Research Papers (2009 – 2013)

Abstract

Over the last decade, substantial technological progress has been achieved in the discipline of computer vision, driven by the inexpensiveness of video sensors and of computer processing. Techniques of particular importance have been developed in the area of automated road user detection and tracking in video sequences. The applications of computer vision techniques in the disciplines of traffic engineering and road safety are numerous and effectively address well- entrenched challenges in these fields. Traditionally, traffic conflict techniques (TCT) have been advocated to empower the weaknesses of collision data by relying on relatively more frequent and costless events. Traffic conflict techniques have been challenged since their inception by observer subjectivity and the costliness of conducting field surveys. The adoption of computer vision techniques in conducting TCT has been advocated by the authors and demonstrated in a number of successful applications. This study demonstrates an extended before-and-after analysis for the evaluation of a pedestrian safety treatment. The results obtained using the automated video analysis system are consistent with previously published results using based on human observers.

Karim Ismail