To evaluate the sustainability of an enormous number of urban intersections, a novel assessment model is proposed, along with an indicator system and corresponding methods to determine the indicators. Considering mainly the demands and feelings of the urban residents, the three aspects of safety, functionality, and image perception are taken into account in the indicator system. Based on technologies such as street view picture crawling, image segmentation, and edge detection, GIS spatial data analysis, a rapid automated assessment method, and a corresponding multi-source database are built up to determine the indicators. The improved information entropy method is applied to obtain the entropy weights of each indicator. A case study shows the efficiency and applicability of the proposed assessment model, indicator system and algorithm.
Internet open location-based service (LBS) is an emerging Internet service that can provide users with a wide range of functions derived from location data. Using Internet open LBS and computer vision algorithms, this paper proposes a new approach to the large-scale automated extraction of built environment characteristics related to pedestrian safety at urban intersections. Through the open LBS, the longitude and latitude position of urban intersections are first acquired and then the street view images of each intersection are downloaded. Hough transform algorithm is applied to the street view images to extract one of the main characteristics of the built environment related to pedestrian safety: crosswalk presence. A case study in Shenzhen, China, proves the applicability and efficiency of the proposed approach.