Paper Info
Title
A Fast and Accurate Forward Vehicle Start Alarm by Tracking Moving Edges Obtained from Dashboard Camera
Abstract
The Forward Vehicle Start Alarm (FVSA) is a technique for determining whether or not a forward vehicle begins to start when the traffic signal changes from STOP to GO. And it is expected to be one of the important functions of Advanced Driver Assistance System (ADAS). In this paper, we implement the FVSA function by using only dashboard camera. We consider the moving edges which are obtained from the frame difference as a key feature to detect the movement of a forward vehicle. We project the moving edges horizontally to find the positions of the dominant edge in the Region of Interest (ROI) and create the edge profile by temporally accumulating them. By examining the change pattern of edge profile, we can successfully track whether the forward vehicle is departing. Through simulation, we confirmed that the proposed method works reliably in various environments and showed that it can solve the typical case of the malfunction in existing commercial ADAS products
Year
DOI
Venue
2018
10.1109/AVSS.2018.8639162
2018 15th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)
Keywords
Field
DocType
Image edge detection,Feature extraction,Tracking,Cameras,Advanced driver assistance systems,Machine learning algorithms,Reliability
Computer vision,Advanced driver,Traffic signal,Computer science,ALARM,Advanced driver assistance systems,Frame difference,Feature extraction,Artificial intelligence,Region of interest,Dashboard (business)
Conference
ISBN
Citations 
PageRank 
978-1-5386-9294-3
1
0.35
References 
Authors
0
2
Name
Order
Citations
PageRank
Kang Yi1466.66
Kyeong-hoon Jung2187.03