Using re-identification data and spatial-temporal analysis for pedestrian origin-destination and travel time estimation

The traditional ways to derive pedestrian Origin-Destination(OD) information are conducting interview and/or questionnaire surveys. However, the traditional approaches involve tremendous time-consuming process, high investment and unable to be performed in real-time basis. Recently, image processing technique have been developed to recognize individual pedestrian based on their face. The face recognition technique could violate personal privacy in some circumstances. Therefore, this study aims to propose an alternative approach for OD and walking time estimation based on pedestrian re-identification. Pedestrian feature extraction and spatial-temporal analysis are adopted for pedestrian re-identification purpose, instead of using face recognition techniques. The pedestrian features including color, texture and shape can be considered as a feature vector which can be used for pedestrian re-identification between OD pairs. Moreover, the results of pedestrian re-identification can be used for estimating travel time between each OD pair. In this study, a pedestrian walkway is used as a case study for pedestrian re-identification. Two cameras are installed at both ends of the walkway to identify an individual pedestrian. Probability-based approaches will be adopted to match the individual pedestrian based on the pedestrian features at two locations. This paper will demonstrate the accuracy of pedestrian re-identification and travel time estimation.