Research On The Perceptual Quality Of Urban Street Space Based On Baidu Street Views

Urban streets are the link between various public facilities in the city.The quality of urban street space is a combination of urban physical environment composition and user experience.In this paper,it is embodied in the diversity and comfort of the scene.That is,the perceptual quality of street space.The quality of street space directly affects people's impression for the city and is an important part of evaluating urban quality.Combine the advantages of street view with the algorithmic capabilities of machine learning,this paper applies “transfer learning” and “learning to ranking” techniques to construct a neural network model for the perception of urban street space quality from the perspective of image perception.The model is used to evaluate the distribution of street space quality in the main urban area of Wuhan,and the correlation analysis with road network data and house price data respectively verifies the feasibility of the method.Through this study found:(1)Using “transfer learning” and “learning to ranking” to construct a perception model of street space quality with good precision and simple structure.It also verified that the research of labeling datasets is more conducive to specific problems under the guidance of prior knowledge,and extracting highdimensional image features can effectively improve the accuracy of the model.(2)The quality of street space has spatial autocorrelation in the study area.The center of the area is generally better than the edge of the area,but its internal differences are greater.The quality of street space is related to the level of road function,and is related to the housing price in the geographical space.The model evaluation results are basically consistent with the actual situation in the study area.Although the study constructs a perceptual model that can be used to assess the quality of urban street space,it only applied Baidu Street Views.Later,we will consider how to effectively combine with other data to build a more comprehensive street space quality assessment model.This paper studies the quality of street space by applying new data and new methods,which promotes the development of space quality research to some extent.The assessment of street space quality provides a quantitative method for planners to provide reference and guidance for urban planning and construction.