Title :
Detection of pedestrians from still images
Author :
Tetik, Yusuf Engin ; Bolat, Bülent
Author_Institution :
Elektron. ve Haberlesme Muhendisligi Bolumu, Yildiz Teknik Univ., Istanbul, Turkey
Abstract :
In this work, a pedestrian detection system based on still images taken from monocular cameras is proposed. This framework aims to distinctively find each of the sub-objects that defines the main object, and then detects the complete main object by exploiting these sub-objects. In other words, this framework utilizes bottom-to-top approach. Modeling of the objects is realized by using approach of ratio-templates.
Keywords :
object detection; bottom-to-top approach; monocular cameras; pedestrian detection system; ratio-template approach; still images; Computer vision; Conferences; Levee; Object detection; Object recognition; Pattern recognition; Signal processing;
Conference_Titel :
Signal Processing and Communications Applications (SIU), 2011 IEEE 19th Conference on
Conference_Location :
Antalya
Print_ISBN :
978-1-4577-0462-8
Electronic_ISBN :
978-1-4577-0461-1
DOI :
10.1109/SIU.2011.5929739
