A Rear-lamp Recognition System Based on D-S Evidence Theory
Download as PDF
DOI: 10.23977/meimie.2019.43015
Author(s)
Yuze Wang, Jindong Zhang, Jiale Qiao, Sai Gao, Chenhui Yu
Corresponding Author
Yuze Wang
ABSTRACT
In this paper, a recognition system of headlamp language based on D-S evidence theory is proposed. For each recognition, the features of the pictures captured by multiple cameras are extracted and brought into the trained BPA by using the features of HSV color space, which is transformed into the probability of different decision support. Finally, D-S evidence theory is used to calculate the final fusion results. Experiments show that the algorithm effectively improves the recognition accuracy, and can work normally even if a single sensor fails as well as the recognition rate is improved. In addition, the algorithm is simple in calculation and meets the requirement of high real-time performance for unmanned vehicle system
KEYWORDS
Language recognition, calculation, sensor