Anzeige
Erschienen in:
2017 | OriginalPaper | Buchkapitel
Multi-Field Depth Vehicle Headlight Detection by Model Construction and Long Trajectory Extraction in Nighttime City Traffic
verfasst von : Chunming Tang, Yancheng Dong, Xiangqing Lin, Wenna Xiao
Erschienen in: International Symposium for Intelligent Transportation and Smart City (ITASC) 2017 Proceedings
Verlag: Springer Singapore
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Due to the limitation of headlights detection algorithm, obtained vehicles’ tracking trajectories are rather short in existing traffic surveillance systems. A novel vehicle tracking system is proposed in this paper to deal with nighttime traffic surveillance videos. It consists of three parts. An effective headlight detection model is firstly constructed based on the optical imaging principle, noises are then filtered out according to the field depth among the far, middle and near regions by different evaluations. Parallel perspective principle is secondly applied to remove the LED lights disturbance. The headlights are tracked and then paired according to vehicles’ type. Vehicles’ tracking is realized finally via trajectory feedback correction. Experiments are presented to show the proposed system’s superiority over several state-of-the-art methods in headlight detection, pairing and tracking.