Research On The Change Of Urban Street Space Quality Based On Street View Image Recognition

The evaluation of urban street space quality changes and the study of quality improvement have always been hot topics in the academic circles.In recent years,with the continuous advancement of refined urban management,urban physical examinations and research on urban issues based on big data have received increasing attention.Street view image data with high information accuracy,wide coverage,and low cost of collection has become an important data source for the study of urban spatial quality changes.This paper takes the analysis and identification of urban street space quality changes as the fundamental goal,takes street view pictures based on Baidu Street View Map platform as the data source,and selects the street space in Shijingshan District,Beijing as a research case.This paper analyzes the characteristics of urban street space quality changes based on street scene image recognition,summarizes the influencing factors that produce changes in space quality,and then proposes countermeasures to improve urban street space quality.First of all,this study collected streetscape pictures of the same street in the central city of Shijingshan in different ages,and established a database of street spatial quality changes.Secondly,through the semantic segmentation algorithm of computer deep learning,all kinds of objective elements in the street view pictures are identified,and the evaluation system of street space quality change is established based on five indexes,namely green looking ratio,motorization rate,enclosure degree,sky openness and walkability.Thirdly,according to the evaluation system,perform a refined quantitative analysis of various evaluation indicators,compare the changes of various indicators longitudinally,dynamically identify changes in street space quality and analyze influencing factors,and summarize the overall changes in street space quality in Shijingshan District feature.Finally,the streets with significant changes in spatial quality are divided into policy-influenced and micro-renewal categories according to their types.Combined with specific street samples,this paper studies the detailed trend and situation of different types of streets,and discusses the influencing factors that cause various changes,and then specific and targeted spatial quality optimization strategies are put forward.This paper focuses on the applicability of street view image recognition technology,and aims to use new technology to provide new methods and ideas for the study of large-scale changes in the quality of urban street space,and also to provide some reference for the dynamic monitoring and management of street space quality in Shijingshan District and even other cities in China.