using the shortest path method to obtain equidistant node graph structure data containing common open features;

[0012] Based on the equidistant node graph structure data containing public and open features, accessibility analysis is performed on each node to obtain a feature dataset including the coordinates and accessibility scores of each node in the equidistant node graph structure.

[0013] The feature dataset is input into a pre-trained extreme gradient boosting algorithm model to obtain point set data of informal urban spaces to be identified.

[0014] Optionally, the multi-source data includes: social media check-in data and map data, wherein the map data includes at least a variety of POI, AOI, street view, road network, water system, water area, building outline, and LBS data;

[0015] The process of processing the multi-source data to obtain the first and second basic feature data of the informal urban space to be identified includes:

[0016] Based on road network, water system, water area, building outline, and AOI data, surface data of the areas to be identified with "openness" and "undefined weight attributes" are obtained from the surface area of ​​the study area and used as the first basic feature data;

[0017] Based on POI, street view, social media check-in, and LBS data, we analyzed and obtained second basic feature data, including check-in point popularity value, street view element area ratio and number, and / or, POI type and location related to public vitality.

[0018] The acquisition of the above-mentioned basic feature data targets several different types of spaces, including waterfront spaces, corner green spaces, and mixed-use blocks. While the acquisition approach is the same, the specific data processing methods differ, including:

[0019] For cases where the informal urban space type to be identified is waterfront space, the acquisition of the area data of the area to be identified includes subtracting buildings, water bodies and all AOIs with actual use functions from the area of ​​the study area;

[0020] For cases where the informal urban space type to be identified is a street corner green space, the acquisition of the area data of the area to be identified includes subtracting the building and street areas from the street area data of the study area;

[0021] For cases where the informal urban space type to be identified is a complex block, the acquisition of the area data of the area to be identified includes, based on the linear data describing the streets in the study area, creating a buffer zone for each street, and subtracting buildings and areas with defined ownership from the buffer zone.

[0022] In addition to the methods mentioned above, mixed-use blocks typically lack clearly defined land use rights or boundaries and may include temporary markets, roadside vendors, etc. Their geographic scope can also be determined based on social media data and street view images. Corner green spaces usually lack official planning and are maintained spontaneously by the community. They are characterized by their location at the intersection of buildings and streets, and their geographic scope can also be determined based on satellite imagery, building data, and street view maps.

[0023] The specific methods for processing multi-source data to obtain the second basic feature data can be found in existing technologies and will not be elaborated further. The first basic feature data can be obtained using existing tools in the ArcGIS platform, such as "erasure" and "segmentation." The second basic feature data is "common" feature data, stored as a point set Shapefile. Subsequently, programs such as Python can be used to distribute the point set label data to the nearest virtual road network nodes based on the shortest path algorithm.

[0024] Optionally, obtaining the area data of the informal urban space to be identified based on the first basic feature data includes:

[0025] If the category of informal urban public space to be identified is waterfront space, then based on the water system Shapefile, water area Shapefile, and area data Shapefile of the area to be identified in the first basic feature data, an outer buffer of a set width is created for the water system and water area within the area to be identified; the original water area within the area to be identified is removed, and the buffers for water area and water system are merged to eliminate the overlapping parts between different buffers; based on all the merged buffer area data and the set coordinate reference system (CRS), a GeoDataFrame file of the informal urban space to be identified, i.e., the urban waterfront area without defined ownership, is generated.

[0026] The above describes the method for obtaining the corresponding unclaimed area data when the informal urban space to be identified is a waterfront space. For cases where the informal urban space to be identified is a mixed-use block or corner green space, the area data of the area to be identified can be directly used to construct a GeoDataFrame file for the corresponding type of informal urban space area.

[0027] Optionally, the step of converting the informal urban space to be identified into an equidistant node graph structure virtual road network based on the area data includes:

[0028] Based on the area data of the informal urban spaces to be identified, define the center point and boundaries of the research scope;

[0029] Within the study area, a virtual road network including nodes, horizontal line segments, and vertical line segments is generated according to the set grid radius. A corresponding GeoDataFrame file is created, and the nodes and line segments of the virtual road network are saved as Shapefile files respectively.

[0030] Optionally, the step of allocating the second basic feature to associated nodes using the shortest path method based on the second basic feature and the equidistant node graph structure virtual road network to obtain equidistant node graph structure data containing common open features includes:

[0031] The feature labels of each feature point in the feature point set corresponding to each "commonality" type in the second basic feature are used to draw perpendicular lines from the feature point to the adjacent line segments based on the feature point coordinates and the node coordinates and line segments in the equidistant node graph structure virtual road network. The line segment connected by the shortest perpendicular line is selected, and the corresponding feature label is assigned to the virtual road network node associated with both ends of the line segment. The association information between the virtual road network node and the corresponding commonality feature label is recorded at the corresponding node in the virtual road network node Shapefile file and at the corresponding feature point in the point set Shapefile file.

[0032] The above technical solution enables bidirectional allocation of the shortest path between public feature labels and virtual road network nodes, providing a data foundation for subsequent objective weighting algorithms.

[0033] Optionally, the reachability analysis of each node based on the equidistant node graph structure data containing public open features includes:

[0034] For each node, calculate its reachability score to nodes with different labels under each common feature type, using the following formula:

[0035]

[0036] In the formula, For the node The reachability score to a certain type of label node under a certain common feature type. This indicates the feature labels assigned to this type in the equidistant node graph structure. The first node in the nth node The nth node, and the nth Each node reaches another node along the road network. Actual walking distance Within the set walking tolerance distance;

[0037] Based on the reachability scores of each node under different labels corresponding to each common feature type, the comprehensive reachability score of each node under each common feature type is obtained.

[0038] Optionally, the step of obtaining a comprehensive reachability score for each node under each common feature type based on the reachability scores of each node for different labels under each common feature type includes:

[0039] For each common feature type, the reachability scores of each node under different labels are normalized using the following formula:

[0040]

[0041] In the formula, For the node Assigned to the equidistant node graph structure The reachability score of all nodes for the class feature label. This represents the number of nodes assigned to each node in the equidistant node graph structure. The set of reachability scores for nodes with class feature labels. and These are the maximum and minimum values ​​in the set, respectively. To The normalized value;

[0042] At this point, for each node, we can obtain the reachability score normalized dataset for all labels under each common feature type. Integrating all N nodes in the equidistant node graph structure, we can obtain the following reachability score normalized dataset matrix for each common feature type:

[0043]

[0044] For each feature label under each common feature type, the weight values ​​of each node corresponding to each feature label are calculated using the CRITIC weighting method based on the node reachability score normalized dataset.

[0045] Based on the weight values ​​of various feature labels corresponding to each node and the normalized value of the accessibility score, calculate the comprehensive accessibility score of each node for each common feature type.

[0046] Optionally, the step of calculating the weight values ​​of each node corresponding to various feature labels using the CRITIC weighting method includes:

[0047] The dataset is normalized based on the reachability scores of all nodes corresponding to each feature label class. Calculate the mean respectively and standard deviation The formula is:

[0048] ,

[0049] Calculate the labels of each feature class and all features under the same common feature type. The formula for determining the degree of contradiction in feature labels is:

[0050]

[0051] In the formula, For the first The contradictory values ​​between the class feature label and all feature labels under the common feature type; To utilize the Pearson correlation coefficient method, based on the normalized dataset of reachability scores corresponding to a single common feature type for all nodes, the 1st... Class feature labels and common feature types The correlation coefficient of class feature labels is calculated using the following formula:

[0052]

[0053] In the above formula, This represents the average of the normalized reachability scores of all nodes corresponding to the Kth class feature label, and its calculation principle is the same as... ;

[0054] Calculate the first Information carrying capacity of class feature labels The formula is: ;

[0055] Calculate the first All class feature labels under the same common feature type Normalized weights in feature labels The formula is:

[0056]

[0057] The formula for calculating the comprehensive reachability score of each node for each common feature type is as follows:

[0058]

[0059] In the formula, For nodes The comprehensive accessibility score for a specific common feature type.

[0060] At this point, for each node, we can obtain a set of <node number, latitude, longitude, first common feature type comprehensive accessibility score, second common feature type comprehensive accessibility score...>. The set of all node arrays is the input to the pre-trained extreme gradient boosting algorithm model. Through supervised training using the labels of each node in a known similar dataset during the pre-training stage, the extreme gradient boosting algorithm model can efficiently identify whether each node is an informal point within the urban space.

[0061] Beneficial effects

[0062] Compared with the prior art, the present invention has the following advantages and advancements:

[0063] First, the area to be identified is delineated through preprocessing, and a virtual road network is constructed. The set of public open feature points is dynamically associated with road network nodes to form an isometric map structure. Then, the CRITIC weighting method is used to objectively assign weights to multi-dimensional public activity feature data, quantifying the overall openness of nodes. Finally, an identification model is constructed based on the extreme gradient boosting algorithm, and combined with accessibility analysis and business distribution patterns to achieve accurate extraction of spatial point sets. This method overcomes the limitations of traditional planning data, achieving accurate identification of informal urban public spaces through dynamic quantification of social activity characteristics, providing data support for flexible urban planning.

[0064] Meanwhile, this invention can be cross-validated through GIS spatial visualization and digital landscape technology. Regarding the presentation of the calculation results, it can achieve vectorized, two-dimensional display, that is, using specific numerical values ​​to identify public spaces, and the results can be intuitively presented in the form of area regions, which helps to increase the accuracy and intuitiveness of the identification results. Attached Figure Description

[0065] Figure 1 The diagram shown is a flowchart of one embodiment of the method of the present invention;

[0066] Figure 2 The diagram shown is a comparison of the visualization results before and after pretreatment of a local water area and water system in an embodiment of the present invention.

[0067] Figure 3 The figure shown is an embodiment of the present invention. Figure 2A schematic diagram of the visualization results of the "urban waterfront area with undefined ownership" obtained by making a buffer zone outward from the central water system;

[0068] Figure 4 As shown Figure 3 A visualization of the lower left corner and extended area;

[0069] Figure 5 The figure shown is a comprehensive public open score matrix of road network nodes and each single POI type in the study area in an embodiment of the present invention.

[0070] Figure 6 The table shown is a weight table of different POI types obtained by analyzing the CRITIC method in an embodiment of the present invention.

[0071] Figure 7 The table shown is a comprehensive score table of public openness of virtual road network nodes in the study area in an embodiment of the present invention;

[0072] Figure 8 The figure shown is a schematic diagram of the recognition results in ArcGIS Pro for the study area in an embodiment of the present invention;

[0073] Figure 9 As shown Figure 8 A partial view within the image. Detailed Implementation

[0074] The following description, in conjunction with the accompanying drawings and specific embodiments, provides further details.

[0075] The identification results of urban public spaces are influenced by multi-source data, including:

[0076] Data directly obtained from open-source map platforms includes: POIs (Points of Interest), city roads, waterways, water bodies, AOIs (Areas of Interest), satellite images, etc.

[0077] Information about social media check-ins includes the location, content, and photos taken at the check-in.

[0078] Other data from online map platforms include street view images, building outlines, and LBS (Location Based Services).

[0079] The technical concept of this invention is as follows: Based on the above-mentioned multi-source data, by combining the shortest path algorithm, the objective weighting method (Criteria Importance Through Intercriteria Correlation, CRITIC), and the extreme gradient boosting algorithm (eXtreme Gradient Boosting, XGBoost), the multi-source data associated with informal urban public spaces are processed to obtain the identification results of informal urban public spaces.

[0080] Example 1

[0081] This embodiment introduces a method for identifying informal urban public spaces based on multi-source big data, referencing... Figure 1 The methods include:

[0082] Obtain multi-source data on the study area where the informal urban public spaces to be identified are located, as well as information on the types of informal urban spaces to be identified;

[0083] The multi-source data is processed to obtain the first basic feature and the second basic feature data of the informal urban space to be identified. The first basic feature corresponds to the basic characteristics of the undefined ownership characteristics and the type of informal urban space to be identified, and the second basic feature corresponds to the public nature of the informal urban space to be identified.

[0084] Based on the first basic feature, obtain the area data of the informal urban space to be identified;

[0085] Based on the area data, the informal urban space to be identified is converted into an equidistant node graph structure virtual road network;

[0086] Based on the second basic feature and the equidistant node graph structure virtual road network, the second basic feature is assigned to the associated nodes using the shortest path method to obtain equidistant node graph structure data containing common open features;

[0087] Based on the equidistant node graph structure data containing public and open features, accessibility analysis is performed on each node to obtain a feature dataset including the coordinates and accessibility scores of each node in the equidistant node graph structure.

[0088] The feature dataset is input into a pre-trained extreme gradient boosting algorithm model to obtain point set data of informal urban spaces to be identified.

[0089] The specific implementation of this embodiment involves the following aspects:

[0090] I. Training of an extreme gradient boosting algorithm model based on multi-source data of informal urban spatial relationships of known types and ranges.

[0091] This section involves the processing and feature extraction of multi-source data. Based on the extracted features, a virtual road network is constructed. The extracted common feature labels are assigned to road network nodes using the shortest path algorithm. The reachability value of each node is calculated based on the virtual road network containing common features. The data of nodes, node coordinates, and reachability values ​​are combined with the labels of informal urban spatial point sets of known types and ranges, and input into an extreme gradient boosting algorithm to complete the training of the stage gradient boosting algorithm model, thereby obtaining an informal urban space identification model for the identification of informal urban spatial point sets of specified types.

[0092] The processing of multi-source data and feature extraction, virtual road network construction, shortest path algorithm allocation of common feature labels, and reachability value calculation during model training are all the same as the corresponding steps in the actual application of the method, which will be described in detail below.

[0093] II. Processing of Multi-Source Data

[0094] refer to Figure 1 This embodiment considers multi-source data including social media check-in data and map data. The map data includes at least POI, AOI, street view, road network, water system, water area, building outline, and LBS data. AOI, road network, water system, water area, and building outline are used to extract area data of the region to be identified and the informal urban space to be identified, in order to construct a virtual road network on this area. Social media check-in, POI, street view, and LBS data are used to extract the common features of points in the region to be identified, for subsequent calculation of accessibility feature values.

[0095] By processing the above two sets of multi-source data, we can obtain the first and second basic characteristics of the study area containing the informal urban spaces to be identified. The first basic characteristic corresponds to the characteristics of undefined ownership, openness, and the basic characteristics of the informal urban space type to be identified. The second basic characteristic corresponds to the public nature of the informal urban spaces to be identified. The processing procedure specifically includes:

[0096] Based on road network, water system, water area, building outline, and AOI data, area data of the areas to be identified with "openness" and "undefined rights attributes" are obtained from the area of ​​the study region as the first basic feature data. Openness means that the area does not contain any area belonging to the interior of any building, and undefined rights attributes mean that the area does not contain any area belonging to any functional AOI or any water area. The basic characteristic of the informal space type to be identified is that, for waterfront space, its basic characteristic is that it is an area that meets the requirements of undefined rights attributes and openness, and the area is an area that meets the water interface buffer zone.

[0097] Based on POI, street view, social media check-in, and LBS data, a second set of basic feature data is obtained, including check-in point popularity values, street view element area ratios and numbers, and POI types and locations related to public vitality. The second set of basic feature data can contain multiple types of public features. Correspondingly, when input into the extreme gradient boosting algorithm, each road network node should have a reachability feature value corresponding to each type of public feature, so that the algorithm model can combine multi-dimensional data to achieve more reliable learning and recognition results during training and recognition.

[0098] The acquisition of the above-mentioned basic characteristic data targets several different informal urban space types, including waterfront spaces, corner green spaces, and mixed-use blocks. The approach is the same: first, all plots within the study area with defined ownership are removed to characterize their "undefined ownership" basic characteristic. Then, existing water system and road network data, such as Shapefile files, are used to further process the water system and road alignment data to characterize the unique basic characteristics of "waterfront," "corner green spaces," and "mixed-use blocks." Specific data processing methods differ, including the following:

[0099] For cases where the informal urban space type to be identified is waterfront space, the acquisition of the area data of the area to be identified includes subtracting buildings, water bodies and all AOIs with actual use functions from the area of ​​the study area;

[0100] For cases where the informal urban space type to be identified is a street corner green space, the acquisition of the area data of the area to be identified includes subtracting the building and street areas from the street area data of the study area;

[0101] For cases where the informal urban space type to be identified is a complex block, the acquisition of the area data of the area to be identified includes, based on the linear data describing the streets in the study area, creating a buffer zone for each street, and subtracting buildings and areas with defined ownership from the buffer zone.

[0102] In addition to the methods mentioned above, mixed-use blocks typically lack clearly defined land use rights or boundaries and may include temporary markets, roadside vendors, etc. Their geographic scope can also be determined based on social media data and street view images. Corner green spaces usually lack official planning and are maintained spontaneously by the community. They are characterized by their location at the intersection of buildings and streets, and their geographic scope can also be determined based on satellite imagery, building data, and street view maps.

[0103] The specific methods for processing multi-source data to obtain the second basic feature data can be found in existing technologies and will not be elaborated further. The first basic feature data can be obtained using existing tools in the ArcGIS platform, such as "erasure" and "segmentation." The second basic feature data is "common" feature data, stored as a point set Shapefile. Subsequently, programs such as Python can be used to distribute the point set label data to the nearest virtual road network nodes based on the shortest path algorithm.

[0104] After obtaining the first basic feature data, the area data of the informal urban space to be identified can be further obtained based on the area data of the "unclaimed" area and the line data of water systems, road networks, etc. For example, if the category of the informal urban public space to be identified is waterfront space, the method for obtaining the area data of the informal urban space to be identified based on the first basic feature data includes:

[0105] Based on the water system Shapefile, water area Shapefile, and area data Shapefile of the region to be identified from the first basic feature data, outer buffers of a set width are created for the water system and water area within the area to be identified. The original water area within the area to be identified is removed, and the buffers for water area and water system are merged to eliminate the overlap between different buffers. Based on all the merged buffer area data and the set coordinate reference system (CRS), a GeoDataFrame file of the informal urban space to be identified, i.e., the urban waterfront area without defined ownership, is generated.

[0106] The above describes the method for obtaining the corresponding unclaimed area data when the informal urban space to be identified is a waterfront space. For cases where the informal urban space to be identified is a mixed-use block or corner green space, the area data of the area to be identified can be directly used to construct a GeoDataFrame file of the corresponding type of informal urban space area, making it suitable for processing and analysis using Python programs.

[0107] III. Construction of Virtual Road Network Structure with Equidistant Nodes

[0108] After obtaining the area data of the informal urban spaces to be identified, the informal urban spaces are converted into an equidistant nodal graph structure virtual road network, including:

[0109] Based on the area data of the informal urban spaces to be identified, define the center point and boundaries of the research scope;

[0110] Within the study area, a virtual road network structure (i.e., a fishing net) including equidistant nodes, horizontal line segments, and vertical line segments is generated according to a set grid radius. Using the concept of "representing surfaces with points," the area study object is decomposed into a set of points with latitude and longitude coordinates that can be used for further calculations. Corresponding GeoDataFrame files are created, and the nodes and line segments of the virtual road network are saved as Shapefile files within the GeoDataFrame for easy access in subsequent calculations.

[0111] IV. Shortest Path Algorithm: Bidirectional Allocation

[0112] In this section, this embodiment associates the public feature label in the second basic feature with the nodes in the virtual road network to characterize the publicness of each node within the study area, which is used for subsequent accessibility analysis, and then to determine whether each point within the study area is within the informal urban space to be identified.

[0113] Specifically, based on the second basic feature and the equidistant node graph structure virtual road network, the shortest path method is used to assign the second basic feature to associated nodes, resulting in equidistant node graph structure data containing common open features, including:

[0114] The feature labels of each feature point in the feature point set corresponding to each "common" type in the second basic feature are used to draw perpendicular lines from the feature point to the adjacent line segments based on the feature point coordinates and the node coordinates and line segments in the equidistant node graph structure virtual road network. The line segment connected by the shortest perpendicular line is selected, and the corresponding feature label is assigned to the virtual road network node associated with both ends of the line segment. The association information between the virtual road network node and the corresponding common feature label is recorded at the corresponding node in the virtual road network node Shapefile file and at the corresponding feature point in the common feature point set Shapefile file. In this way, the equidistant node graph structure containing common feature labels within the research scope can be obtained.

[0115] The above technical solution realizes the bidirectional allocation of the shortest path between public feature labels and virtual road network nodes, providing a data foundation for subsequent objective weighting algorithms. The public hot certificate labels mentioned above include, as mentioned earlier, the heat value of check-in points, the area ratio and number of street view elements, and POI types related to public vitality. Among them, POI types include, but are not limited to, ["shopping", "transportation facilities", "food", "life services", "leisure and entertainment", "education and training", "leisure and entertainment", "culture and media", "green space", "government agencies"].

[0116] V. Accessibility Analysis

[0117] Using the aforementioned equidistant node graph structure file containing common feature labels within the research scope as the basis for accessibility analysis, the automated analysis program implemented in Python takes the line segment Shapefile file and the node Shapefile file recording the associated common feature labels in the equidistant node graph structure GeoDataFrame file as input data. First, the coordinate reference system is normalized, then the road network is converted into a multimap, and then accessibility calculation is performed, as detailed below.

[0118] 5.1) For each node, calculate its reachability score to nodes with different labels under each common feature type, using the following formula:

[0119]

[0120] In the formula, For the node The reachability score to a certain type of feature label node under a certain common feature type. This indicates the feature labels assigned to this type in the equidistant node graph structure. The first node in the nth node The nth node, and the nth Each node reaches another node along the road network. Actual walking distance Within the set walking tolerance distance.

[0121] 5.2) Based on the reachability scores of each node under different feature labels corresponding to each common feature type, obtain the comprehensive reachability score of each node under each common feature type, including:

[0122] 5.2.1) For each common feature type, the reachability scores of each node under different feature labels are normalized using the following formula:

[0123]

[0124] In the formula, For the node Assigned to the equidistant node graph structure The reachability score of all nodes for the class feature label. This represents the number of nodes assigned to each node in the equidistant node graph structure. The set of reachability scores for nodes with class feature labels. and These are the maximum and minimum values ​​in the set, respectively. To The normalized value;

[0125] At this point, for each node, we can obtain the reachability score normalized dataset for all labels under each common feature type. Integrating all N nodes in the equidistant node graph structure, we can obtain the following reachability score normalized dataset matrix for each common feature type:

[0126]

[0127] 5.2.2) For each feature label under each common feature type, calculate the weight value of each node corresponding to each feature label using the CRITIC weighting method based on the node reachability score normalized dataset. Specifically:

[0128] The dataset is normalized based on the reachability scores of all nodes corresponding to each feature label class. Calculate the mean respectively and standard deviation The formula is:

[0129] ,

[0130] Calculate the labels of each feature class and all features under the same common feature type. The formula for determining the degree of contradiction in feature labels is:

[0131]

[0132] In the formula, For the first The contradictory values ​​between the class feature label and all feature labels under the common feature type; To utilize the Pearson correlation coefficient method, based on the normalized dataset of reachability scores corresponding to a single common feature type for all nodes, the 1st... Class feature labels and common feature types The correlation coefficient of class feature labels is calculated using the following formula:

[0133]

[0134] In the above formula, This represents the average of the normalized reachability scores of all nodes corresponding to the Kth class feature label, and its calculation principle is the same as... ;

[0135] Calculate the first Information carrying capacity of class feature labels The formula is: ;

[0136] Calculate the first All class feature labels under the same common feature type Normalized weights in feature labels The formula is:

[0137]

[0138] 5.2.3) Based on the weight values ​​of various feature labels corresponding to each node and the normalized values ​​of the reachability scores, the formula for calculating the comprehensive reachability score of each node for each common feature type is as follows:

[0139]

[0140] In the formula, For nodes The comprehensive accessibility score for a specific common feature type.

[0141] At this point, for each node, a set of <node number, latitude, longitude, first common feature type comprehensive accessibility score, second common feature type comprehensive accessibility score...> can be obtained. The set of all node arrays is the input to the pre-trained extreme gradient boosting algorithm model. Then, the extreme gradient boosting algorithm model can be used to efficiently identify whether each node is within the informal urban space, thereby clarifying the range and boundary identification of the informal urban space of the specified type to be identified within the research area.

[0142] Subsequently, the recognition results can be imported into ArcGIS Pro for visualization. After importing, the platform uses different diameters and colors to visualize points identified as "1" (non-urban public space) and "0" (urban public space). Nodes identified as "1" are drawn with a darker color and a larger diameter. The area covered by a cluster of darker-colored coordinate points on the map represents the informal urban public space identification results, such as... Figure 8 and Figure 9 As shown. Furthermore, expertise in digital landscape and digital urban design research can be used to further examine the identification results and verify their accuracy.

[0143] Example 2

[0144] This embodiment uses a square area with a side length of 5 kilometers centered on the Sipailou Campus of Southeast University in Nanjing City, Jiangsu Province, as the study area to identify the scope and boundaries of the waterfront space. The specific implementation method is explained below.

[0145] Step S1: Determine the specific type of the area to be identified, which in this embodiment is waterfront space. Utilizing multi-source big data of different types, through data preprocessing and integration, the basic characteristics of "undefined ownership" and the unique basic characteristics of "waterfront" are characterized, constructing a Shapefile for the area to be identified to determine the type of informal public space.

[0146] Preliminary data cleaning was conducted based on the fundamental characteristics of social media check-ins, street view images, and POI data. From the social media check-in data, latitude and longitude coordinates and check-in point types were extracted, and popularity values ​​were calculated according to the platform's popularity ranking rules. From the street view data, image semantic segmentation was used to calculate the area ratio and number of different types of elements (such as street trees, pedestrians, non-motorized vehicles, and shop signs). From the POI data, types and latitude and longitude coordinates related to public vitality were extracted, and "publicness" characteristic data were obtained through analysis.

[0147] As input data for subsequent analysis procedures, the main program first needs to import the Shapefile files of water systems (linear features), water bodies (area features), and the Shapefile of the area to be identified (area features) obtained from the OpenStreetMap platform into the buffer generation program, setting the buffer width to 100 meters. For example... Figure 2 The buffer zone generation program first removes the waterways within the water area, then extends the waterways not currently in the water area 100 meters away from the center, and finally erases the original water area. The union of the buffer zones generated by the waterways and the water area is then exported as a "Unclaimed Urban Waterfront Area" Shapefile file. The resulting area is as follows: Figure 3 and Figure 4 As shown.

[0148] Shapefile is a geographic information system (GIS) data format developed by ESRI (Environmental Systems Research Institute) in the United States. It is used to store geospatial data and attribute information. It is a non-topological, simple format primarily used to store vector data such as points, lines, and polygons, and typically consists of the following files:

[0149] .shp files: store the geometric shapes of geographic data, such as the coordinate information of points, lines, and polygons;

[0150] .shx files: These are index files for .shp files, used to speed up random access to data within .shp files;

[0151] .dbf file: is a dBase format database file that stores the attribute information of geographic data, such as name and type;

[0152] .prj files contain projection information for geographic data and describe the geographic coordinate system used for the data;

[0153] Other optional files include .sbn and .sbx files (spatial index information), and .xml files (metadata).

[0154] Step S2: Within the "unclaimed urban waterfront area" (i.e., the area to be identified), determine the latitude and longitude of the center coordinate point of the study area and the radius of the fishing net extension. Construct fishing nets at certain intervals as a virtual road network and generate a graph structure with equidistant nodes. The purpose here is firstly to associate the "point set data" and "area data" so as to import them into the subsequent accessibility analysis program to obtain the accessibility of common feature points within the walking tolerance distance along the road network; secondly, to ensure that the horizontal and vertical road networks are interconnected and cover the entire study area.

[0155] In the program that generates the graph structure with equidistant nodes, the center point of the study area is first set, the road network spacing is selected as 10 meters, and after coordinate transformation, the bounding box is calculated to generate the equidistant nodes, horizontal line segments, and vertical line segments of the virtual road network. Then, the program creates a GeoDataFrame and saves the line segments and nodes as different Shapefile files for easy import into subsequent programs.

[0156] Step S3: Based on the public nature of informal urban public spaces, select the public nature feature types that have a significant impact on the identification of informal urban public spaces. This embodiment only uses POI features as an example for illustration. The POI types participating in the subsequent calculation include ["Shopping", "Transportation Facilities", "Food", "Life Services", "Leisure and Entertainment", "Education and Training", "Culture and Media", "Green Space", "Government Agencies"].

[0157] Based on the analysis of the "publicity" feature data obtained in step S1, the shortest path algorithm is used to assign the feature point labels, i.e. POI types, in the publicity feature data corresponding to the selected POI type to the two nearest virtual road network nodes, thus obtaining the equidistant node graph structure containing POI type labels in the area to be identified.

[0158] Step S4 involves performing reachability analysis on the equidistant node graph structure containing POI data within the area to be identified. The program automatically performs coordinate reference system normalization, uniformly setting all input Shapefiles to CRS = EPSG:32650. The Python program first segments the surface region of the informal urban space to be identified according to the study area. Then, the filtered road network is converted into a multimap and edges are added.

[0159] During the accessibility analysis, the accessibility of reaching a specified POI type node from each road network node within the study area along the virtual road network within a walking tolerance distance of 400m is calculated.

[0160] After the calculation, the program stores the node data in a CSV file and exports it in the format of <node number, latitude, longitude, calculation result of the "shopping" public opening score, calculation result of the "transportation facilities" public opening score, …>, for reference Figure 5 as shown. This file, as Intermediate Product 1, is continuously input into the program.

[0161] Use the CRITIC weight method to objectively assign weights to each category of public characteristics (such as POI types, social media check-in subtypes), and calculate the comprehensive accessibility score of each road network node to different public characteristic points based on the weights. The program first performs normalization processing and then uses the CRITIC weight method for analysis. Export a CSV file in the format of <POI type, weight>, such as Figure 6 , as Intermediate Product 2.

[0162] After that, each road network point calculates the final comprehensive node accessibility score based on Intermediate Product 1 and Intermediate Product 2. Then, the program stores the nodes in a CSV file and exports them in the format of <node number, latitude, longitude, comprehensive accessibility score of public characteristic 1, comprehensive accessibility score of public characteristic 2, …>, for reference Figure 7 .

[0163] Step S5: Import the obtained scored result CSV file into the pre-trained Extreme Gradient Boosting algorithm program, and the program will give the recognition result of the informal urban public space. Import the recognition result into ArcGIS Pro, and use markers with different diameters and different color mappings for visualization of the points recognized as 1 (non-urban public space) and 0 (urban public space). For the nodes recognized as "1", their plotted colors are darker and their diameters are larger. The area covered by the coordinate points marked with darker colors concentrated and contiguous on the map is the recognition result of the waterfront space type in the informal urban public space within the study area. Such as Figure 7 is the overall map of the recognition result in ArcGIS Pro, Figure 8 is the partial map of the recognition result in ArcGIS Pro.

[0164] Step S6: For the recognition result of the urban public space in the area to be recognized obtained, use digital landscape and digital urban design expertise, combined with satellite maps and statistical analysis methods for further verification to verify the accuracy of the result.

[0165] Embodiment 3

[0166] Based on the same inventive concept as Embodiments 1 to 3, this embodiment introduces a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the method for identifying informal urban public spaces based on multi-source big data as described in Embodiment 1 or 2.

[0167] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0168] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0169] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0170] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0171] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for identifying informal urban public spaces based on multi-source big data, characterized in that, include: Obtain multi-source data on the study area where the informal urban public spaces to be identified are located, as well as information on the types of informal urban spaces to be identified; The multi-source data is processed to obtain the first basic feature and the second basic feature data of the informal urban space to be identified. The first basic feature corresponds to the basic characteristics of the undefined ownership characteristics and the type of informal urban space to be identified, and the second basic feature corresponds to the public nature of the informal urban space to be identified. Based on the first basic feature, obtain the area data of the informal urban space to be identified; Based on the area data, the informal urban space to be identified is converted into an equidistant node graph structure virtual road network; Based on the second basic feature and the equidistant node graph structure virtual road network, the second basic feature is assigned to the associated nodes using the shortest path method to obtain equidistant node graph structure data containing common open features; Based on the equidistant node graph structure data containing public and open features, accessibility analysis is performed on each node to obtain a feature dataset including the coordinates and accessibility scores of each node in the equidistant node graph structure. The feature dataset is input into a pre-trained extreme gradient boosting algorithm model to obtain point set data of informal urban spaces to be identified; The reachability analysis of each node is performed based on the equidistant node graph structure data containing public open features, including: For each node, calculate its reachability score to nodes with different labels under each common feature type, using the following formula: In the formula, For the node The reachability score to a certain type of label node under a certain common feature type. This indicates the feature labels assigned to this type in the equidistant node graph structure. The first node in the nth node The nth node, and the nth Each node reaches another node along the road network. Actual walking distance Within the set walking tolerance distance; Based on the reachability scores of each node under different labels for each common feature type, a comprehensive reachability score for each node under each common feature type is obtained, including: For each common feature type, the reachability scores of each node under different labels are normalized using the following formula: In the formula, For the node Assigned to the equidistant node graph structure The reachability score of all nodes for the class feature label. This represents the number of nodes assigned to each node in the equidistant node graph structure. The set of reachability scores for nodes with class feature labels. and These are the maximum and minimum values ​​in the set, respectively. To The normalized value; For each feature label under each common feature type, the weight values ​​of each node corresponding to each feature label are calculated using the CRITIC weighting method based on the node reachability score normalized dataset. Based on the weight values ​​of various feature labels corresponding to each node and the normalized value of the accessibility score, calculate the comprehensive accessibility score of each node for each common feature type.

2. The method according to claim 1, characterized in that, The multi-source data includes: social media check-in data and map data. The map data includes at least a variety of data such as POI, AOI, street view, road network, water system, water area, building outline, and LBS data. The multi-source data is processed to obtain the first and second basic feature data of the informal urban space to be identified, including: Based on road network, water system, water area, building outline, and AOI data, surface data of the areas to be identified with "openness" and "undefined weight attributes" are obtained from the surface regions of the study area and used as the first basic feature data; Based on POI, street view, social media check-in, and LBS data, we analyzed and obtained second basic feature data, including check-in point popularity value, street view element area ratio and number, and / or, POI type and location related to public vitality.

3. The method according to claim 2, characterized in that, The process of obtaining surface data of the region to be identified with "openness" and "undefined weight attributes" from the surface regions of the study area includes: For cases where the informal urban space type to be identified is waterfront space, buildings, water bodies, and all AOIs with actual use functions are subtracted from the area of ​​the study area, and the area data of the area to be identified is obtained based on the remaining area. For cases where the informal urban space type to be identified is a street corner green space, the building and street areas are subtracted from the street area data of the study area, and the area data of the area to be identified is obtained based on the remaining area. For cases where the informal urban space type to be identified is a complex block, a buffer zone is created for each street based on the linear data describing the streets within the study area. Buildings and areas with defined ownership are subtracted from the buffer zone, and the area data of the area to be identified is obtained based on the remaining area.

4. The method according to claim 2 or 3, characterized in that, The step of obtaining the area data of the informal urban space to be identified based on the first basic feature data includes: If the category of informal urban public space to be identified is waterfront space, then based on the water system Shapefile, water area Shapefile, and area data Shapefile of the area to be identified in the first basic feature data, an outer buffer of a set width is created for the water system and water area within the area to be identified; the original water area within the area to be identified is removed, and the buffers for water area and water system are merged to eliminate the overlapping parts between different buffers; based on all the merged buffer area data and the set coordinate reference system (CRS), a GeoDataFrame file of the informal urban space to be identified, i.e., the urban waterfront area without defined ownership, is generated.

5. The method according to claim 1, characterized in that, Based on the area data, the informal urban space to be identified is transformed into an equidistant node graph structure virtual road network, including: Based on the area data of the informal urban spaces to be identified, define the center point and boundaries of the research scope; Within the study area, a virtual road network including nodes, horizontal line segments, and vertical line segments is generated according to the set grid radius. A corresponding GeoDataFrame file is created, and the nodes and line segments of the virtual road network are saved as Shapefile files respectively.

6. The method according to claim 1, characterized in that, Based on the second basic feature and the equidistant node graph structure virtual road network, the second basic feature is assigned to associated nodes using the shortest path method, resulting in equidistant node graph structure data containing common open features, including: The feature labels of each feature point in the feature point set corresponding to each "commonality" type in the second basic feature are used to draw perpendicular lines from the feature point to the adjacent line segments based on the coordinates of the feature point and the coordinates of the nodes and line segments in the equidistant node graph structure virtual road network. The line segment connected by the shortest perpendicular line is selected, and the corresponding feature label is assigned to the virtual road network nodes associated with both ends of the line segment. The association information between the virtual road network node and the corresponding commonality feature label is recorded at the corresponding node in the virtual road network node Shapefile file and at the corresponding feature point in the point set Shapefile file.

7. The method according to claim 1, characterized in that, The calculation of the weight values ​​of each node corresponding to various feature labels using the CRITIC weighting method includes: The dataset is normalized based on the reachability scores of all nodes corresponding to each feature label class. Calculate the mean respectively and standard deviation The formula is: ， Calculate the labels of each feature class and all features under the same common feature type. The formula for determining the degree of contradiction in feature labels is: In the formula, For the first The contradictory values ​​between the class feature label and all feature labels under the common feature type; The first value was obtained using the Pearson correlation coefficient method. Class feature labels and common feature types The correlation coefficient of class feature labels is calculated using the following formula: In the above formula, This represents the average of the normalized reachability scores of all nodes corresponding to the Kth class feature label; Calculate the first Information carrying capacity of class feature labels The formula is: ; Calculate the first All class feature labels under the same common feature type Normalized weights in feature labels The formula is: The formula for calculating the comprehensive reachability score of each node for each common feature type is as follows: In the formula, For nodes The comprehensive accessibility score for a specific common feature type.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the informal urban public space identification method based on multi-source big data as described in any one of claims 1-7.

Images