A method and system for urban land use recommendation based on multi-scale spatial colocation patterns

Abstract

The present invention provides a method and system for urban land use recommendation based on multi-scale spatial co-location patterns. The method comprises the following steps: S1, resampling of POI points based on a multi-scale grid within the city; S2, setting land use types based on POI categories; S3, second-order spatial co-location pattern analysis of land use types; S4, land use type recommendation for specific locations. The system comprises: a parameter setting module, a POI data input module, a POI data processing module, a spatial co-location pattern recognition module, and a land use type recommendation module. The present invention considers the land use association characteristics of multi-scale spatial co-location patterns, mines the land use association characteristics in cities of different sizes and different spatial co-location patterns, and recommends land use types under the influence of neighboring land use based on the spatial co-location pattern characteristics of land use, so that the determination of land use types can achieve quantitative, automatic and scientific recommendation.

Description

Technical Field

[0001] The present invention relates to the field of spatial big data analysis, and in particular to a method and system for recommending urban land use based on multi-scale spatial colocation patterns. Background Art

[0002] Urban renewal is the necessary, planned reconstruction of urban areas that are no longer suitable for modern urban life. In most urban renewal projects, the integration and reallocation of various land uses is a key element and core component. Scientific and prudent planning and design of various land uses, from a compliant and pragmatic perspective, can facilitate the smooth implementation of urban renewal projects and truly achieve the goal of promoting high-quality development in the region.

[0003] Spatial co-location pattern analysis is a common method in urban data mining, effectively mining the spatial association characteristics of spatial elements. From the perspective of spatial scope, it can be divided into global spatial association pattern mining and local spatial association pattern mining. Global spatial association pattern mining focuses on the overall significant characteristics of a region. It was first proposed by Shekhar et al. and can be divided into frequent neighborhood-based mining methods and density-based mining methods. Local spatial association pattern mining focuses more on the significant spatial co-location patterns of local areas and mines spatial association characteristics through partitioning. The main partitioning methods include quadtree partitioning, K-nearest neighbor graph partitioning, clustering-based partitioning methods, and visualization-based local spatial co-location pattern mining methods. Urban land is a typical spatial element, and the distribution patterns between urban land types can be mined based on spatial co-location pattern analysis techniques. However, urban land distribution patterns vary at different scales. Both global and local spatial co-location pattern mining methods fail to take into account the differences in the impact range and distribution patterns of urban land at different scales. In addition, there is a lack of a mature methodological system for quantitative recommendations of urban land types for specific locations, especially those that consider the current status of neighboring urban land, which makes it difficult to meet the actual application needs of urban renewal. Summary of the Invention

[0004] In order to overcome the shortcomings of the existing technology, the present invention targets the quantitative requirements of land use types in urban planning, construction, and renewal, constructs multi-scale spatial co-location patterns of urban land and identifies spatial correlation characteristics of land use, and proposes an urban land recommendation method and system based on multi-scale spatial co-location patterns.

[0005] The present invention provides an urban land recommendation method based on multi-scale spatial colocation patterns, comprising the following steps:

[0006] S1. POI point resampling based on a city-wide multi-scale grid, including:

[0007] S11, city boundary extraction;

[0008] S12, construct multi-scale grid;

[0009] Set the grid size to K = {k1, k2, ..., k n …,k N}, where N represents the total number of grid scales, k n Indicates the nth grid scale value, and constructs the city-wide grid according to the set grid scale parameter K;

[0010] S13, resampling the POI points;

[0011] For a grid size of k n The POI data in a single grid is extracted based on the spatial location and resampled according to the POI data category. The resampling is performed by calculating the geometric center point of the grid. After resampling, the POI points of different categories in each grid are represented by the geometric center of the corresponding category.

[0012] S2. Setting land use type based on POI category;

[0013] The resampled POI data are matched according to the conversion relationship between POI categories and land use types. The geometric centers of POIs of different categories in each grid become the instance points of the land use type in each grid.

[0014] S3, second-order spatial co-location pattern analysis of land use types;

[0015] Spatial colocation pattern analysis of land use types is used to mine association rules between land use types. Different land use type instances are connected to obtain a set of spatially adjacent instances of different land use type instances as a set of spatial colocation candidate instances. Spatial colocation candidate instances to be mined are screened based on the screening threshold. Finally, based on the spatial association characteristics of land use in spatial colocation patterns, a second-order spatial colocation pattern correlation PI table of pairwise land use types at all scales is obtained and saved.

[0016] S4. Recommended land use types for specific locations, including:

[0017] S41. Setting the urban land location: setting the proposed land location (X0, Y0);

[0018] S42, setting the spatial neighborhood calculation scale, setting the scale parameter k0, k0∈K;

[0019] S43. Determine the type of land use in the neighborhood;

[0020] S44, land use type recommendation;

[0021] First, calculate the cumulative correlation value of the second-order spatial co-location pattern of the instance point land use type x within the neighborhood range The specific calculation formula is:

[0022]

[0023] Where, Indicates the yth instance point of land use type x in the neighborhood and the instance point L of the land use location to be recommended p The spatial co-location pattern PI value of In, L xy The land use type is x, L p The land use type is p, x∈I, p∈I′, R is the total land use type, a xy Indicates whether to retain the spatial parity pattern;

[0024] Calculate the associated values of all land use types within the neighborhood in sequence Will Sort from high to low, extract by sort The land use type p in the table is used to obtain the land use type recommendation table for the proposed land use location.

[0025] Preferably, the resampling in step S13 is performed by calculating the geometric center point of the grid, specifically:

[0026] The coordinates of a POI category L in grid j after resampling (X L (j), Y L (j)) is the geometric center of all POI points of POI category L in the grid j, expressed as:

[0027]

[0028] Among them, (X L,i (j), Y L,i (j)) is the coordinate of the i-th POI point with POI category L in the grid j, n L (j) represents the number of POI points of POI category L in grid j;

[0029] POI points are resampled for all city-wide grids of grid scale values in grid scale K. After resampling, POI points of different categories in each grid are represented by the geometric center of the corresponding category.

[0030] Preferably, the second-order spatial co-location pattern analysis of land use types in step S3 comprises the following specific steps:

[0031] S31. Select two land use types and a grid scale value;

[0032] S32, generating a set of spatially adjacent instances;

[0033] Spatial neighbor instance set {A s (A s , B t , d)} represents each instance point B of land use type B t With core point A s Connect the pairs and calculate (A s , B t ) The distance d between two points; the set of spatial neighboring instances {B t (A s , B t , d)} represents each instance point A of land use type A s With core point B t Connect the pairs and calculate (A s , B t ) the distance d between two points;

[0034] S33, determining a set of spatially co-located candidate instances;

[0035] Find the set {A s (A s , B t , d)} and the set {B t (A s , B t , d)}, and obtain the spatial co-location candidate instance set {A s , B t , d};

[0036] S34, determining spatial co-location patterns;

[0037] The spatial co-location candidate instance set is subjected to long-distance noise removal according to the threshold value. Only the spatial co-location candidate instances whose distance between two points is within the threshold value can be retained, and the spatial co-location candidate instances to be mined are screened. s , B t , d|d<d ave}, where d ave is the screening threshold;

[0038] S35. Identification of spatial correlation features of land use based on spatial colocation patterns;

[0039] The participation index of the spatial colocation pattern is used as the characteristic value of the land use spatial association characteristics. The definition formula of the participation index is:

[0040] PI(cp (A，B) )=min{Pr(cp (A，B) , YD A ), Pr(cp (A，B) , YDB )}

[0041] Pr(cp (A，B) , YD A )=N(cp (A，B) , YD A ) / N(YD A )

[0042] Pr(cp (A，B) , YD B )=N(cp (A，B) , YD B ) / N(YD B )

[0043] Where: PI(cp (A,B) ) represents the spatial co-location pattern participation index of the two land use types A and B; Pr(cp (A，B) , YD A ) represents the participation rate of land use A in the spatial co-location pattern of two land use types A and B, N(cp (A，B) , YD A ) represents the candidate instance of the same location to be mined {A s , B t , d|d<d ave The number of instances of land use type A in}, N(YD A ) represents the total number of instance points of land use type A; similarly, Pr(cp (A,B) , YD B ) represents the participation rate of land use B in the spatial co-location pattern of two land use types A and B, N(cp (A，B) , YD B ) represents the candidate instance of the same location to be mined {A s , B t , d|d<d ave The number of instances of land use type B in}, N(YD B ) represents the total number of instance points with land use type B.

[0044] Preferably, the screening threshold d in step S34 ave is a set of spatially adjacent instances {A s , B t , the average distance of d}.

[0045] Preferably, the step S43 determines the type of neighborhood land use, and the specific steps are as follows:

[0046] S431. Based on the set scale parameter k0, with the land location (X0, Y0) as the center and the scale parameter k0 as the radius, a neighborhood range of the land location is delineated in the city range grid with the grid scale value k0;

[0047] S432. Obtain all instance points of all land use types within the neighborhood through spatial connection, where the yth instance point of land use type x is represented by L xy ;

[0048] S433, calculate each instance point L within the neighborhood xy Distance d from the location of the recommended land (X0, Y0) xy .

[0049] The present invention also discloses an urban land use recommendation system based on multi-scale spatial co-location patterns, which includes the following modules: a parameter setting module, a POI data input module, a POI data processing module, a spatial co-location pattern recognition module, and a land use type recommendation module, specifically:

[0050] A parameter setting module is used to set the required parameters, including a city setting unit, a grid scale parameter setting unit, a location setting unit for recommended land, and a recommended land scale setting unit; wherein the city setting unit is used to set the name of a prefecture-level city or county-level city; the scale parameter setting unit is used to set the scale for participating in the spatial co-location pattern calculation; the location setting unit for recommended land is used to set the latitude and longitude coordinates of the location of the recommended land; and the recommended land grid scale setting unit is used to set the scale for calculating the recommended spatial neighborhood.

[0051] The POI data input module is used to select a city POI data file according to the city set in the parameter setting module and import the city POI data file into the system;

[0052] The POI data processing module is used for extracting the city range based on POI data, resampling the POI data at multiple scales, and setting the land use type of the POI data; specifically, it includes a city range delineation unit, a POI resampling unit, and a land use type setting unit; the city range delineation unit is used to automatically delineate the city range based on the kernel density analysis results of the POI data; the POI resampling unit is used to generate grid data based on the set grid scale and POI data, and resample the POI data at each scale according to the grid geometric center point; the land use type setting unit is used to match the POI category to the corresponding land use type;

[0053] The second-order spatial co-location pattern analysis module is used to obtain the second-order spatial co-location pattern correlation PI table; it includes: a spatial instance collection submodule and a land use spatial correlation feature identification submodule;

[0054] The spatial neighboring instance collection submodule is used to obtain and save various spatial instances;

[0055] The spatial correlation feature recognition submodule is used to traverse all combinations of land use types and grid scales, obtain PI values and save them;

[0056] The land use type recommendation module is used to output a land use type recommendation list, including: a neighborhood land use extraction submodule and a land use recommendation submodule.

[0057] The neighborhood land extraction submodule is used to extract the neighborhood land types within the neighborhood of the recommended land location at a specified grid scale; based on the recommended land location (X0, Y0) and scale parameters, all instance points within the neighborhood are obtained through spatial connection, and the distance d between all instance points within the neighborhood and the recommended land location (X0, Y0) is calculated. xy ;

[0058] The land use recommendation submodule is used to recommend the land use type at (X0, Y0). First, the second-order spatial co-location pattern cumulative association value of the instance point with land use type x in the neighborhood is calculated. Then Sort from high to low, and extract the land use types that belong to the I′ set as the final recommended land use type for the location.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] (1) Considering the analysis of urban land use correlation characteristics in multi-scale spatial colocation patterns, the urban land use correlation characteristics in cities of different sizes and different spatial colocation patterns are explored.

[0061] (2) An urban land use recommendation method based on spatial co-location pattern characteristics is used to quantitatively recommend urban land use types under the influence of neighboring land use. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flow chart of the urban land recommendation method based on multi-scale spatial colocation patterns of the present invention;

[0063] Figure 2 This is a diagram showing the extraction result of the Shizhu County area based on POI data according to an embodiment of the present invention;

[0064] Figure 3 This is a graph showing the resampling results of the 500m-scale grid and POI data in an embodiment of the present invention;

[0065] Figure 4 The first 10 spatial co-location pattern results of the second-order spatial co-location pattern correlation PI table at a scale of 500m output by an embodiment of the present invention;

[0066] Figure 5 This is a schematic diagram of the land use recommendation neighborhood and the top three recommendation results according to an embodiment of the present invention;

[0067] Figure 6 Schematic diagram of the urban land recommendation system based on multi-scale spatial colocation pattern of the present invention. DETAILED DESCRIPTION

[0068] In order to make the content of the invention and technical solutions of the present invention easier to understand and implement, the present invention will be described in more detail below with reference to the accompanying drawings and examples. It should be understood that for those skilled in the art, the present invention can be implemented without some of these specific details. The following description of the embodiments is merely intended to provide a better understanding of the present invention by illustrating examples of the present invention. Therefore, the specific embodiments described herein are intended only to explain the present invention and are not intended to limit the present invention.

[0069] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0070] According to the present invention, a method for recommending urban land use based on multi-scale spatial colocation patterns is provided. Figure 1 As shown, the method includes the following steps:

[0071] S1. POI (Point of Interest) resampling based on a city-wide multi-scale grid.

[0072] For cities that require land use recommendations, we first use POI data for multi-scale resampling preprocessing, which includes the following subdivision steps:

[0073] S11. Extract the city boundary.

[0074] Kernel density analysis based on POI data is used to define the area that meets the POI density distribution threshold as the urban area.

[0075] This example uses Shizhu County, Chongqing City, as an example to illustrate the multi-scale spatial colocation model land use recommendation method. First, the urban area of Shizhu County is extracted. Figure 2 It is based on the kernel density analysis of Shizhu County POI data, and the extracted urban area is approximately 10.55 square kilometers.

[0076] S12. Construct a multi-scale grid.

[0077] According to the multi-scale spatial co-location pattern analysis requirements, the required scale parameters are set and the grid scale is defined as K = {k1, k2, ..., k n ..., k N}, where N represents the total number of grid scales, k nRepresents the nth grid scale value, and then constructs the city-wide grid according to the set grid scale parameter K. Because there are N grid scale values, a total of N city-wide grids with different grid scales need to be constructed.

[0078] S13. Resample the POI points.

[0079] For a grid size of k n The POI data in a single grid is extracted according to the spatial location and resampled according to the POI data category. The POI data categories usually include commercial and residential, science and education and culture, financial institutions, government agencies and groups, cultural activities, catering services, shopping services, life services, sports and leisure services, medical care services, accommodation services, scenic spots, etc. These are the attributes of the POI data.

[0080] Resampling is performed by calculating the geometric center point of the grid. After resampling, the coordinates (X L (j), Y L (j)) is the geometric center of all POI points of POI category L in the grid j, which can be expressed as:

[0081]

[0082] Among them, (X L,i (j), Y L,i (j)) is the coordinate of the i-th POI point with POI category L in the grid j, n L (j) represents the number of POI points of POI category L in grid j. In particular, when n L When (j)=1, the resampled POI point is equal to the original POI point position.

[0083] POI points are resampled for all city-wide grids of grid scale values in grid scale K. After resampling, POI points of different categories in each grid are represented by the geometric center of the corresponding category.

[0084] In this embodiment, the grids in Shizhu County, Chongqing City are resampled as follows Figure 3 shown.

[0085] S2. Setting the land use type based on the POI category.

[0086] The resampled POI data are matched according to the conversion relationship between POI categories and land use types.

[0087] The relationship between POI categories and land use types is pre-set based on the actual POI data category and the selected land use type. For example, if the POI category is set to catering services, it will be associated with the land use type of catering land. If the POI data category is set to cultural activities and sports and leisure services, it will be associated with entertainment and recreation land.

[0088] For cities, there are many classifications of land use types, which are recorded as a set R, and the set R includes all land use types to be used subsequently. The embodiment of the present invention is based on the "Guidelines for the Classification of Land and Sea Use for National Land Space Survey, Planning, and Use Control (Trial)" (hereinafter referred to as the Land Classification Guidelines), and selects 22 land use types in the land classification guide, including wholesale market land, retail commercial land, catering land, other commercial service land, business and financial land, government and group land, industrial land, urban residential land, entertainment and sports land, hotel land, public utility business outlets land, education land, medical and health land, sanitation land, logistics and warehousing land, park green land, sports land, cultural activity land, square land, scientific research land, forest land, and social welfare land for the elderly. Users can select land types according to actual needs.

[0089] Once the POI categories are matched to the land use types, the geometric centers of the POIs of different categories in each grid become the instance points of the land use types in each grid; in effect, the corresponding land use types within a certain scale spatial range of the land use type instance points are obtained. It should be noted that the "certain scale spatial range" here does not refer to the grid range, but rather the range related to the subsequent algorithm. For example, in this application, due to the subsequent use of the Voronoi diagram, the certain scale spatial range actually refers to the Thiessen polygon.

[0090] S3. Second-order spatial colocalization pattern analysis of land use types.

[0091] Spatial colocation pattern analysis of land use types is mainly used to discover association rules between land use types. The present invention uses a Voronoi diagram (also known as Thiessen polygon) to connect different land use type instances to obtain a set of spatially adjacent instances of different land use type instances as a set of spatial colocation candidate instances. Spatial colocation patterns of land use types with strong spatial associations are then filtered based on a specific threshold. The specific steps are as follows:

[0092] S31. Select two land use types and a grid scale value. The present invention adopts second-order spatial co-location pattern analysis. The embodiment of the present invention takes a 500m grid scale value as an example to illustrate the second-order spatial co-location pattern analysis of land use types A and B.

[0093] S32: Generate a set of spatially adjacent instances.

[0094] First, the Voronoi diagram is constructed with the instance point representing land use type A as the core point. s Thiessen polygons with core points represent each instance point B of land use type B. t With core point A s Connect the pairs and calculate (A s , B t ) The distance d between two points, and the set of spatial neighboring instances {A s (A s , B t , d)}, s, t are positive integers. When s is taken as all instance points of land use type A, the spatial neighboring instance set {A s (A s , B t , d)} includes the combination and distance d between all instance points of land use type A and all instance points of land use type B.

[0095] Then generate a set of spatial neighboring instances of land use type B [B t (A s , B t , d)}, at this time with B t is the core point, representing each instance point A of land use type A s With core point B t Connect the pairs and calculate (A s , B jt ) The distance d between two points, when t is taken as all instance points of land use type B, the spatial adjacent instance set {B t (A s , B t , d)} includes the combinations and distances between all instance points of land use type B and all instance points of land use type A.

[0096] S33: Determine a set of spatially co-located candidate instances.

[0097] Find the set {A s (A s , B t , d)} and the set {B t (A s , B t , d)}, and obtain the spatial co-location candidate instance set {A s , B jt , d}. Spatial co-location candidate instance set {A s , B t , d} includes the combination of any instance point of land use type A and any instance point of land use type B and the distance between the two points.

[0098] S34. Determine the spatial co-location pattern.

[0099] The spatial co-location candidate instance set is subjected to long-distance noise removal based on the threshold. Only the spatial co-location candidate instances whose distance between two points is within the threshold are retained, and the spatial co-location candidate instances to be mined are screened. Calculate the spatial neighboring instance set {A s , B t ,d} the average distance d ave As the screening threshold, close instances with a distance less than the average distance are retained as spatial co-location candidate instances to be mined {A s , B t ,d|,d<d ave}.

[0100] S35. Identification of spatial correlation characteristics of land use based on spatial co-location patterns.

[0101] The participation index of the spatial colocation pattern is used as the characteristic value of the land use spatial association feature to measure the association between two land use types. The definition formula of the participation index is:

[0102] PI(cp (A，B) )=min{Pr(cp (A，B) , YD A ), Pr(cp (A，B) , YD B )}

[0103] Pr(cp (A，B) , YD A )=N(cp (A，B) , YD A ) / N(YD A )

[0104] Pr(cp (A，B) , YD B )=N(cp (A，B) , YD B ) / N(YD B )

[0105] Where: PI(cp (A，B) ) represents the spatial colocation pattern participation index of the two land use types A and B. The higher the value, the more frequently the spatial colocation pattern appears in the selected spatial range. (A,B) , YD A ) represents the participation rate of land use A in the spatial co-location pattern of two land use types A and B, N(cp (A，B) , YD A ) represents the candidate instance of the same location to be mined {A s , B t , d|d<d ave The number of instances of land use type A in}, N(YDA ) represents the total number of instance points of land use type A. Similarly, Pr(cp (A，B) , YD B ) represents the participation rate of land use B in the spatial co-location pattern of two land use types A and B, N(cp (A，B) , YD B ) represents the candidate instance of the same location to be mined {A s , B t , d|d<d ave The number of instances of land use type B in}, N(YD B ) represents the total number of instance points with land use type B.

[0106] For example, the instance points of land use type A are A1, A2, and A3, and the instance points of land use type B are B1, B2, B3, and B4. According to S31-S34, the candidate instances of spatial co-location to be mined are: {(A1, B2, 0.5), (A3, B2, 0.8), (A3, B1, 0.3), (A3, B4, 0.7)}, then:

[0107] Pr(cp (A，B) , YD A )=N(cp (A，B) , YD A ) / N(YD A )=2 / 3=0.67

[0108] Pr(cp (A，B) , YD B )=N(cp (A，B) , YD B ) / N(YD B )=3 / 4=0.75

[0109] PI(cp (A，B) )=min{Pr(cp (A，B) , YD A ), Pr(cp (A，B) , YD B )}=0.67.

[0110] S36. Traverse all combinations of pairwise land use types and all grid scales in sequence, and obtain and save the second-order spatial co-location pattern correlation PI table of pairwise land use types at all scales.

[0111] For example, the second-order spatial co-location pattern correlation PI at a spatial scale of 500m is shown as follows: Figure 4 shown.

[0112] S4. Recommendation of urban renewal land types for specific locations.

[0113] The land use type that is suitable for the current spatial location is recommended through the spatial co-location pattern under the spatial association of the neighborhood of the land use to be recommended.

[0114] S41, site location setting, set the proposed site location (X0, Y0), recorded as instance point L p It should be noted that the proposed land location (X0, Y0) must be within the city range extracted in step S11, otherwise a prompt will be given to reset the proposed land location.

[0115] S42: Setting the scale of the spatial neighborhood calculation, setting the scale parameter k0. The scale parameter k0 can only be selected from N grid scale values of the grid scale K and cannot be set arbitrarily.

[0116] S43. Determine the type of land use in the neighborhood. The specific steps are as follows:

[0117] S431. Based on the set scale parameter k0, with the land location (X0, Y0) as the center and the scale parameter k0 as the radius, a neighborhood range of the land location is delineated in the city range grid with a grid scale value of k0.

[0118] S432. Obtain all instance points of all land use types within the neighborhood through spatial connection, where the yth instance point of land use type x is represented by L xy , the land use types of all instance points within the neighborhood are represented by set I, and Y is the number of instance points of land use type x.

[0119] S433, calculate each instance point L within the neighborhood xy Distance d from the location of the recommended land (X0, Y0) xy .

[0120] S44. Land use type recommendation.

[0121] First, calculate the cumulative correlation value of the second-order spatial co-location pattern of instance points with land use type x within the neighborhood The specific calculation formula is:

[0122]

[0123] Where: Indicates the yth instance point of land use type x in the neighborhood and the instance point L of the land use location to be recommended p The spatial co-location pattern PI value of In the example, the land use type of Lxy is x, L p The land use type is p, x∈I, p∈I′, R is the total land use type, which is obtained by searching the PI table of second-order spatial co-location pattern correlation at the k0 scale. xyIndicates whether to retain the spatial co-location pattern, using the average distance as the screening threshold. If the distance d between the neighborhood instance point and the center of the recommended land location is xy Less than the average distance d of the second-order spatial colocalization pattern ave , then the second-order parity pattern is retained; otherwise, it is not retained. In particular, for land use type x, if Y = 1 and a xy =1,

[0124] Because the location instance point L p The land use type is not determined, so the method of traversing I′ is adopted. The total result is the total result obtained by combining all land use types I and land use types I′ in the neighborhood. Due to the limitation of grid size, the present invention will not have the situation where all land use types R are equal to land use types I in the neighborhood, so the situation where land use type I′ is empty is not considered.

[0125] Finally Sort from high to low, and extract the land use types that belong to the I' set in order, which is the final recommended land use type for the location. In the embodiment of the present invention, it is preferred to display the top ten land use types, and the final recommended list is as follows: Figure 5 shown.

[0126] In order to implement the urban land recommendation method based on multi-scale spatial co-location pattern, this embodiment also provides an urban land recommendation system based on multi-scale spatial co-location pattern, such as Figure 6 As shown, the system includes the following modules: parameter setting module 210, data input module 220, data processing module 230, spatial co-location pattern recognition module 240 and land use type recommendation module 250, specifically:

[0127] The parameter setting module 210 is used to set the required parameters, including a city setting unit, a grid scale parameter setting unit, a recommended land location setting unit, a recommended land scale setting unit, etc., wherein the city setting unit is used to set the name of a prefecture-level city or a county-level city; the scale parameter setting unit is used to set the scale involved in the spatial co-location pattern calculation, that is, the value in the grid scale K. The grid scale parameter needs to select a grid scale value, and the grid scale value that can be input is in meters. In this embodiment, the minimum value of the input grid scale value is not less than 50 meters, and the maximum value does not exceed 2000 meters; the recommended land location setting unit is used to set the latitude and longitude coordinates of the recommended land location; the recommended land grid scale setting unit is used to set the recommended spatial neighborhood calculation scale. The recommended spatial neighborhood calculation scale can only select the scale input by the grid scale parameter setting unit.

[0128] The POI data input module 220 is used to select a city POI data file based on the city set in the parameter setting module and import the city POI data file into this system; fuzzy matching or exact matching of the input city name and the city POI data file; the city POI data input file should include POI category attributes.

[0129] The POI data processing module 230 is used to extract the city range based on POI data, resample the POI data at multiple scales, and set the land use type of the POI data. Specifically, it includes a city range delineation unit, a POI resampling unit, and a land use type setting unit. The city range delineation unit is used to automatically delineate the city range based on the kernel density analysis results of the POI data, thereby implementing step S11. The POI resampling unit is used to generate grid data based on the set grid scale and POI data, and resample the POI data at each scale according to the grid geometric center point, thereby implementing step S13. The land use type setting unit is used to match the POI category to the corresponding land use type, thereby implementing step S2.

[0130] The second-order spatial co-location pattern analysis module 240 is used to obtain a second-order spatial co-location pattern correlation PI table; it includes: a spatial instance collection submodule and a land use spatial correlation feature identification submodule.

[0131] The spatial neighboring instance collection submodule is used to obtain and save various spatial instances. It first obtains a set of spatial neighboring instances, then a set of spatial co-located candidate instances, and finally a set of spatial co-located candidate instances to be mined. This submodule is used to implement steps S32-S34.

[0132] The spatial correlation feature identification submodule is used to obtain PI values for combinations of two land use types and one grid scale value. The two land use types are selected from the land use types in the POI data processing module 230, and the grid scale value is selected from the grid scale values set in the grid scale parameter setting unit. The spatial correlation feature identification submodule needs to traverse all combinations of two land use types and all grid scales to obtain and save the PI values. The spatial correlation feature identification submodule is used to implement steps S35-S36.

[0133] The land use type recommendation module 250 is used to output a land use type recommendation list. The land use type recommendation module 250 includes: a neighborhood land use extraction submodule and a land use recommendation submodule.

[0134] The neighborhood land use extraction submodule is used to extract the neighborhood land use types within the neighborhood of the recommended land use location at a specified grid scale; based on the recommended land use location (X0, Y0) and the scale parameter k0, all instance points within the neighborhood are obtained through spatial connection. These instance points are divided into I types of land use, and the distance d between all instance points within the neighborhood and the recommended land use location (X0, Y0) is calculated respectively. xy The neighborhood land extraction submodule is used to implement step S43.

[0135] The land use recommendation submodule is used to recommend the land use type at (X0, Y0). First, the correlation value between the neighboring land use type x and the land use type to be recommended p is calculated. where x∈I, p∈I′, I∪I′=R, R is all land use types, I is all neighboring land use types, I′ is the remaining land use types after removing all neighboring land use types; then the associated value Sort from high to low, and extract the land use type L that appears in the sort p , obtaining a recommended land use type table for the proposed land use location. Because I' does not include the neighboring land use type I, the final recommended land use type for this location must also be different from the I land use types within the neighboring area. The land use recommendation submodule is used to implement step S44.

[0136] Finally, it should be noted that the embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for urban land use recommendation based on multi-scale spatial colocation patterns, characterized by: It includes the following steps: S1. POI point resampling based on a city-wide multi-scale grid, including: S11, city boundary extraction; S12, construct multi-scale grid; Set the grid size to K = {k1, k2, ..., k n …,k N }, where N represents the total number of grid scales, k n Indicates the nth grid scale value, and constructs the city-wide grid according to the set grid scale parameter K; S13, resampling the POI points; For a grid size of k n The POI data in a single grid is extracted based on the spatial location and resampled according to the POI data category. The resampling is performed by calculating the geometric center point of the grid. After resampling, the POI points of different categories in each grid are represented by the geometric center of the corresponding category. S2. Setting land use type based on POI category; The resampled POI data are matched according to the conversion relationship between POI categories and land use types. The geometric centers of POIs of different categories in each grid become the instance points of the land use type in each grid. S3, second-order spatial co-location pattern analysis of land use types; Spatial colocation pattern analysis of land use types is used to mine association rules between land use types. Different land use type instances are connected to obtain a set of spatially adjacent instances of different land use type instances as a set of spatial colocation candidate instances. Spatial colocation candidate instances to be mined are screened based on the screening threshold. Finally, based on the spatial association characteristics of land use in spatial colocation patterns, a second-order spatial colocation pattern correlation PI table of pairwise land use types at all scales is obtained and saved. S4. Recommended land use types for specific locations, including: S41. Setting the location of urban land: setting the proposed location of the land (X0, Y0); S42, setting the scale of spatial neighborhood calculation, setting the scale parameter k0, k0∈K; S43. Determine the type of land use in the neighborhood; S44, land use type recommendation; First, calculate the cumulative correlation value of the second-order spatial co-location pattern of the instance point land use type x within the neighborhood range ), the specific calculation formula is: Where, Indicates the yth instance point of land use type x in the neighborhood and the instance point L of the land use location to be recommended p The spatial co-location pattern PI value of In, L xy The land use type is x, L p The land use type is p, x∈I, p∈I', R is the total land use type, a xy Indicates whether to retain the spatial parity pattern; Calculate the associated values of all land use types within the neighborhood in sequence Will Sort from high to low, extract by sort The land use type p in the table is used to obtain the land use type recommendation table for the proposed land use location.

2. The urban land use recommendation method based on multi-scale spatial colocation patterns according to claim 1, characterized in that: In step S13, resampling is performed by calculating the geometric center point of the grid, specifically: The coordinates of a POI category L in grid j after resampling (X L (j),Y L (j)) is the geometric center of all POI points of POI category L in the grid j, expressed as: Among them, (X L,i (j),Y L,i (j)) is the coordinate of the i-th POI point with POI category L in the grid j, n L (j) represents the number of POI points of POI category L in grid j; POI points are resampled for all city-wide grids of grid scale values in grid scale K. After resampling, POI points of different categories in each grid are represented by the geometric center of the corresponding category.

3. The urban land use recommendation method based on multi-scale spatial colocation patterns according to claim 1, characterized in that: Step S3: Second-order spatial co-location pattern analysis of land use types. The specific steps are as follows: S31. Select two land use types and a grid scale value; S32, generating a set of spatially adjacent instances; Spatial neighbor instance set {A s (A s ,B t ,d)} represents each instance point B of land use type B t With core point A s Connect the pairs and calculate (A s ,B t ) The distance d between two points; the set of spatial neighboring instances {B t (A s ,B t ,d)} represents each instance point A of land use type A s With core point B t Connect the pairs and calculate (A s ,B t ) the distance d between two points; S33, determining a set of spatially co-located candidate instances; Find the set {A s (A s ,B t ,d)} and the set {B t (A s ,B t ,d)}, we get the spatial co-location candidate instance set {A s ,B t ,d}; S34, determining spatial co-location patterns; The spatial co-location candidate instance set is subjected to long-distance noise removal according to the threshold value. Only the spatial co-location candidate instances whose distance between two points is within the threshold value can be retained, and the spatial co-location candidate instances to be mined are screened. s ,B t ,d|d <d ave }, where d ave is the screening threshold; S35. Identification of spatial correlation features of land use based on spatial colocation patterns; The participation index of the spatial colocation pattern is used as the characteristic value of the land use spatial association characteristics. The definition formula of the participation index is: PI(cp (A，B) )=min{Pr(cp (A，B) ,YD A ), Pr(cp (A，B) ,YD B )} Pr(cp (A，B) ,YD A )=N(cp (A，B) ,YD A ) / N(YD A ) Pr(cp (A,B) ,YD B )=N(cp (A，B) ,YD B ) / N(YD B ) Where: PI(cp (A，B) ) represents the spatial co-location pattern participation index of the two land use types A and B; Pr(cp (A，B) , YD A ) represents the participation rate of land use A in the spatial co-location pattern of two land use types A and B, N(cp (A，B) , YD A ) represents the candidate instance of the same location to be mined {A s , B t , d|d<d ave The number of instances of land use type A in}, N(YD A ) represents the total number of instance points of land use type A; similarly, Pr(cp (A，B) , YD B ) represents the participation rate of land use B in the spatial co-location pattern of two land use types A and B, N(cp (A，B) , YD B ) represents the candidate instance of the same location to be mined {A s , B t , d|d<d ave The number of instances of land use type B in}, N(YD B ) represents the total number of instance points with land use type B.

4. The urban land recommendation method based on multi-scale spatial colocation patterns according to claim 3 is characterized by: The screening threshold d in step S34 ave is a set of spatially adjacent instances {A s , B t , the average distance of d}.

5. The urban land recommendation method based on multi-scale spatial colocation patterns according to claim 1 is characterized by: Step S43 determines the type of land use in the neighborhood. The specific steps are as follows: S431. Based on the set scale parameter k0, with the land location (X0, Y0) as the center and the scale parameter k0 as the radius, a neighborhood range of the land location is delineated in the city range grid with the grid scale value k0; S432. Obtain all instance points of all land use types within the neighborhood through spatial connection, where the yth instance point of land use type x is represented by L xy ; S433, calculate each instance point L within the neighborhood xy Distance d from the location of the recommended land (X0, Y0) xy .

6. An urban land recommendation system based on multi-scale spatial co-location patterns, implementing the urban land recommendation method based on multi-scale spatial co-location patterns as claimed in any one of claims 1 to 5, characterized in that: It includes the following modules: Parameter setting module, POI data input module, POI data processing module, spatial co-location pattern recognition module and land use type recommendation module, specifically: A parameter setting module is used to set the required parameters, including a city setting unit, a grid scale parameter setting unit, a location setting unit for recommended land, and a recommended land scale setting unit; wherein the city setting unit is used to set the name of a prefecture-level city or county-level city; the scale parameter setting unit is used to set the scale for participating in the spatial co-location pattern calculation; the location setting unit for recommended land is used to set the latitude and longitude coordinates of the location of the recommended land; and the recommended land grid scale setting unit is used to set the scale for calculating the recommended spatial neighborhood. The POI data input module is used to select a city POI data file according to the city set in the parameter setting module and import the city POI data file into the system; The POI data processing module is used for extracting the city range based on POI data, resampling the POI data at multiple scales, and setting the land use type of the POI data; specifically, it includes a city range delineation unit, a POI resampling unit, and a land use type setting unit; the city range delineation unit is used to automatically delineate the city range based on the kernel density analysis results of the POI data; the POI resampling unit is used to generate grid data based on the set grid scale and POI data, and resample the POI data at each scale according to the grid geometric center point; the land use type setting unit is used to match the POI category to the corresponding land use type; The second-order spatial co-location pattern analysis module is used to obtain the second-order spatial co-location pattern correlation PI table; it includes: a spatial instance collection submodule and a land use spatial correlation feature identification submodule; The spatial neighboring instance collection submodule is used to obtain and save various spatial instances; The spatial correlation feature recognition submodule is used to traverse all combinations of land use types and grid scales, obtain PI values and save them; The land use type recommendation module is used to output a list of recommended land use types, including: a neighborhood land use extraction submodule and a land use recommendation submodule; The neighborhood land extraction submodule is used to extract the neighborhood land types within the neighborhood of the recommended land location at a specified grid scale; based on the recommended land location (X0, Y0) and scale parameters, all instance points within the neighborhood are obtained through spatial connection, and the distance d between all instance points within the neighborhood and the recommended land location (X0, Y0) is calculated. xy ; The land use recommendation submodule is used to recommend the land use type at (X0, Y0). First, the second-order spatial co-location pattern cumulative association value of the instance point with land use type x in the neighborhood is calculated. Then Sort from high to low, and extract the land use types that belong to the I′ set as the final recommended land use type for the location.

Images