WO2020151528A1 - 一种综合业态大数据与建筑形态的城市用地自动识别系统 - Google Patents

一种综合业态大数据与建筑形态的城市用地自动识别系统 Download PDF

Info

Publication number
WO2020151528A1
WO2020151528A1 PCT/CN2020/071915 CN2020071915W WO2020151528A1 WO 2020151528 A1 WO2020151528 A1 WO 2020151528A1 CN 2020071915 W CN2020071915 W CN 2020071915W WO 2020151528 A1 WO2020151528 A1 WO 2020151528A1
Authority
WO
WIPO (PCT)
Prior art keywords
land
business
urban
data
plot
Prior art date
Application number
PCT/CN2020/071915
Other languages
English (en)
French (fr)
Inventor
杨俊宴
邵典
王桥
张宇豪
曾宪涵
刘志成
叶晟之
Original Assignee
东南大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 东南大学 filed Critical 东南大学
Priority to US16/963,506 priority Critical patent/US11270397B2/en
Priority to EP20745903.3A priority patent/EP3916668A4/en
Publication of WO2020151528A1 publication Critical patent/WO2020151528A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/26—Government or public services
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2431—Multiple classes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/16—Real estate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/176—Urban or other man-made structures

Definitions

  • the invention belongs to the field of urban planning, and relates to an urban land automatic identification system, in particular to an urban land automatic identification system that integrates big data of business conditions and architectural forms.
  • Urban land is the planning and design basis of the urban planning discipline, and is a collective term for land with certain uses and functions.
  • the layout of urban land has become more and more complicated, which has increased the difficulty and time cost of measuring the nature of urban land.
  • the layout and scale of urban land plots show distinct characteristics of differentiation, and the differences in business and architectural forms within the same type of land have become more obvious. Identifying the nature of urban land use is the basis for the development of various urban planning and design work. By identifying urban land use and analyzing the current construction characteristics, spatial pattern and land use issues of the city, the current construction evaluation and planning and design work are carried out.
  • the common urban land identification method is to use manual on-site surveying and mapping, combined with the current topographic map, and based on the results of comprehensive judgments on the buildings, business conditions, functions, and public spaces of each block.
  • Such identification has a long time for surveying and mapping and requires manpower.
  • Large material resources and land use identification involve problems such as human brain judgment, construction of complex plots, easy to judge errors and large randomness; the other is to perform unsupervised clustering identification only through the business state POI, such identification ignores the relationship between the urban building form and urban land Due to the single data dimension, the error coefficient of the recognition result is large, and it can only identify the large-scale land use, which cannot achieve the purpose of finely identifying the small-scale urban land.
  • the objective of the present invention is to provide a system for automatically recognizing urban land by comprehensively extracting the data distribution characteristics of urban business status points and the multi-dimensional morphological characteristics of three-dimensional entities in urban space, which can efficiently and automatically generate urban land results for cities of different sizes. And give the corresponding confidence.
  • an urban land automatic identification system integrating business big data and architectural morphology includes:
  • the data acquisition and input module is used to acquire and store the spatial vector data and business point data in the built-up areas of case cities and target cities of different scales, and input them into the geographic information system;
  • case cities of different scales are divided according to the latest city scale classification standards issued by the State Council.
  • Five types of case cities are selected: super large cities, extra large cities, large cities, medium cities, and small cities.
  • Each type of city generates a database and Machine learning classification model;
  • the spatial vector data in the urban built-up area includes a polygonal block with a closed contour; wherein the block containing more than one contour is a closed polygonal block, and the block containing more than one contour is A closed polygonal building, which has information on the number of floors or height;
  • the space vector platform is used for vector data processing
  • the space vector platform includes: ArcGIS and CAD;
  • the business type point data includes business type point names, geographic coordinates, and business type feature type information; wherein, the business type feature type information is industry information to which the business type point belongs, or business type point function information or business type point classification information.
  • the database construction module is used to calibrate and correlate data according to the set spatial automatic calibration method through the geographic information system to obtain the land parcel database of the associated business point, and weight the land parcel database of the correlated business point according to the importance of the business status Processing, automatically calculate the highest height, average height, average building base area, floor area ratio index of all buildings in each plot according to the set building form feature index, and associate it with the plot, generate and store the weighted format features,
  • the spatial automatic calibration method includes spatial calibration of business status points and urban blocks, spatial calibration of business status points and land parcels, and association of business status points with the properties of the land parcels where they are located and urban blocks. Dilate operation (Dilate), and then perform spatial connection (Spatial Join) with business data to compensate for data errors;
  • the land parcel database of the associated business status points is composed of the characteristic frequency of each business status in the statistical land parcel;
  • the weighting processing of the land parcel database of the associated business format point includes the TF-IDF algorithm conversion on the frequency of the business format feature, wherein the TF method normalizes the format data, and a certain format feature is in a lot The ratio of the frequency of the business type to the frequency of this business type characteristic in the entire city is obtained; the IDF method measures the importance of the business type characteristic by the ratio of the total number of land parcels in the city to the number of land parcels containing this type of business characteristic.
  • the TF-IDF algorithm is specifically as follows:
  • i is the parcel number
  • j is the format feature number
  • n i,j is the frequency of the j-th format feature of the i-th block
  • K is the dimension of the format feature
  • is the total number of plots in the city
  • ⁇ J:t i ⁇ D j ⁇ is the number of plots with a feature frequency of the j-th format that is not 0.
  • the highest height index of all buildings in the plot is the maximum height of all buildings in each plot
  • the average height index of all the buildings in the plot is the average value of the heights of all the buildings in each plot
  • the average base area index of all buildings in the plot is the average value of the geometric area of all closed polylines of all buildings in each plot
  • the floor area ratio index of all buildings in the plot is the ratio of the sum of the product of the closed polyline geometric area of each building in each plot and the number of building floors to the geometric area of the plot closed polyline.
  • the machine learning training module is used to input the standard data on the nature of land use in case cities of different scales.
  • the standard data on the nature of land use is classified according to the sub-categories of land use and used as machine learning labels;
  • a supervised classification learning algorithm is used to construct the database to include Machine learning training is carried out on the case city database of different scales with weighted business characteristics and morphological characteristics, and multiple machine learning classification models corresponding to different city scales are generated, and they are combined to form a cluster of automatic urban land identification models;
  • the supervised classification learning algorithm is based on different city scales, calling the database construction module of different scale case city databases containing weighted business characteristics and morphological characteristics, and according to the actual measured land property data set of typical cities, according to the city scale. Carry out model training and parameter optimization;
  • the data set of actually measured land use properties is divided into training set, validation set and test set according to proportion sampling, and the machine learning model with excellent classification performance and generalization performance is selected through cross-validation, parameter optimization, and generalization test. The final model.
  • the automatic recognition module is used to obtain and input the spatial vector data and business point data in the built-up area of the target city in the data acquisition and input module after the target city area is selected, according to the weighted business characteristics and shape generated in the database building module Characteristic target city database, using the trained urban land automatic recognition model cluster, automatically identifying and outputting the possible land use category and its confidence level of each plot in the built area of the target city;
  • the data output module is used to obtain the land properties corresponding to each land parcel output by the automatic identification module, and fill the land parcels of the same type with one color with the accuracy of sub-category land properties to identify the urban land in the area
  • the recognition result is displayed as a colored block of the city's current land property map.
  • the present invention has the following beneficial effects:
  • the standard data on the nature of urban land used in the machine learning and the land use data identified by the system are accurate to the sub-category of land use, ensuring the accuracy and practicality of the automatic identification results of urban land use;
  • the automatic identification of urban land avoids the problems of long surveying and mapping time, large investment of manpower and material resources, land use identification involving human judgment, construction of complex plots, easy judgment errors and large randomness in traditional manual field surveying and mapping, and realizes
  • the high-efficiency, full-process automation, and intelligent identification of land-use properties provide an efficient and convenient surveying and mapping approach and reference for the measurement and mapping of the current urban construction land, effectively saving the time and cost of manual surveying and mapping.
  • Figure 1 is a flowchart of a method according to an embodiment of the present invention.
  • FIG. 2 is a spatial connection (Spatial Join) diagram between a format point and a land parcel according to an embodiment of the present invention
  • FIG. 3 is a performance comparison diagram of various machine learning models according to an embodiment of the present invention.
  • FIG. 4 is a diagram of hyperparameter tuning of a GBDT model according to an embodiment of the present invention.
  • Figure 5.1 is a map of the current urban land use properties according to an embodiment of the present invention
  • Figure 5.2 is a partial enlarged view of Figure 5.1.
  • an embodiment of the present invention discloses an automatic urban land identification system that integrates big data of business and architectural forms, and the system includes:
  • the data acquisition and input module is used to acquire and store the space vector data and business point data in the built-up areas of case cities and target cities of different scales, and input them into the space vector platform.
  • case cities of different scales are divided according to the latest city scale classification standards issued by the State Council, and five types of case cities of super large cities, mega cities, large cities, medium cities, and small cities are selected, and a database and machine are generated for each type of city.
  • Learning classification model
  • the spatial vector data and business status point data can be obtained through government departments or data providers in other related fields, and can also be processed through geographic information system processing software or image data software, and processed and acquired by combining software code programming.
  • the space vector data usually contains the vector data of the closed polygonal block with the outline of the city (it can also be generated by the road red line), and the vector data of the closed polygonal block with the outline (usually provided by the government department, or through the internal block Branch roads, residential roads, etc.
  • business point data usually includes geographic coordinate data of the business point (the data coordinates can be converted through the projection operation in ArcGIS, including projection coordinates, geographic The switching between coordinates and the conversion between different coordinate systems), the name of the business type, and the information of the business type feature type (the type of business type information is the industry information of the business point or the functional information of the business point or the classification information of the business point).
  • the above data can be XLS Format, CSV format, DWG format or SHP format.
  • the space vector platform is used for vector data processing, including: ArcGIS and CAD;
  • the database construction module constructs a land parcel database including weighted business characteristics and morphological characteristics through spatial automatic calibration correlation and architectural form characteristic index calculation.
  • the land parcel database includes a target city database and a database of case cities of different scales.
  • the spatial automatic correction and correlation refers to the automatic spatial correlation and correction between the plots and the business point using the expansion method and the spatial connection method in consideration of the inevitable drift and loss of accuracy of the data coordinates. It should be noted that there is inevitably a drift phenomenon in the coordinates of the business status, that is, a few meters of error and misalignment with the real space position. At the same time, due to the accuracy of floating-point numbers, the GPS coordinates also have the problem of accuracy loss.
  • the expansion method refers to a certain scale enlargement and extension of the boundary of each plot, so that each plot can contain the business point of itself and its neighbors, which has a higher drift error and precision loss. Tolerance.
  • the spatial connection method refers to the land parcel layer and the industry point layer as two superimposed layers, and the spatial connection index can be established if there is a spatial inclusion relationship.
  • the above methods can be implemented by tools in ArcGIS, and can also be implemented by geopandas software package programming.
  • TF-IDF algorithm is mainly used to weight the characteristics of business status points.
  • the TF method normalizes the characteristic data of the business, which is obtained from the ratio of the characteristic frequency of a certain business in a plot to the characteristic frequency of the business in the entire city;
  • the IDF method measures the importance of the business characteristics, which is determined by the The logarithm of the ratio of the total number of land parcels to the number of land parcels containing this type of business feature is obtained.
  • the TF-IDF algorithm is specifically as follows:
  • i is the parcel number
  • j is the format feature number
  • n i,j is the frequency of the j-th format feature of the i-th block
  • K is the dimension of the format feature
  • is the total number of plots in the city
  • ⁇ J:t i ⁇ D j ⁇ is the number of plots with a feature frequency of the j-th format that is not 0.
  • the building form index includes the highest height index, the average height index, the average building base area index, and the floor area ratio index of all buildings in the plot.
  • the area of each plot and the base area of each building can be obtained by calculating the polygonal space geometry composed of closed polylines.
  • the highest height index is the maximum value of all building heights in the plot; the average height index is the average value of all building heights in the plot; the average building base area index is the average geometric area of all closed polylines of the building in the plot; volume The rate index is the ratio of the sum of the product of the geometric area of each closed polyline of each building and the number of building floors to the geometric area of the closed polyline of the plot.
  • the machine learning training module uses the standard data of urban land properties of different scales as learning labels, and uses a supervised classification learning algorithm to perform machine learning training on the case city databases of different scales that contain the weighted business characteristics and morphological characteristics obtained by the database building module.
  • a machine learning classification model corresponding to different city scales is combined to form a cluster of automatic urban land recognition models.
  • the characteristic data of typical cities in the database of case cities of different scales collected by the data acquisition and input module are obtained, and used as training data. Train independent models according to different city scales to meet the needs of various city identification.
  • the training data is divided into training set, validation set and test set through proportional sampling.
  • the input of the model is the weighted business characteristics and building form characteristics of the plot, and the output is the possible land use category, and the label is the land used by field measurement. Nature provided.
  • Commonly used supervised learning classification algorithms include logistic regression, tree model, support vector machine, ensemble model, etc. Considering that the plot labels are divided into sub-categories, the number of categories is large, and the feature dimension of the plot is relatively high, which includes both the weighted format feature dimension and the building form feature dimension of the plot, in order to improve the classification of the model For accuracy, it is recommended to use optimized ensemble tree models for classification, such as gradient boosting decision tree model (GBDT), xgboost model, etc.
  • GBDT gradient boosting decision tree model
  • xgboost model etc.
  • the automatic identification module uses the trained urban land automatic identification model cluster to identify the land properties and confidence of each plot.
  • the automatic identification module calls the model clusters that have been trained in the machine learning training module.
  • the model clusters correspond to various city sizes.
  • the automatic identification module only needs to set parameters to determine the size of the target city area, and then automatically call the corresponding classification model. .
  • the system needs to automatically identify the land use function of a certain target city area, first select the corresponding classification model according to the city scale of the target city area, and then use the database to obtain the collected business and architectural features of the target city area as features The vector is input into the classification model, and finally the possible land-use property categories of each block in the target city area are generated, and the classification confidence is given.
  • the data output module enters the different color blocks corresponding to the nature of the land into the plot vector file and annotates its confidence level, and prints the results into a drawing to obtain a map of the current urban land use nature.
  • the 8 major categories, 35 middle categories and 35 middle categories in the urban construction land classification in the urban land classification and planning construction land standard are usually adopted.
  • 42 sub-category standards it is also possible to use the city classification land use standards of each region and city as the accuracy, fill the land parcels of the same type of land with one color, and automatically mark the confidence level of each land recognition result.
  • the result of urban land recognition in the recognition area is printed and displayed as a flat vector image of colored blocks through a full-color inkjet printer.
  • Utilizing the urban land automatic identification system that integrates big data of business types and architectural forms of the embodiments of the present invention can efficiently, automatically and finely identify the nature of urban land of different scales, thereby generating urban land identification results and giving corresponding confidence levels. It avoids the time-consuming, labor-intensive, and easy judgment errors that are likely to occur in traditional manual on-site surveying and mapping, and provides an efficient and convenient surveying and mapping approach and reference for the measurement and mapping of the current urban construction land, effectively saving the time and cost of manual surveying and mapping .
  • the data acquisition and input modules respectively acquire the space vector data and business status point data in the built-up areas of each city, and enter them into the space vector platform; specifically including:
  • the database construction module numbers the city's plot units, spatially associates and calibrates the business points, and counts the number of business points with different characteristics in each plot, and generates a plot database of related business points. Weighting the features of the business status, and then performing calculations on the various building shape data in the plot to generate a plot feature attribute table containing the weighted business features and building features, and generate a database of the land parcels containing the weighted business features and morphological features; Specifically:
  • TF-IDF algorithm processing is performed on the completed land parcel database of the associated business point, and the TF-IDF algorithm is specifically shown in the following formula:
  • i is the parcel number
  • j is the format feature number
  • n i,j is the frequency of the j-th format feature of the i-th block
  • K is the dimension of the format feature
  • is the total number of plots in the city
  • ⁇ J:t i ⁇ D j ⁇ is the number of plots where the characteristic frequency of the j-th format is not 0;
  • each format feature can be re-weighted according to the frequency and importance of its appearance, and a plot feature attribute table containing the weighted format feature can be obtained;
  • Hmax is the highest height index of the plot
  • Have is the average height index of the plot
  • Save is the average base area index of the building
  • FAR is the floor area ratio index
  • A is the base area of the building
  • F is the number of building floors
  • B is the area of the plot.
  • N is the total number of buildings in the plot.
  • the machine learning training module obtains data from the five cities of Shanghai, Nanjing, Harbin, Zhenjiang, and Zhangjiagang from the data acquisition and input module, and performs data set sampling and segmentation, model selection and training, and parameter optimization and crossover After verification, the final model is obtained, and multiple models trained on data of different city sizes are combined into a model cluster.
  • Shenzhen which represents a megacity, as an example, (3) specifically includes:
  • the machine learning training module extracts typical urban land parcel data including business and architectural morphological characteristics and land-use property tags from the database construction module. Each typical city represents a city scale;
  • the automatic identification module automatically recognizes the nature of land use in the target city area to be identified in Shenzhen. From the target city database containing business and architectural features generated by the database construction module, the business features and architectural features are obtained. The population of the city determines its city scale, selects the Shanghai urban land model, which is also a super large-scale city, from the model cluster, inputs Shenzhen plot data, generates and outputs a table of possible land use nature categories for each plot in Shenzhen, and gives Out classification confidence;
  • the data output module inputs the output properties of the target city land corresponding to different color blocks into the plot vector file and annotates its confidence level, and prints the results into drawings to obtain a map for the city, including:

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Tourism & Hospitality (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Medical Informatics (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Marketing (AREA)
  • Human Resources & Organizations (AREA)
  • General Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Primary Health Care (AREA)
  • Economics (AREA)
  • Educational Administration (AREA)
  • Development Economics (AREA)
  • Mathematical Physics (AREA)
  • Databases & Information Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Instructional Devices (AREA)

Abstract

本发明公开了一种综合业态大数据与建筑形态的城市用地自动识别系统，包括数据获取及输入模块、数据库建构模块、机器学习训练模块、自动识别模块及数据输出模块，该系统通过提取城市业态点的数据分布特征及城市空间三维实体的多维度形态特征进行城市用地的自动识别。本发明能够应对城市规划设计领域对城市地块用地性质的判定，实现基于人工智能系统对不同规模城市用地性质的高效自动化精细识别，为城市现状建设用地的测度和绘制提供了高效便捷的测绘途径及参考，有效节约人工测绘的时间成本。

Description

一种综合业态大数据与建筑形态的城市用地自动识别系统 技术领域
本发明属于城市规划领域，涉及一种城市用地自动识别系统，特别涉及一种综合业态大数据与建筑形态的城市用地自动识别系统。
背景技术
城市用地是城市规划学科的规划及设计基础，是具有一定用途及功能土地的统称。随着城市化进程快速稳定发展，城市规模在不断扩张的同时，城市用地布局也变得愈加复杂，这为城市用地性质的测度加大了难度及时间成本。另一方面，不同规模等级的城市，其城市用地地块布局及尺度呈现出明显的差异化特征，同类型用地内的业态特征、建筑形态差异愈发明显。识别城市用地性质是城市各类规划及设计工作开展的基础，通过识别城市用地分析城市现状建设特征、空间格局及土地利用问题，进而开展现状建设评价及规划设计工作。
目前常见的城市用地识别方法，一种是通过人工现场测绘，结合现状地形图，根据各地块的建筑、业态、职能、公共空间等综合判断的结果，这样的识别存在测绘时间长、投入人力物力大、用地识别涉及人脑判断、建设复杂地块易判断失误且随意性大等问题；另一种是仅通过业态POI进行无监督聚类识别，这样的识别忽略城市建筑形态与城市用地间的关联性，因数据维度的单一导致识别结果的误差系数较大，且仅能识别至用地大类，无法达到精细识别城市小类用地的目的。
发明内容
发明目的：本发明目的在于提供一种综合提取城市业态点的数据分布特征及城市空间三维实体的多维度形态特征进行城市用地自动识别的系统，能够对不同规模的城市高效自动化精细生成城市用地结果并给出相对应置信度。
技术方案：为实现上述目的，本发明所述的一种综合业态大数据与建筑形态的城市用地自动识别系统，该系统包括：
数据获取及输入模块，用于获取并存储不同规模案例城市及目标城市建成区域内的空间矢量数据、业态点数据，输入地理信息系统；
进一步地，所述不同规模案例城市按照国务院印发的最新城市规模划分标准划分， 选取超大城市、特大城市、大城市、中等城市、小城市五类规模的案例城市，每类城市对应生成一个数据库及机器学习分类模型；
进一步地，所述城市建成区域内的空间矢量数据包括轮廓为闭合的多边形街区；其中，所述街区内包含一个以上的轮廓为闭合的多边形地块，所述地块内包含一个以上的轮廓为闭合的多边形建筑，所述建筑具备层数或高度信息；
所述空间矢量平台用于矢量数据处理；
进一步地，所述空间矢量平台包括：ArcGIS、CAD；
所述业态点数据包括业态点名称、地理坐标及业态特征类型信息；其中，所述业态特征类型信息为业态点所属行业信息或业态点职能信息或业态点分类信息。
数据库建构模块，用于通过地理信息系统，按照设定的空间自动校准方法对数据进行校准并关联，得到关联业态点的地块数据库，根据业态特征重要性对关联业态点的地块数据库进行加权处理，按照设定的建筑形态特征指标自动计算出每个地块内所有建筑的最高高度、平均高度、建筑平均基底面积、容积率指标，并与地块关联，生成并存储包含加权业态特征、形态特征的地块数据库，所述地块数据库包含目标城市数据库和不同规模案例城市数据库；
进一步地，所述空间自动校准方法，包括业态点与城市街区的空间校准、业态点与用地地块的空间校准、业态点与所在用地地块及城市街区的属性关联，先对城市地块进行膨胀操作(Dilate)，再与业态数据进行空间连接(Spatial Join)，以弥补数据误差；
进一步地，所述关联业态点的地块数据库，由统计地块中的各个业态特征频数构成；
进一步地，所述对关联业态点的地块数据库进行加权处理，包括对业态特征频数进行TF-IDF算法转化，其中TF方法对业态数据进行归一化处理，由某业态特征在一个地块中的频数与以这一业态特征在整个城市中的频数的比值得到；IDF方法对业态特征的重要性进行度量，由城市中的地块总数与包含这一业态特征的地块数的比值的对数值得到；
所述TF-IDF算法具体如下公式所示：
Figure PCTCN2020071915-appb-000001
其中，i为地块编号，j为业态特征编号，n i,j第i个地块的第j个业态特征的频数，K为业态特征的维数，|D|为城市中的地块总数，{j:t i∈D j}为第j个业态特征频数不为 0的地块数量。
进一步地，所述地块内所有建筑的最高高度指标为每个地块内所有建筑高度的最大值；
进一步地，所述地块内所有建筑的平均高度指标为每个地块内所有建筑高度的平均值；
进一步地，所述地块内所有建筑的平均基底面积指标为每个地块内所有建筑闭合多段线的几何面积的平均值；
进一步地，所述地块内所有建筑的容积率指标为每个地块内每个建筑闭合多段线几何面积与建筑层数乘积之总和与地块闭合多段线几何面积的比值。
机器学习训练模块，用于输入不同规模案例城市的用地性质标准数据，所述用地性质标准数据按照用地小类划分，并作为机器学习标签；采用有监督分类学习算法，对数据库建构模块得到的包含加权业态特征、形态特征的不同规模案例城市数据库进行机器学习训练，产生多个对应不同城市规模的机器学习分类模型，将其组合形成城市用地自动识别模型簇；
进一步地，所述有监督分类学习算法是根据不同城市规模，调用数据库建构模块中包含加权业态特征、形态特征的不同规模案例城市数据库，根据其中典型城市的实测用地性质数据集，按照城市规模分别进行模型的训练和参数优化；
进一步地，所述实测用地性质数据集，按比例抽样切分为训练集、验证集和测试集，通过交叉验证、参数优化、泛化检验选取分类性能与泛化性能均优的机器学习模型作为最终模型。
自动识别模块，用于在选择目标城市区域后，获取并输入数据获取及输入模块中的目标城市建成区域内的空间矢量数据、业态点数据，根据数据库建构模块中生成的包含加权业态特征、形态特征的目标城市数据库，采用训练完成的城市用地自动识别模型簇，自动识别并输出目标城市建成区域内每个地块可能的用地性质类别及其置信度；
数据输出模块，用于获取自动识别模块输出的每个地块对应的用地性质，以小类用地性质为精度，对同一类用地性质的地块进行一种颜色的填充，将识别区域的城市用地识别结果以着色块的城市现状用地性质图进行显示。
有益效果：本发明具有以下有益效果：
1、综合业态点特征及建筑形态特征，对照实测所得的城市用地性质标准数据，利用有监督分类学习算法进行训练，最大程度提高系统识别的准确性；
2、对业态点数据根据其重要程度进行加权，避免不同标签类型的业态因数量级差异过大而导致的识别误差；
3、纳入多个建筑形态特征指标，弥补仅通过业态点识别城市用地导致的局限性，最大程度逼近人工识别城市用地的常用方法；
4、根据不同规模城市建构对应的数据库，并训练产生多个对应不同城市规模的机器学习分类模型，缩小因城市规模不同而带来的识别误差，确保城市用地自动识别系统适用于不同规模城市的用地识别；
5、机器学习中所对照的城市用地性质标准数据及系统识别出的用地数据均精确至用地小类，保证了城市用地自动识别结果的精确性及实用性；
6、对城市用地进行自动识别避免了传统人工现场测绘中测绘时间长、投入人力物力大、用地识别涉及人脑判断、建设复杂地块易判断失误且随意性大的问题，实现对不同规模城市用地性质的高效、全流程自动化、用地性质精确化的智能识别，为城市现状建设用地的测度和绘制提供了高效便捷的测绘途径及参考，有效节约人工测绘的时间成本。
附图说明
图1为本发明实施例的方法流程图；
图2为本发明实施例的业态点与地块的空间连接(Spatial Join)图；
图3为本发明实施例的各机器学习模型性能比较图；
图4为本发明实施例的GBDT模型超参数调优图；
图5.1为本发明实施例的城市现状用地性质图；图5.2为图5.1的局部放大图。
具体实施方式
下面结合附图和实施例对本发明的技术方案作进一步详细的说明。
如图1所示，本发明实施例公开了一种综合业态大数据与建筑形态的城市用地自动识别系统，该系统包括：
数据获取及输入模块，用于获取并存储不同规模案例城市及目标城市建成区域内的空间矢量数据、业态点数据，输入空间矢量平台。
其中，所述不同规模案例城市按照国务院印发的最新城市规模划分标准划分，选取超大城市、特大城市、大城市、中等城市、小城市五类规模的案例城市，每类城市对应生成一个数据库及机器学习分类模型；
所述空间矢量数据及业态点数据可以通过政府部门或其他相关领域的数据提供商获得，也可以通过地理信息系统处理软件或图像数据软件进行处理，并结合软件代码编程自行加工获取。其中，空间矢量数据通常包含城市的轮廓为闭合的多边形街区矢量数据(也可以通过道路红线围合成面生成)、轮廓为闭合的多边形地块矢量数据(通常为政府部门提供，或通过街区内部的支路、小区路等将街区切分生成)、轮廓为闭合的多边形建筑矢量数据、每个闭合建筑的高度/层数信息(在没有高度的情况下，通过建筑层数推算建筑高度，建筑高度＝建筑层数*3米)，以上数据可以为DWG格式或SHP格式等；业态点数据通常包含业态点的地理坐标数据(可以通过ArcGIS中的projection操作对数据坐标进行转换，包括投影坐标、地理坐标间的切换以及不同坐标系之间的转换)、业态点名称、业态特征类型信息(业态特征类型信息为业态点所属行业信息或业态点职能信息或业态点分类信息)，以上数据可以为XLS格式、CSV格式、DWG格式或SHP格式。
所述空间矢量平台用于矢量数据处理，包括：ArcGIS、CAD；
数据库建构模块，通过空间自动校准关联及建筑形态特征指标计算，建构包含加权业态特征、形态特征的地块数据库，所述地块数据库包含目标城市数据库和不同规模案例城市数据库。
所述空间自动校正关联指的是在考虑到不可避免的数据坐标存在漂移和精度损失的情况下，利用膨胀法和空间连接法进行地块与业态点的自动空间关联和校正。需要说明的是，业态点坐标不可避免的存在漂移现象，即与真实空间位置存在几米的误差和错位，同时，由于浮点数精度的问题，GPS坐标也存在精度损失问题。膨胀法是指针对每个地块，对地块的边界进行一定尺度的放大和延展，使得每个地块可以包含自身及其近邻位置的业态点，对漂移误差和精度损失有了更高的容忍度。空间连接法是指把地块图层和业态点图层看做两个叠加图层，存在空间包含关系的即可建立空间连接索引。以上方法均可通过ArcGIS中的工具进行实现，亦可通过geopandas软件包编程实现。
在完成膨胀和空间连接后即可通过统计地块内业态特征频数得到关联业态点的地块属性表，并生成关联业态点的地块数据库，下一步需要对数据库中的业态特征根据其重要程度进行加权处理，并加入建筑形态特征指标。
一般而言，对业态点特征加权主要采用TF-IDF算法实现。其中TF方法对业态特征数据进行归一化处理，由一个地块中的某业态特征频数与整个城市中该业态特征频数的 比值得到；IDF方法对业态特征的重要性进行度量，由城市中的地块总数与包含这一业态特征的地块数的比值的对数值得到。所述TF-IDF算法具体如下公式所示：
Figure PCTCN2020071915-appb-000002
其中，i为地块编号，j为业态特征编号，n i,j第i个地块的第j个业态特征的频数，K为业态特征的维数，|D|为城市中的地块总数，{j:t i∈D j}为第j个业态特征频数不为0的地块数量。
按照设定的建筑形态特征指标自动计算出每个地块的建筑形态指标数据，生成包含加权业态特征、形态特征的地块属性表，生成并存储包含加权业态特征、形态特征的地块数据库。所述建筑形态指标包括地块内所有建筑的最高高度指标、平均高度指标、建筑平均基底面积指标、容积率指标。通过对闭合多段线构成的多边形空间几何计算可以获得每个地块的面积及每个建筑的基底面积(包括多边形轮廓面积)。最高高度指标是地块内所有建筑高度的最大值；平均高度指标是地块内所有建筑高度的平均值；建筑平均基底面积指标是地块内所有建筑闭合多段线的几何面积的平均值；容积率指标是地块内每个建筑闭合多段线几何面积与建筑层数乘积之总和与地块闭合多段线几何面积的比值。
机器学习训练模块，以不同规模城市用地性质标准数据作为学习标签，采用有监督分类学习算法，对数据库建构模块得到的包含加权业态特征、形态特征的不同规模案例城市数据库进行机器学习训练，产生多个对应不同城市规模的机器学习分类模型，将其组合形成城市用地自动识别模型簇。
通过调用数据库建构模块获得由数据获取及输入模块采集得到的不同规模案例城市数据库中典型城市的特征数据，并将其作为训练数据。根据不同的城市规模分别训练独立的模型，以适应各类城市识别需要。通过按比例抽样将训练数据切分为训练集、验证集和测试集，其中模型输入为地块的加权业态特征和建筑形态特征，输出为其可能的用地性质类别，标签由实地测量得到的用地性质提供。
所述有监督学习分类算法常见的有：逻辑回归、树模型、支持向量机、集成模型等。由于考虑到地块标签按照小类进行划分，类别数较多，且地块的特征维度较高，既包含了加权的业态特征维度也包含了地块的建筑形态特征维度，为了提高模型的分类准确度， 这里建议采用优化的集成树模型来进行分类，如梯度提升决策树模型(GBDT)、xgboost模型等，在参数优化的过程中，根据验证集的性能，逐步调整节点数量、最大深度、学习率等参数，使得在验证集上的性能达到同等情况下的最优同时，采用K折交叉验证法进行交叉验证，避免模型的过拟合。
自动识别模块，根据数据库建构模块中生成的包含加权业态特征、形态特征的目标城市数据库，采用训练完成的城市用地自动识别模型簇识别每个地块的用地性质及置信度。
所述自动识别模块调用机器学习训练模块中已经训练完成的模型簇，该模型簇对应了各类城市规模，自动识别模块只需要设置参数确定目标城市区域的规模大小，即可自动调用相应分类模型。
当系统需要对某目标城市区域进行自动识别用地功能时，首先根据目标城市区域的城市规模选择相应分类模型，然后通过调用数据库得到采集到的目标城市地块的业态特征和建筑形态特征，作为特征向量输入到该分类模型中，最后生成该目标城市区域各地块的可能的用地性质类别，并给出分类置信度。
数据输出模块，将用地性质对应不同色块录入地块矢量文件中并标注其置信度，将成果打印成图纸得到城市现状用地性质图。
在得出每个地块对应的用地性质后，以小类用地性质，通常采用城市用地分类与规划建设用地标准(GB50137-2011)中的城市建设用地分类中的8大类、35中类、42小类标准；亦可采用各地区、市的城市分类用地标准为精度，对同一类用地性质的地块进行一种颜色的填充，并自动标注每个用地识别结果的置信度。通过全彩色喷墨打印机将识别区域的城市用地识别结果以着色块的平面矢量图像进行打印并展示。
利用本发明实施例的综合业态大数据与建筑形态的城市用地自动识别系统，能够对不同规模的城市用地性质进行高效、自动化、精细的识别，进而生成城市用地识别结果并给出相对应置信度，避免了传统人工现场测绘中容易出现的耗时长、人力投入大、易判断失误等问题，为城市现状建设用地的测度和绘制提供了高效便捷的测绘途径及参考，有效节约人工测绘的时间成本。
实施例
以下将以深圳市城市用地自动识别为例对本发明的技术方案进行详细说明。
(1)以上海(超大规模城市)、南京(特大规模城市)、哈尔滨(大规模城市)、镇 江(中等规模城市)、张家港(小规模城市)作为五类不同规模案例城市，并以深圳作为目标城市，数据获取及输入模块分别获取各城市建成区域内的空间矢量数据、业态点数据，并录入空间矢量平台；具体包括：
(1.1)通过各城市对应规划部门，获得以上城市的空间矢量数据，包含每个城市的现状闭合街区CAD文件、现状闭合用地地块CAD文件、现状闭合建筑及层数CAD文件。
(1.2)将空间矢量数据中的现状闭合街区CAD文件、现状闭合用地地块CAD文件导入ArcGIS软件，并导出闭合多段线(Polyline)的SHP格式；将现状闭合建筑及层数CAD文件导入ArcGIS软件，并导出建筑闭合面(Polyline)的SHP格式及层数点(Point)的SHP格式；将建筑闭合面与建筑层数点进行空间关联，将每个建筑附上其层数信息。
(1.3)通过百度坐标拾取器，获取每个城市建成区域西北角及东南角的坐标数据，通过Python编程，获取对应区域内的业态信息XLS文件。
(1.4)将业态信息XLS文件导出CSV格式，在ArcGIS软件中导入XY数据(Add XY data)，并导出SHP格式，得到包含业态点名称、地理坐标及业态特征类型信息的业态点数据。
(2)数据库建构模块对城市的地块单元进行编号，将业态点进行空间关联和校准并统计每个地块内的各类不同特征的业态点数量，生成关联业态点的地块数据库。对业态点特征进行加权，再对地块内的各类建筑形态数据进行运算，生成包含加权业态特征和建筑形态特征的地块特征属性表，生成含加权业态特征、形态特征的地块数据库；具体包括：
(2.1)对由地块构成的空间单元进行编号i：1,2，···,|D|，对每个地块单元在空间上进行膨胀操作，即在地块边界的基础上向外延展数米。
(2.2)如图2所示，对业态点图层和膨胀后的地块单元图层进行空间连接，统计得到每个地块空间单元所拥有的K类业态点数量，即每个地块的业态特征频数,生成关联业态点的地块属性表及数据库。本系统采用业态特征编号为1,2，··,j，··,20，各个业态特征如表1所示：
表1业态特征类别表
公司企业 购物 美食 出入口 房地产
生活服务 道路 交通设施 丽人 政府机构
汽车服务 医疗 金融 休闲娱乐 酒店
运动健身 教育培训 旅游景点 文化传媒 自然地物
(2.3)对完成的关联业态点的地块数据库进行TF-IDF算法处理，所述TF-IDF算法具体如下公式所示：
Figure PCTCN2020071915-appb-000003
其中，i为地块编号，j为业态特征编号，n i,j第i个地块的第j个业态特征的频数，K为业态特征的维数，|D|为城市中的地块总数，{j:t i∈D j}为第j个业态特征频数不为0的地块数量；
通过这样的计算，可以对各个业态特征根据其出现的次数和重要性重新加权，得到包含加权的业态特征的地块特征属性表；
(2.4)对建筑图层(包含层数信息、轮廓为闭合的多边形建筑)与地块单元图层进行空间连接，对每个地块的四项建筑形态指标进行计算并添加到包含加权业态特征的地块特征属性表中，生成并存储包含加权的业态特征、建筑形态特征的地块属性表及数据库。建筑形态特征指标如表2所示：
表2建筑形态特征指标表
Figure PCTCN2020071915-appb-000004
Figure PCTCN2020071915-appb-000005
其中，Hmax为地块最高高度指标，Have为地块平均高度指标里，Save为建筑平均基底面积指标，FAR为容积率指标，A为建筑基底面积，F为建筑层数，B为地块面积，n为地块内建筑总数。
(3)机器学习训练模块向数据获取及输入模块获取上海、南京、哈尔滨、镇江、张家港这五个城市数据，对其进行数据集采样、切分，进行模型选择和训练，通过参数优化和交叉验证，得到最终模型，将不同城市规模数据所训练得到的多个模型组合成模型簇。
下面仅以代表特大城市的深圳市为例，(3)具体包括：
(3.1)机器学习训练模块从数据库建构模块中抽取包含业态特征和建筑形态特征以及用地性质标签的典型城市地块数据，每个典型城市代表一种城市规模；
(3.2)将地块特征属性表等比例采样，按照6:2:2的比例切分为训练集、验证集和测试集，采用GBDT、SVM、LASSO等模型进行模型训练，根据其分类准确率确定合适的模型，模型性能比较如图3所示。在确定选用的模型后，对模型进行参数优化，这里以GBDT算法最大深度max_depth为例，如图4所示，选择合适的模型超参数可以进一步优化分类性能；
(3.3)在将各个城市规模的典型城市分别进行模型训练后，得到适应各类城市规模的最终模型簇，模型训练完毕；
(4)自动识别模块对待识别目标城市区域深圳市进行用地性质自动识别，从数据库建构模块中生成的包含业态特征和建筑形态特征的目标城市数据库中，获取其业态特征和建筑形态特征，根据该城市所拥有的人口确定其城市规模，在模型簇中选取同为超大规模城市的上海城市用地模型，输入深圳地块数据，生成并输出深圳市各个地块的可能的用地性质类别表，并给出分类置信度；
(5)数据输出模块将输出的目标城市用地性质对应不同色块录入地块矢量文件中并标注其置信度，将成果打印成图纸得到城市用地图，具体包括：
(5.1)以小类用地性质通常采用城市用地分类与规划建设用地标准(GB50137-2011)中的城市建设用地分类中的8大类、35中类、42小类标准；亦可采用各地区、市的城 市分类用地标准为精度，对同一类用地性质的地块进行一种颜色的填充，并自动标注每个用地识别结果的置信度。
(5.2)通过全彩色喷墨打印机将识别区域的城市用地识别结果以着色块的平面矢量图像进行打印并展示，如图5.1所示，图5.2为局部放大图。

Claims (10)

  1. 一种综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于，该系统包括：
    数据获取及输入模块，用于获取并存储不同规模案例城市及目标城市建成区域内的空间矢量数据、业态点数据，输入地理信息系统，其中，所述不同规模案例城市按照国务院印发的最新城市规模划分标准划分，选取超大城市、特大城市、大城市、中等城市、小城市五类规模的案例城市，每类城市对应生成一个数据库及机器学习分类模型；所述业态点数据包括业态点名称、地理坐标及业态特征类型信息；其中，所述业态特征类型信息为业态点所属行业信息或业态点职能信息或业态点分类信息；
    数据库建构模块，用于通过地理信息系统，按照设定的空间自动校准方法对数据进行校准并关联，得到关联业态点的地块数据库，根据业态特征重要性对关联业态点的地块数据库进行加权处理，按照设定的建筑形态特征指标自动计算出每个地块内所有建筑的最高高度、平均高度、建筑平均基底面积、容积率指标，并与地块关联，生成并存储包含加权业态特征、形态特征的地块数据库，所述地块数据库包含目标城市数据库和不同规模案例城市数据库；
    机器学习训练模块，用于输入不同规模案例城市的用地性质标准数据，所述用地性质标准数据按照用地小类划分，并作为机器学习标签；采用有监督分类学习算法，对数据库建构模块得到的包含加权业态特征、形态特征的不同规模案例城市数据库进行机器学习训练，产生多个对应不同城市规模的机器学习分类模型，将其组合形成城市用地自动识别模型簇；
    自动识别模块，用于根据数据库建构模块中生成的包含加权业态特征、形态特征的目标城市数据库，采用训练完成的城市用地自动识别模型簇，自动识别目标城市建成区域内每个地块对应的城市用地性质及其置信度；
    数据输出模块，用于将用地性质按小类对应不同色块录入地块矢量文件中，并标注数据置信度，得到城市现状用地性质图。
  2. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于：所述城市建成区域内的空间矢量数据包括轮廓为闭合的多边形街区；其中，所述街区内包含一个以上的轮廓为闭合的多边形地块，所述地块内包含一个以上的轮廓为闭合的多边形建筑，所述建筑具备层数或高度信息。
  3. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于：所述空间自动校准方法，包括业态点与城市街区的空间校准、业态点与用 地地块的空间校准、业态点与所在用地地块及城市街区的属性关联，先对城市地块进行膨胀操作，再与业态点数据进行空间连接。
  4. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于：所述地理信息系统用于矢量数据处理，包括：ArcGIS、CAD。
  5. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于，所述对关联业态点的地块数据库进行加权处理的实现方式为按照如下公式对完成的关联业态点的地块数据库进行TF-IDF算法处理：
    Figure PCTCN2020071915-appb-100001
    其中，i为地块编号，j为业态特征编号，n i,j第i个地块的第j个业态特征的频数，K为业态特征的维数，|D|为城市中的地块总数，{j:t i∈D j}为第j个业态特征频数不为0的地块数量。
  6. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于：所述地块内所有建筑的最高高度指标为每个地块内所有建筑高度的最大值；所述地块内所有建筑的平均高度指标为每个地块内所有建筑高度的平均值；所述地块内所有建筑的平均基底面积指标为每个地块内所有建筑闭合多段线的几何面积的平均值；所述地块内所有建筑的容积率指标为每个地块内每个建筑闭合多段线几何面积与建筑层数乘积之总和与地块闭合多段线几何面积的比值。
  7. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其步骤特征在于：所述有监督分类学习算法是根据不同城市规模，调用数据库建构模块中包含加权业态特征、形态特征的不同规模案例城市数据库，根据其中典型城市的实测用地性质数据集，按照城市规模分别进行模型的训练和参数优化。
  8. 根据权利要求6所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于：所述实测用地性质数据集，按比例抽样切分为训练集、验证集和测试集，通过交叉验证、参数优化、泛化检验选取分类性能与泛化性能均优的机器学习模型作为最终模型。
  9. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于：所述自动识别模块，在选择目标城市区域后，获取并输入数据获取及输入模块中的目标城市建成区域内的空间矢量数据、业态点数据，利用数据库建构模块得到的包含业态特征、形态特征的目标城市数据库，再根据其城市规模选择对应的机器学习 模型进行分类，输出该城市每个地块的可能的用地性质类别，并给出其分类置信度。
  10. 根据权利要求1所述的综合业态大数据与建筑形态的城市用地自动识别系统，其特征在于：所述数据输出模块，获取自动识别模块输出的每个地块对应的用地性质，以小类用地性质为精度，对同一类用地性质的地块进行一种颜色的填充，将识别区域的城市用地识别结果以着色块的城市现状用地性质图进行显示。
PCT/CN2020/071915 2019-01-25 2020-01-14 一种综合业态大数据与建筑形态的城市用地自动识别系统 WO2020151528A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US16/963,506 US11270397B2 (en) 2019-01-25 2020-01-14 Automatic urban land identification system integrating business big data with building form
EP20745903.3A EP3916668A4 (en) 2019-01-25 2020-01-14 SYSTEM FOR AUTOMATIC IDENTIFICATION OF URBAN LAND WITH INTEGRATION OF INDUSTRIAL LARGE DATA AND BUILDING SHAPE

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910072439.8 2019-01-25
CN201910072439.8A CN109816581A (zh) 2019-01-25 2019-01-25 一种综合业态大数据与建筑形态的城市用地自动识别系统

Publications (1)

Publication Number Publication Date
WO2020151528A1 true WO2020151528A1 (zh) 2020-07-30

Family

ID=66605059

Family Applications (2)

Application Number Title Priority Date Filing Date
PCT/CN2019/081818 WO2020151089A1 (zh) 2019-01-25 2019-04-08 一种综合业态大数据与建筑形态的城市用地自动识别系统
PCT/CN2020/071915 WO2020151528A1 (zh) 2019-01-25 2020-01-14 一种综合业态大数据与建筑形态的城市用地自动识别系统

Family Applications Before (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/081818 WO2020151089A1 (zh) 2019-01-25 2019-04-08 一种综合业态大数据与建筑形态的城市用地自动识别系统

Country Status (4)

Country Link
US (1) US11270397B2 (zh)
EP (1) EP3916668A4 (zh)
CN (1) CN109816581A (zh)
WO (2) WO2020151089A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112967286A (zh) * 2021-05-19 2021-06-15 航天宏图信息技术股份有限公司 一种新增建设用地检测方法及装置

Families Citing this family (36)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109816581A (zh) 2019-01-25 2019-05-28 东南大学 一种综合业态大数据与建筑形态的城市用地自动识别系统
CN110288134A (zh) * 2019-06-06 2019-09-27 国网湖北省电力有限公司孝感供电公司 一种城市分类用地面积快速自动统计方法
CN111127258B (zh) * 2019-12-26 2023-05-26 深圳集智数字科技有限公司 用于确定具有多业态的地块中业态容积率的方法及装置
CN111914328B (zh) * 2020-07-31 2023-11-17 同济大学 一种基于人工智能的建筑平面辅助设计方法及系统
CN112085357B (zh) * 2020-08-27 2021-06-04 东南大学 一种出让地块规划条件冲突要点识别与处理的系统与方法
CN112115641A (zh) * 2020-09-11 2020-12-22 同济大学 一种智能城市信息基础设施规划系统
CN112666588B (zh) * 2020-11-06 2021-10-22 南京航空航天大学 一种城市峡谷环境下基于景象匹配与机器学习的定位方法
CN112508336B (zh) * 2020-11-09 2023-09-19 东南大学 一种基于结构方程模型的空间与环境效能关联测度方法
CN112597948B (zh) * 2020-12-29 2023-03-28 同济大学 一种城市土地利用变化预测方法
CN112766718A (zh) * 2021-01-18 2021-05-07 华南理工大学 城市商圈边界识别方法、系统、计算机设备及存储介质
CN112884208A (zh) * 2021-01-26 2021-06-01 中国测绘科学研究院 一种城市街区智能划分方法
CN112926857A (zh) * 2021-02-25 2021-06-08 重庆市规划设计研究院 一种高效城市规划容积率自动校正系统及方法
CN113033484B (zh) * 2021-04-21 2022-11-22 河北工程大学 一种面向无人机应急网络部署的城市分类方法
CN112990605B (zh) * 2021-04-22 2021-07-30 武汉市规划研究院 一种全流程线上规划编制的方法和装置
CN113672788B (zh) * 2021-07-22 2024-04-09 东南大学 一种基于多源数据和权重系数法的城市建筑功能分类方法
CN113610165B (zh) * 2021-08-10 2024-02-13 河南大学 基于多源高维特征的城市土地利用分类确定方法和系统
CN113642902A (zh) * 2021-08-17 2021-11-12 上海图源素数字科技有限公司 一种国土空间多元关联信息的处理方法
CN113449936B (zh) * 2021-08-31 2022-03-18 北京市城市规划设计研究院 城市空间演进模拟预测方法、装置、电子设备及存储介质
CN113706997B (zh) * 2021-09-06 2023-06-27 深圳市指跃未来科技有限公司 城乡规划图纸标准化处理方法、装置及电子设备
CN113935549B (zh) * 2021-11-22 2022-07-19 国家电投集团雄安能源有限公司 一种综合智慧能源优化调度系统
CN114139827B (zh) * 2021-12-09 2024-02-09 北京建筑大学 一种城市功能区功能绩效的智能感知与优化方法
CN114218640B (zh) * 2021-12-15 2023-10-31 东南大学 一种基于人工智能的假山体量设计方法
CN114461869B (zh) * 2021-12-21 2022-11-22 北京达佳互联信息技术有限公司 业务特征数据处理方法、装置、电子设备及存储介质
CN114089033B (zh) * 2022-01-24 2022-04-26 天津安力信通讯科技有限公司 一种基于频谱分析的异常信号检测方法及系统
CN114639027B (zh) * 2022-05-20 2023-02-03 山东省地质科学研究院 根据土地利用分类数据对城镇低效用地识别的系统及方法
CN114969007A (zh) * 2022-06-01 2022-08-30 南京大学 一种基于功能混合度和集成学习的城市功能区识别方法
CN115238584B (zh) * 2022-07-29 2023-07-11 湖南大学 一种基于多源大数据的人口分布识别方法
CN115292789B (zh) * 2022-08-12 2023-04-28 东南大学建筑设计研究院有限公司 城市设计中基于形态类型的建筑体量数字化生成方法
CN115859765B (zh) * 2022-09-29 2023-12-08 中山大学 城市扩张的预测方法、装置、设备及存储介质
CN115620165B (zh) * 2022-11-07 2023-03-28 中国科学院空天信息创新研究院 城市建成区慢行系统设施评价方法、装置、设备及介质
CN115578643B (zh) * 2022-12-06 2023-02-17 东莞先知大数据有限公司 一种农田区域建筑检测方法、电子设备和存储介质
CN116129202B (zh) * 2023-04-20 2023-06-09 北京新兴科遥信息技术有限公司 存量用地分析方法、装置和存储介质
CN116578564B (zh) * 2023-05-13 2024-05-07 杭州市勘测设计研究院有限公司 一种多测合一生产质检一体化应用服务系统
CN117077005B (zh) * 2023-08-21 2024-05-10 广东国地规划科技股份有限公司 一种城市微更新潜力的优化方法和系统
CN117196135A (zh) * 2023-08-24 2023-12-08 北京市市政工程设计研究总院有限公司 一种海绵城市中径流控制率的调整方法及装置
CN117454253B (zh) * 2023-12-08 2024-04-02 深圳市蕾奥规划设计咨询股份有限公司 建筑的分类方法、装置、终端设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521273B (zh) * 2011-11-23 2013-06-19 中国科学院地理科学与资源研究所 一种高分辨率遥感的多功能城市用地空间信息生成方法
CN105893544A (zh) * 2016-03-31 2016-08-24 东南大学 一种基于poi业态数据生成城市空间大数据地图的方法
CN106897417A (zh) * 2017-02-22 2017-06-27 东南大学 一种基于多源大数据融合的城市空间全息地图的构建方法
CN109816581A (zh) * 2019-01-25 2019-05-28 东南大学 一种综合业态大数据与建筑形态的城市用地自动识别系统

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7970642B2 (en) * 2005-12-27 2011-06-28 Alex Anas Computer based system to generate data for implementing regional and metropolitan economic, land use and transportation planning
US8117138B2 (en) * 2008-03-11 2012-02-14 International Business Machines Corporation Method and apparatus for location evaluation and site selection
US9535927B2 (en) * 2013-06-24 2017-01-03 Great-Circle Technologies, Inc. Method and apparatus for situational context for big data
CN106682800A (zh) * 2015-11-10 2017-05-17 星际空间（天津）科技发展有限公司 一种建设项目快速选址方法
CN105912764B (zh) * 2016-04-06 2019-02-05 东南大学 一种基于噪声分区模型的城市容积率优化方法
US20190066137A1 (en) * 2017-04-24 2019-02-28 The Chicago TREND Corporation Systems and methods for modeling impact of commercial development on a geographic area
CN107679229B (zh) * 2017-10-20 2021-06-01 东南大学 城市三维建筑高精度空间大数据的综合采集及分析方法
CN108052887A (zh) * 2017-12-07 2018-05-18 东南大学 一种融合slam/gnss信息的疑似违法用地自动识别系统及方法
AU2019101466A4 (en) * 2019-11-27 2020-01-16 Henan University Method for determining future urban expansion mode

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521273B (zh) * 2011-11-23 2013-06-19 中国科学院地理科学与资源研究所 一种高分辨率遥感的多功能城市用地空间信息生成方法
CN105893544A (zh) * 2016-03-31 2016-08-24 东南大学 一种基于poi业态数据生成城市空间大数据地图的方法
CN106897417A (zh) * 2017-02-22 2017-06-27 东南大学 一种基于多源大数据融合的城市空间全息地图的构建方法
CN109816581A (zh) * 2019-01-25 2019-05-28 东南大学 一种综合业态大数据与建筑形态的城市用地自动识别系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3916668A4 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112967286A (zh) * 2021-05-19 2021-06-15 航天宏图信息技术股份有限公司 一种新增建设用地检测方法及装置

Also Published As

Publication number Publication date
EP3916668A1 (en) 2021-12-01
CN109816581A (zh) 2019-05-28
EP3916668A4 (en) 2022-10-19
WO2020151089A1 (zh) 2020-07-30
US11270397B2 (en) 2022-03-08
US20210217117A1 (en) 2021-07-15

Similar Documents

Publication Publication Date Title
WO2020151528A1 (zh) 一种综合业态大数据与建筑形态的城市用地自动识别系统
WO2020233152A1 (zh) 基于城市建筑空间数据的建成区边界识别方法及设备
WO2017166370A1 (zh) 一种基于区域城际流强度测算模型的划定大都市圈的方法
CN110533038A (zh) 一种基于信息数据的城市活力区和中心城区边界识别的方法
CN110188228A (zh) 基于草图检索三维模型的跨模态检索方法
CN108153894A (zh) 一种olap数据模型自动建模的方法、分类器
WO2020073430A1 (zh) 一种城市空间形态自动分区方法与系统
CN115757604B (zh) 一种基于夜光影像数据的gdp时空演变分析方法
WO2023029678A1 (zh) 基于gis的农业服务管理方法及系统
CN109147322A (zh) 一种城市交通大数据处理中多源数据自适应融合方法
CN106021499A (zh) 基于志愿者地理信息的建设用地分类方法和装置
CN106202762A (zh) 一种基于ArcGIS工具的用户水量数据自动导入建模软件方法
CN113240209A (zh) 一种基于图神经网络的城市产业集群发展路径预测方法
CN116415499B (zh) 一种社区舒适感模拟预测方法
CN109509254A (zh) 三维地图构建方法、装置及存储介质
CN114170441B (zh) 基于地理国情数据和影像分类的路旁树自动提取方法
CN116151686A (zh) 一种科技企业孵化器的孵化效益评价方法
CN107220615B (zh) 一种融合兴趣点大数据的城市不透水面信息提取方法
CN107563327A (zh) 一种基于自步反馈的行人重识别方法及系统
Qin et al. Urban Functional Zone Identification by Considering the Heterogeneous Distribution of Points of Interests
CN114612800A (zh) 一种核算城市群建筑物质存量及时空变化的方法、系统
CN113569946A (zh) 开源地图与专业数据源路网自适应匹配法
CN106250828A (zh) 一种基于改进的lbp算子的人群计数方法
Wang et al. Classification of Rural Tourism Features Based on Hierarchical Clustering Analysis Knowledge Recognition Algorithm
CN105447875A (zh) 一种电子地形图自动几何校正方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20745903

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020745903

Country of ref document: EP

Effective date: 20210825