Affiliations: [a] Center for Biomimetic Sensing and Control Research, Institute of Intelligent Machine, Chinese Academy of Science, Hefei, Anhui, China
| [b] University of Science and Technology of China, West Campus of University of Science and Technology of China, Hefei, Anhui, China
Correspondence:
[*]
Corresponding author. Bin Wang, University of Science and Technology of China, Hefei, Anhui 230027,
China. Tel./Fax: +86 551 65591168; E-mail: [email protected].
Abstract: In this paper, we have proposed a novel traffic sign recognition algorithm based on sparse representation and dictionary learning. In the past period of research and applications of traffic sign recognition, most of the traffic sign recognition algorithms are based on statistical learning, neural networks and template matching algorithm. In these algorithms, they need high-dimensional mapping during classification, resulting in huge amount of calculation. Meanwhile, when the external environment changes, such as illumination, deformation and occlusion, the recognition rate will be further reduced. The proposed sparse representation theory has much better performance to solve the problems of external environment changed and while we use dictionary learning method to build a traffic sign over-complete redundant dictionary, the experimental results clearly showed that the algorithm we proposed has much better performance than traditional algorithm and also has much higher recognition rates.
