PEDESTRIAN COLLISION WARNING SYSTEMS USING NEURAL NETWORKS BASED ON A SINGLE CAMERA

摘要

This paper proposes a method of achieving fast detection of pedestrians, while simultaneouslyrnmaintaining good performance regardless of variation in illumination, and in both shape andrnscale of pedestrians with a single camera. Regions of interest (ROIs) are acquired by opticalrnflow fields using the Lucas-Kanade algorithm, and classified by convolutional neuralrnnetworks (CNNs) whether they are pedestrians. Detected pedestrians are tracked by using arnparticle filter based on adaptive fusion frameworks. The CNNs allow the proposed system tornbe robust to variation in illumination and in both shape and scale of pedestrians; and proposedrnmethods of setting ROIs and tracking pedestrians allow this system to detect a dangerousrnsituation and warn it to a driver fast. A single camera is only used to conduct this method, thus,rnthe proposed system is also economically efficient.