Detecting New Pedestrian Facilities from VGI Data Sources

Pedestrian facility (e.g. footbridge, pedestrian crossing and underground passage) information is an important basic data of location based service (LBS) for pedestrians. However, timely updating pedestrian facility information challenges due to facilities change frequently. Previous pedestrian facility information collecting and updating tasks are mainly completed by highly trained specialized persons. However, this conventional approach has several disadvantages such as high cost, long update cycle and so on. Volunteered Geographic Information (VGI) has proven efficiency to provide new, free and fast growing spatial data. Pedestrian trajectory, which can be seen as measurements of real pedestrian road, is one of the most valuable information of VGI data. Although the accuracy of the trajectories is not too high, due to the large number of measurements, an improvement of quality of the road information can be achieved. Thus, we develop a method for detecting new pedestrian facilities based on the current road network and pedestrian trajectories. Specifically, 1) by analyzing speed, distance and direction, those outliers of pedestrian trajectories are removed, 2) a road network matching algorithm is developed for eliminating redundant trajectories, and 3) a space-time cluster algorithm is adopted for detecting new walking facilities. The performance of the method is evaluated with a series of experiments conducted on a part of the road network of Heifei and a large number of real pedestrian trajectories, and verified the results by using Tencent Street map. The results show that the proposed method is able to detecting new pedestrian facilities from VGI data accurately. We believe that the proposed method provides an alternative way for general road data acquisition, and can improve the quality of LBS for pedestrians.