Lane-mark Extraction by Frequency-based Saliency Visual Attention
Lane mark extraction and detection system gives vital information for many intelligent transportation systems. There are several active research projects on the system, but none of them is using frequencybased saliency techniques to extract lane regions for Driving Assistant Systems (DASs). In our research, two frequency: based saliency map building techniques, Phase Fourier Transform (PFT) and Quadrature Phase Fourier Transform (QPFT), are implemented in the system and compared with each other according to detection rate. Through the comparison, PQFT, which simultaneously processes on four different channel shows its advantages over PFT, which works on only one intensity feature. The offered methods are tested on real traffic scenes images from UNMC-VIER Autovision database and it could detect discrete lane marks accurately.
lane detection lane extraction frequency-based saliency PFT QPFT saliency map
Anh Cat Le Ngo Li-Minn Ang Kah Phooi Seng Guoping Qiu
The University of Nottingham, Malaysia Campus Electrical and Electronic Department Jalan Broga, 43500 Semenyih, Selangor Darul Ehsan
国际会议
成都
英文
55-59
2010-07-07（万方平台首次上网日期，不代表论文的发表时间）