The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLIII-B4-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIII-B4-2022, 537–543, 2022
https://doi.org/10.5194/isprs-archives-XLIII-B4-2022-537-2022
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIII-B4-2022, 537–543, 2022
https://doi.org/10.5194/isprs-archives-XLIII-B4-2022-537-2022
 
02 Jun 2022
02 Jun 2022

ANALYSIS THE INFLUENCING FACTORS OF URBAN TRAFFIC FLOWS BY USING NEW AND EMERGING URBAN BIG DATA AND DEEP LEARNING

Y. Li1, Q. Zhao1, and M. Wang2 Y. Li et al.
  • 1Urban Big Data Centre, School of Social and Political Sciences, University of Glasgow, Glasgow G12 8RZ, UK
  • 2School of Geographical and Earth Sciences, University of Glasgow, Glasgow G12 8RZ, UK

Keywords: Traffic Flow, Urban Big Data, Spatial Analysis, Deep Learning, Computer Vision

Abstract. Urban traffic analysis has acted an important role in the process of urban development, which can provide insights for urban planning, traffic management and resource allocation. Meanwhile, the advancement of Intelligent Transportation Systems has produced a variety of traffic-related data from sensors and cameras to monitor urban traffic conditions in high spatio-temporal resolution. This research applies spatial regression models combined with computer vision and deep learning to analyse traffic flow distributions via various factors in the urban areas and traffic flow data. We include road characteristics and surrounding environments such as land use/cover, nearby points of interest (POI) and Google Street View images. The results show that the daily average traffic flow on main roads is much higher than smaller roads, and nearby POIs numbers have positive effect on traffic flows. The impact of land cover type is insignificant in the linear regression model, while demonstrates significant contribution to traffic flows in spatial regression models. Although the spatial autocorrelation still exists after the spatial regression, the spatial error model generates a better fit on the dataset. Further analysis will focus on extend the current model with the time parameters and understand what influence the changes of traffic flow in the different spatio-temporal scales.