Research On Urban Functional Semantic Partition Based On Remote Sensing And POI Data

With the urbanization and the promotion of smart city,the meticulous planning of cities meets with new challenges.Most of cities are mainly nagged by traffic congestion,environmental degradation,air pollution and so on,which have resulted in a heavy burden on these cities and have restricted their development.In order to make reasonable planning and to be helpful to urban management,it is very important to identify the spatial structure of the city dynamically.According to the features of agglomeration,the distribution of functions can be recognized and the space layout of the urban can be optimized.Traditional field survey and questionnaire interviews are the most direct way to accurately understand the city functional distribution,however,it faces lots of problems especially in large-scale cities,such as,a heavy workload,a high human and material consumption and not being promptly updated.Therefore,this paper proposes a Semantic Urban Function Zoning Model(SUFZ)which using semantic information mining method of urban function zoning based on the POI(Point of Interest)data,remote sensing images and other spatial data.Based on the extraction of build-up area in high-resolution images by object-oriented approach,combined with the semantic information of POI data,and the spatial influence of POI data from point to area is extended,which further assists remote sensing technology in identifying and analyzing the urban land functions.It is of great significance for the urbanization construction to make clear the division of urban spatial structure and to strengthen the rational planning of urban functional zones.In this paper,the semantic classification results of urban functional areas are obtained by remote sensing technology and the semantic information mining method,based on the GF-1 and GF-2 images,POI data and road network data.Firstly,extraction of construction land in study areas is based on object-oriented method,and the block partition is recognized by using road network data.Considering that the semantic features from POI data can make the classification of urban construction land much finer,it divides POI data into five types of land use,which are industrial land,commercial and business facilities land,residential land,administration and public services land,and street and transportation land.Then it estimates POI data for each category with kernel density analysis and chooses the optimum extraction threshold,which contains the data points more than 95% that stands for the kernel density intensive region.The establishment of experimental rules is good for city construction land preliminary extraction.After that,the evaluation model of function area category,which is according to the point density influence score rules,is established based on the overlapping region of multiple types of kernel density,and thus the function land classification of experimental area is completed.Selecting the sample plots to verify the regional functional matching degree and comparing the results of the urban functional zoning with the regional plots,the results show that the accuracy of the urban functional semantic partition is relatively high.Finally,according to the results of urban functional zoning,the spatial layout rationality analysis model is further set up,and therefore the layout analysis can be accomplished.Generally,the results of the research shows that the number,the space distribution and the semantic hierarchy of POI all have a strong correlation with city function definition,and the definition of city spatial aggregation from the micro analysis view will be affected by many quantitative factors.The method is unrestricted by the regionalism,and the result of urban partition can basically match with the actual urban area,so it is effective for urban function zoning.