Citation

Song J. Adv. Transp. Stud. 2017; 42(SI 2): 63-70.

Copyright

(Copyright © 2017, Arcane Publishers)

DOI

unavailable

PMID

unavailable

Abstract

In this paper, we focus on the problem of road traffic sign detection from complex outdoor scene images, which is of great importance for intelligent transportation system. We illustrate the structure of road traffic sign detection system, which is made up of two steps. In the detection step, images are segmented to obtain candidate blobs, and then shape recognition of traffic sign is implemented via blob tracking. In the recognition step, SVM classifier is used to convert the road traffic sign detection problem to a multi-class classification problem. Furthermore, 2DPCA algorithm is used to implement the work of dimension reduction for feature vectors. Afterwards, the blob tracking operations are executed to promote the performance of road traffic sign detection. Based on the above operations, SVM classifier is utilized to detect road traffic signs from the tracked blobs. Finally, experimental results demonstrate that our proposed algorithm can achieve higher recognition rate accuracy under complex outdoor environments.

Keywords: road traffic sign; complex outdoor scene; SVM; feature extraction; blob tracking

Language: en