Study Of Spatial Morphology Types And Influencing Factors Of Road Networks In Urban Built-up Areas Based On Image Recognition

With the frequent flow and gathering of population,material and other factors,urbanization in China has entered a rapid development stage.This has led to unchecked urban spatial expansion while promoting economic growth and boosting urban competitiveness,which is a challenging issue to address at this stage of new urbanization in China.Optimizing the carrying capacity and sustainability of the constrained urban area,as well as limiting the unchecked growth of urban space while respecting natural laws,are the solutions to the issue.The urban land use pattern is largely determined by the road network,which also has a complex mutual influence and interlocking interaction with urban growth as the foundation of the urban landscape corridor and the regular operation of the city.In order to further explore the similarities and differences between the spatial pattern categories of built-up areas and road networks,this paper constructs a model of the spatial pattern types,spatial distribution characteristics,and influencing factors of built-up areas in 110 cities across the country using image recognition and machine learning techniques.The main findings of the study are as follows:(1)With significant variation in the number and structural traits of the road networks,the results of the hierarchical clustering of urban road network images exhibit distinct boundaries and close internal connections.Large scale,low density,and unbalanced characteristics characterize the road network images of city C1,whereas small scale,low density,and clear texture characterize those of city C2,and small scale,high density,and balanced distribution characterize those of city C3.The built-up areas of C1 cities are diverse and mixed;those of C2 cities have simple internal structures and obvious zoning,and they are in a state of spatial expansion;those of C3 cities are clearly zoned,have more developed spatial forms,and have higher spatial utilization.In addition,the morphological characteristics of the corresponding built-up areas are also noticeably different.(2)Highly territorial and geographically diverse,the built-up area’s spatial structure and the road system are both.The built-up areas of C1 cities are more impacted by the terrain since they are concentrated in the hilly regions of the south-east and south-west.The area where C2 cities are concentrated is a triangle structure created by the middle section,the north-eastern coast,and the Hu Huanyong line,and the plain terrain offers a benefit for the construction and extension of the spatial structure of the built-up areas.The(3)Multi-scale influence factor model shows that the urbanization rate,GDP per capita,mean slope,built-up area road network density,and the amount of land-average fixed asset investment are the main factors that influence and can distinguish the spatial pattern of built-up areas and the road network structure of C1 and C3 cities,while the mean slope,population density,city class,and land-average fixed asset investment are the main factors that influence C2 cities.(4)The prediction accuracy of learning based on the existing classification results and influencing factors can reach 70%,and some of the cities clustered in C1 and C2 are difficult to distinguish and have a higher misjudgment rate.Further observation combined with the built-up area and road network morphology reveals that these cities do not have obvious characteristics in factors such as population density,road density and urbanization rate leading to misjudgment.