An OSM building completeness assessment method based on stratified random sampling

Abstract

The application discloses an OSM building integrity evaluation method based on hierarchical random sampling, comprising the following steps: (1) based on the reference data of gridding, a classification multivariate linear regression model is used to construct a correlation model between the three indexes of GDP, population and road network length and the reference building area; (2) the fitting value of the reference building in the grid of the region to be evaluated is obtained according to the correlation model, and the fitting value of the OSM building data integrity is further calculated; (3) the fitting value of the OSM building data integrity is taken as a hierarchical index for hierarchical sampling, so as to calculate the integrity of the whole region to be evaluated. The application is based on the publicly available indexes of GDP, population and road network length, then a machine learning method is used to estimate the fitting value of the OSM building integrity, and the hierarchical index for hierarchical sampling is determined, so that the evaluation precision of the OSM building data integrity is greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of geographic information science and technology, specifically relating to the design of an OSM building integrity assessment method based on stratified random sampling. Background Technology

[0002] In recent years, Volunteered Geographic Information (VGI) has become possible. This new paradigm of geographic information, based on Web 2.0, mobile smart terminals, location service technologies, open high-resolution remote sensing imagery, and personal spatial knowledge, is created and shared spontaneously by non-professional surveyors. Open Street Map (OSM) is the most successful and well-known project within VGI, rapidly accumulating over 10 million users with its excellent user experience and rich data. Its goal is to build a comprehensive world map. OSM, with its rich semantic information, high timeliness, and free access, has become an effective supplement to professional geographic data collection and updating. Furthermore, as a popular urban analysis dataset, OSM is widely used in urban planning, land use, sustainable development goal monitoring, disaster management, public health, and other fields. However, the open, public-participatory, and non-professional nature of OSM data means it continues to face challenges related to information reliability, consistency, and usability. Therefore, evaluating the quality of OSM data has become a primary issue to address. Completeness refers to whether the geographic entities contained in the dataset are comprehensive and complete. It is an important indicator for evaluating the quality of OSM data because it is an important reference for OSM applications in urban analysis.

[0003] Currently, OSM data quality assessment is mainly divided into "external" and "internal" methods. "External" quality assessment methods use comparative evaluation with existing professional mapping data. For example, Hecht et al. used map matching (object-based) and various study unit division methods (unit-based) to measure the completeness of OSM building data in Germany, showing that the object-based method performed better. However, reference data is often difficult to obtain, so some research has turned to "internal" quality assessment methods, which analyze the quality information from the inherent characteristics of the data. For example, Barron et al. proposed a framework containing more than 25 methods and indicators, allowing for an approximate description of OSM data quality using only historical data; Zhou et al. approximated the data completeness by studying OSM building data density, and later studied a method for estimating OSM building completeness using population density. While these assessment methods do not require reference data, they are difficult to provide absolute evaluation results for OSM building data quality.

[0004] Previous research has shown that OSM building integrity is highly correlated with several other factors. For example, Tian's research indicates that demographic and socioeconomic factors influence OSM data integrity, and the length of the OSM road network can also be considered, as the OSM project initially aimed to collect road data worldwide. Therefore, it can be assumed that volunteers begin contributing building data as the road network data becomes more complete. Thus, this invention first proposes a calculation method that uses data from existing reference building data areas to fit several readily available indicator data, resulting in a universal formula for calculating reference building data. Substituting this formula into areas with unknown reference building data yields a rough fitted value for OSM building integrity in those areas. However, the development of OSM building data varies across cities, and the resulting rough fitted value cannot be used as the true value. Therefore, this invention proposes a stratified random sampling method using the obtained rough OSM building integrity as a stratification indicator to improve the accuracy of stratified sampling, as the selection of stratification indicators is crucial for stratified sampling. Since sampling only requires a small portion of the study area to be sampled for completeness calculation, the amount of reference data required is greatly reduced. Therefore, reference data can be obtained based on digitized building spatial data from remote sensing images, making sampling a low-cost and time-efficient alternative method. Summary of the Invention

[0005] The purpose of this invention is to address the problem that existing technologies struggle to provide absolute evaluation results for the quality of OSM building data, and to propose a method for assessing the integrity of OSM buildings based on stratified random sampling.

[0006] The technical solution of this invention is: an OSM building integrity assessment method based on stratified random sampling, comprising the following steps:

[0007] S1. Obtain the area data of the impermeable surface distribution of the existing reference building data area, and divide it into a kilometer grid to obtain the first kilometer grid.

[0008] S2. Overlay the reference building data, GDP data, population data, and OSM road network data in the existing reference building data area with the first kilometer grid to obtain the reference building area, total GDP, population, and OSM road network length data in each grid of the existing reference building data area, and normalize them to obtain the first normalized data.

[0009] S3. Based on the first normalized data, a multiple linear regression model is used to establish a regression model between the reference building area in each grid and the total GDP, population, and OSM road network length.

[0010] S4. Obtain the isometric data of the impermeable surface distribution area of ​​the area to be evaluated, and divide it into a kilometer grid to obtain the second kilometer grid.

[0011] S5. Overlay the GDP data, population data, OSM road network data, and OSM building data of the area to be evaluated with the second kilometer grid to obtain the total GDP, population, OSM road network length, and OSM building area data of each grid in the area to be evaluated, and normalize them to obtain the second normalized data.

[0012] S6. Input the total GDP, population and OSM road network length from the second normalized data into the regression model to obtain the normalized value of the reference building area in each grid of the area to be evaluated.

[0013] S7. Calculate the OSM building integrity fit value for each grid in the area to be evaluated based on the normalized value of the reference building area in each grid of the area to be evaluated.

[0014] S8. Using the OSM building integrity fitting value in each grid of the area to be evaluated as a stratification index, the K-means clustering algorithm is used to stratify the entire grid of the area to be evaluated.

[0015] S9. Based on the stratification results, determine the number of samples in each stratum using Neyman assignment, and then perform random sampling to obtain stratified random sampling samples.

[0016] S10. Based on orthophoto remote sensing image data, the area of ​​reference buildings in each grid of the stratified random sampling sample is obtained by artificial digitization, and the OSM building integrity of all samples is calculated accordingly.

[0017] S11. Calculate the standard deviation of OSM building integrity for all samples and determine whether the standard deviation is less than the preset threshold. If so, proceed to step S12; otherwise, return to step S9 to re-perform stratified random sampling.

[0018] S12. Extrapolate the OSM building integrity of the stratified random sampling sample to the entire area to be evaluated, and complete the evaluation of the OSM building integrity of the area to be evaluated.

[0019] Furthermore, the normalization formulas in steps S2 and S5 are as follows:

[0020]

[0021] Where y ij This indicates that the data has been normalized, x ij This represents the original indicator data, max(x) ij) represents the maximum value of the original indicator data, min(x) ij ) represents the minimum value of the original indicator data, the subscript i represents the i-th grid, and the subscript j represents the j-th indicator in each grid.

[0022] Furthermore, step S3 includes the following sub-steps:

[0023] S3-1. Constructing a multiple linear regression model:

[0024] y h =β h，0 +β h，1 *gdp+β h，2 *p+β h，3 *r

[0025] Where y h β represents the area of ​​the reference building in the h-th grid after normalization. h，0 β is a constant term. h，1 β h，2 and β h，3 These are the parameters of the multiple regression model. gdp represents the total GDP in the h-th grid after normalization, p represents the population in the h-th grid after normalization, and r represents the OSM road network length in the h-th grid after normalization.

[0026] S3-2. Divide each grid into one of the following seven types:

[0027] Type 1: gdp≠0, p≠0, r≠0;

[0028] Type 2: gdp = 0, p ≠ 0, r ≠ 0;

[0029] Type 3: gdp≠0, p=0, r≠0;

[0030] Type 4: gdp≠0, p≠0, r=0;

[0031] Type 5: gdp ≠ 0, p = 0, r = 0;

[0032] Type 6: gdp = 0, p ≠ 0, r = 0;

[0033] Type 7: gdp = 0, p = 0, r ≠ 0;

[0034] S3-3. Based on the multiple linear regression model, establish a regression model between the reference building area and the total GDP, population, and OSM road network length within each grid:

[0035] y1=β 1，0 +β 1，1 *gdp+β 1，2 *p+β1，3 *r

[0036] y2=β 2，0 +β 1，1 *gdp+β 1，2 *p

[0037] y3=β 3，0 +β 3，1 *gdp+β 3，2 *r

[0038] y4=β 4，0 ++β 4，1 *p+β 4，2 *r

[0039] y5=β 5，0 +β 5，1 *gdp

[0040] y6=β 6，0 +β 6，1 *p

[0041] y7=β 7，0 +β 7，1 *r.

[0042] Furthermore, the formula for calculating the OSM building integrity fitting value within each grid of the area to be evaluated in step S7 is as follows:

[0043]

[0044] Where c i This represents the OSM building integrity fit value for the i-th grid in the region to be evaluated. b represents the normalized value of the total OSM building area in the i-th grid of the region to be evaluated. i This represents the normalized value of the total area of ​​the reference buildings within the i-th grid of the area to be evaluated.

[0045] Furthermore, step S8 includes the following sub-steps:

[0046] S8-1. Use the OSM building integrity fitting value in each grid of the area to be evaluated as the stratification index for stratified random sampling, and use K-means clustering for stratification.

[0047] S8-2. Determine the optimal number of clusters for K-means clustering using the elbow method based on the sum of squared errors (SSE):

[0048]

[0049] Where C i Let q represent the i-th cluster, and q represent C. i The sample points in the middle, mi C represents i The center of mass, i.e., C i The mean of all samples in the dataset, where M represents the total number of clusters.

[0050] Furthermore, step S9 includes the following sub-steps:

[0051] S9-1. Determine the total sample size for stratified random sampling:

[0052]

[0053] Where N represents the total sample size of the stratified random sampling, Z represents the statistic at the current confidence level, P represents the expected proportion of the target population, and d represents the half-width of the confidence interval, i.e., the allowable error.

[0054] S9-2. The total sample size is allocated using the Neyman partitioning method to obtain the sample size for each stratum:

[0055]

[0056] Where n a N represents the number of samples in layer a. a S represents the number of elements in the a-th layer. a This represents the standard deviation of the a-th layer.

[0057] S9-3. Perform stratified sampling based on the sample size of each stratum to obtain stratified random sampling samples.

[0058] Furthermore, the formula for calculating the OSM building integrity of all samples in step S10 is as follows:

[0059]

[0060] Where C represents the OSM building integrity of the sampled area to be evaluated. B represents the total area of ​​OSM buildings within the i-th sample grid of the region to be evaluated. i represents the total area of ​​reference buildings within the i-th sample grid of the region to be evaluated, and n represents the total number of samples in the stratified random sampling.

[0061] Furthermore, in step S11, the standard deviation SD s The calculation formula is:

[0062]

[0063] Where N represents the total sample size in the stratified random sampling, and C l This represents the OSM building integrity of the l-th stratified random sample. This represents the mean of OSM building integrity across all stratified random samples.

[0064] The beneficial effects of this invention are as follows: Within the existing reference building data area, this invention successfully established the correlation between the area of ​​reference buildings in the OSM (Organizational Structure) and key indicators such as publicly available population, GDP, and OSM road length using a multiple linear regression model. Through the construction of this model, the correlation between the area of ​​reference buildings in the OSM and population, GDP, and road length can be applied to the area to be evaluated. Furthermore, the area of ​​OSM buildings in the area to be evaluated can be calculated, thereby estimating the completeness of OSM building data within that area. In addition, by using the estimated OSM building data completeness as the basis for stratified sampling, the accuracy and efficiency of sampling can be significantly improved. This invention not only improves the accuracy of data analysis but also optimizes resource allocation and data processing workflows, providing an innovative and effective data evaluation tool for related fields. Attached Figure Description

[0065] Figure 1 The diagram shows a flowchart of an OSM building integrity assessment method based on stratified random sampling provided by an embodiment of the present invention. Detailed Implementation

[0066] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary and are intended to illustrate the principles and spirit of the invention, and are not intended to limit the scope of the invention.

[0067] This invention provides an OSM building integrity assessment method based on stratified random sampling, such as... Figure 1 As shown, the process includes the following steps S1 to S12:

[0068] S1. Obtain the area data of the impermeable surface distribution of the existing reference building data area, and divide it into a kilometer grid to obtain the first kilometer grid.

[0069] S2. Overlay the reference building data, GDP data, population data, and OSM road network data in the existing reference building data area with the first kilometer grid to obtain the reference building area, total GDP, population, and OSM road network length data in each grid of the existing reference building data area, and normalize them to obtain the first normalized data.

[0070] In this embodiment of the invention, the normalization formula is:

[0071]

[0072] Where y ij This indicates that the data has been normalized, xij This represents the original indicator data, max(x) ij ) represents the maximum value of the original indicator data, min(x) ij ) represents the minimum value of the original indicator data, the subscript i represents the i-th grid, and the subscript j represents the j-th indicator in each grid.

[0073] S3. Based on the first normalized data, a multiple linear regression model is used to establish a regression model between the reference building area in each grid and the total GDP, population, and OSM road network length.

[0074] Step S3 includes the following sub-steps S3-1 to S3-3:

[0075] S3-1. Constructing a multiple linear regression model:

[0076] y h =β h，0 +β h，1 *gdp+β h，2 *p+β h，3 *r

[0077] Where y h β represents the area of ​​the reference building in the h-th grid after normalization. h，0 β is a constant term. h，1 β h，2 and β h，3 These are the parameters of the multiple regression model. gdp represents the total GDP in the h-th grid after normalization, p represents the population in the h-th grid after normalization, and r represents the OSM road network length in the h-th grid after normalization.

[0078] S3-2. Divide each grid into one of the following seven types:

[0079] Type 1: gdp≠0, p≠0, r≠0;

[0080] Type 2: gdp = 0, p ≠ 0, r ≠ 0;

[0081] Type 3: gdp≠0, p=0, r≠0;

[0082] Type 4: gdp≠0, p≠0, r=0;

[0083] Type 5: gdp ≠ 0, p = 0, r = 0;

[0084] Type 6: gdp = 0, p ≠ 0, r = 0;

[0085] Type 7: gdp = 0, p = 0, r ≠ 0;

[0086] S3-3. Based on the multiple linear regression model, establish a regression model between the reference building area and the total GDP, population, and OSM road network length within each grid:

[0087] y1=β 1，0 +β 1，1 *gdp+β 1，2 *p+β 1，3 *r

[0088] y2=β 2，0 +β 1，1 *gdp+β 1，2 *p

[0089] y3=β 3，0 +β 3，1 *gdp+β 3，2 *r

[0090] y4=β 4，0 ++β 4，1 *p+β 4，2 *r

[0091] y5=β 5，0 +β 5，1 *gdp

[0092] y6=β 6，0 +β 6，1 *p

[0093] y7=β 7，0 +β 7，1 *r.

[0094] S4. Obtain the isometric data of the impermeable surface distribution area of ​​the area to be evaluated, and divide it into a kilometer grid to obtain the second kilometer grid.

[0095] S5. Overlay the GDP data, population data, OSM road network data, and OSM building data of the area to be evaluated with the second kilometer grid to obtain the total GDP, population, OSM road network length, and OSM building area data of each grid in the area to be evaluated, and normalize them to obtain the second normalized data.

[0096] In this embodiment of the invention, the same normalization formula as in step S2 is used.

[0097] S6. Input the total GDP, population and OSM road network length from the second normalized data into the regression model to obtain the normalized value of the reference building area in each grid of the area to be evaluated.

[0098] S7. Calculate the OSM building integrity fitting value for each grid in the area to be evaluated based on the normalized value of the reference building area within each grid. The calculation formula is as follows:

[0099]

[0100] Where c i This represents the OSM building integrity fit value for the i-th grid in the region to be evaluated. b represents the normalized value of the total OSM building area in the i-th grid of the region to be evaluated. i This represents the normalized value of the total area of ​​the reference buildings within the i-th grid of the area to be evaluated.

[0101] S8. Using the OSM building integrity fitting value in each grid of the area to be evaluated as a stratification index, the K-means clustering algorithm is used to stratify the entire grid of the area to be evaluated.

[0102] Step S8 includes the following sub-steps S8-1 to S8-2:

[0103] S8-1. Use the OSM building integrity fitting value in each grid of the area to be evaluated as the stratification index for stratified random sampling, and use K-means clustering for stratification.

[0104] S8-2. Determine the optimal number of clusters for K-means clustering using the elbow method based on the sum of squared errors (SSE):

[0105]

[0106] Where C i Let q represent the i-th cluster, and q represent C. i The sample points in the middle, m i C represents i The center of mass, i.e., C i The mean of all samples in the dataset, where M represents the total number of clusters.

[0107] S9. Based on the stratification results, determine the number of samples in each stratum using Neyman assignment, and then perform random sampling to obtain stratified random sampling samples.

[0108] Step S9 includes the following sub-steps S9-1 to S9-3:

[0109] S9-1. Determine the total sample size for stratified random sampling:

[0110]

[0111] Where N represents the total sample size of the stratified random sampling, Z represents the statistic at the current confidence level (e.g., 1.96 at a 95% confidence level), P represents the expected proportion of the target population (typically 0.5), and d represents the half-width of the confidence interval, i.e., the allowable error.

[0112] S9-2. The total sample size is allocated using the Neyman partitioning method to obtain the sample size for each stratum:

[0113]

[0114] Where n a N represents the number of samples in layer a. a S represents the number of elements in the a-th layer. a This represents the standard deviation of the a-th layer.

[0115] S9-3. Perform stratified sampling based on the sample size of each stratum to obtain stratified random sampling samples.

[0116] S10. Based on orthophoto remote sensing image data, the area of ​​reference buildings in each grid of the stratified random sampling sample is obtained by artificial digitization, and the OSM building integrity of all samples is calculated accordingly.

[0117] In this embodiment of the invention, the formula for calculating the OSM building integrity of all samples is as follows:

[0118]

[0119] Where C represents the OSM building integrity of all samples in the area to be evaluated. B represents the total area of ​​OSM buildings within the i-th sample grid of the region to be evaluated. i represents the total area of ​​reference buildings within the i-th sample grid of the region to be evaluated, and n represents the total number of samples in the stratified random sampling.

[0120] S11. Calculate the standard deviation (SD) of OSM building integrity for all samples. s Judgment standard deviation SD s Is it less than the preset threshold SD? th If yes, proceed to step S12; otherwise, return to step S9 to perform stratified random sampling again.

[0121] In this embodiment of the invention, the standard deviation SD s The calculation formula is:

[0122]

[0123] Where N represents the total sample size in the stratified random sampling, and C l This represents the OSM building integrity of the l-th stratified random sample. This represents the mean of OSM building integrity across all stratified random samples.

[0124] S12. Extrapolate the OSM building integrity of the stratified random sampling sample to the entire area to be evaluated, and complete the evaluation of the OSM building integrity of the area to be evaluated.

[0125] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for assessing the integrity of OSM buildings based on stratified random sampling, characterized in that, Includes the following steps: S1. Obtain the area data of the impermeable surface distribution of the existing reference building data area, and divide it into a kilometer grid to obtain the first kilometer grid; S2. Overlay the reference building data, GDP data, population data, and OSM road network data in the existing reference building data area with the first kilometer grid to obtain the reference building area, total GDP, population, and OSM road network length data in each grid of the existing reference building data area, and normalize them to obtain the first normalized data. S3. Based on the first normalized data, a multiple linear regression model is used to establish a regression model between the reference building area in each grid and the total GDP, population, and OSM road network length. S4. Obtain the isometric data of the impermeable surface distribution area of ​​the area to be evaluated, and divide it into a kilometer grid to obtain the second kilometer grid; S5. Overlay the GDP data, population data, OSM road network data, and OSM building data of the area to be evaluated with the second kilometer grid to obtain the total GDP, population, OSM road network length, and OSM building area data of each grid in the area to be evaluated, and normalize them to obtain the second normalized data. S6. Input the total GDP, population and OSM road network length from the second normalized data into the regression model to obtain the normalized value of the reference building area in each grid of the area to be evaluated. S7. Calculate the OSM building integrity fitting value for each grid in the area to be evaluated based on the normalized value of the reference building area in each grid of the area to be evaluated. S8. Using the OSM building integrity fitting value in each grid of the area to be evaluated as a stratification index, the K-means clustering algorithm is used to stratify the grid of the entire area to be evaluated. S9. Based on the stratification results, determine the number of samples in each stratum using Neyman allocation, and then perform random sampling to obtain stratified random sampling samples. S10. Based on orthophoto remote sensing image data, the area of ​​reference buildings in each grid of the stratified random sampling sample is obtained by artificial digitization, and the OSM building integrity of all samples is calculated accordingly. S11. Calculate the standard deviation of OSM building integrity for all samples and determine whether the standard deviation is less than the preset threshold. If it is, proceed to step S12; otherwise, return to step S9 to re-perform stratified random sampling. S12. Extrapolate the OSM building integrity of the stratified random sampling sample to the entire area to be evaluated, and complete the evaluation of the OSM building integrity of the area to be evaluated.

2. The OSM building integrity assessment method according to claim 1, characterized in that, The normalization formulas in steps S2 and S5 are as follows: in This indicates that the data has been normalized. This represents the original indicator data. This represents the maximum value of the original indicator data. Indicates the minimum value of the original indicator data, subscript Indicates the first Each grid, subscript Represents the first in each grid. One indicator.

3. The OSM building integrity assessment method according to claim 1, characterized in that, Step S3 includes the following sub-steps: S3-1. Constructing a multiple linear regression model: in Indicates the normalized i-th h Reference building area within each grid, For constant terms, , and These are the parameters of the multiple regression model. Indicates the normalized i-th h The total GDP value within each grid. Indicates the normalized i-th h Population value within each grid Represented as the normalized i-th h OSM road network length value within each grid; S3-2. Divide each grid into one of the following seven types: Type 1: ; Type 2: ; Type 3: ; Type 4: ; Type 5: ; Type 6: ; Type 7: ; S3-3. Based on the multiple linear regression model, establish a regression model between the reference building area and the total GDP, population, and OSM road network length within each grid: 。 4. The OSM building integrity assessment method according to claim 1, characterized in that, The formula for calculating the OSM building integrity fitting value in each grid of the area to be evaluated in step S7 is as follows: in Indicates the area to be evaluated. OSM building integrity fit values ​​for each grid. Indicates the area to be evaluated. Normalized values ​​of OSM building area within each grid cell Indicates the area to be evaluated. Normalized values ​​of the reference building area within each grid.

5. The OSM building integrity assessment method according to claim 1, characterized in that, Step S8 includes the following sub-steps: S8-1. Use the OSM building integrity fitting value in each grid of the area to be evaluated as the stratification index for stratified random sampling, and use K-means clustering for stratification. S8-2, Using error summation-based approach SSE The elbow method is used to determine the optimal number of clusters for K-means clustering: in Indicates the first Clusters, express The sample points in express The center of mass, that is The mean of all samples in the sample. This indicates the total number of clusters.

6. The OSM building integrity assessment method according to claim 1, characterized in that, Step S9 includes the following sub-steps: S9-1. Determine the total sample size for stratified random sampling: in This represents the total sample size in stratified random sampling. This represents the statistic at the current confidence level. This represents the expected proportion of the target population. This represents the half-width of the confidence interval, i.e., the allowable error; S9-2. The total sample size is allocated using the Neyman partitioning method to obtain the sample size for each stratum: in Indicates the first Number of layer samples Indicates the first The number of elements in the layer Indicates the first Standard deviation of the layer; S9-3. Perform stratified sampling based on the sample size of each stratum to obtain stratified random sampling samples.

7. The OSM building integrity assessment method according to claim 1, characterized in that, The formula for calculating the OSM building integrity of all samples in step S10 is as follows: in This indicates the OSM building integrity of the sampled area to be evaluated. Indicates the area to be evaluated. OSM building area within each grid Indicates the area to be evaluated. Reference building area within each grid, This indicates the total number of grid cells in the area to be evaluated.

8. The OSM building integrity assessment method according to claim 1, characterized in that, Standard deviation in step S11 The calculation formula is: in This represents the total sample size in stratified random sampling. Indicates the first OSM building integrity of a stratified random sample This represents the mean of OSM building integrity across all stratified random samples.

Images