Fast and robust road sign detection in driver assistance systems

Publication Type:
Journal Article
Citation:
Applied Intelligence, 2018, 48 (11), pp. 4113 - 4127
Issue Date:
2018-11-01
Metrics:
Full metadata record
Files in This Item:
Filename Description Size
20180311_2018AI_Zhang_WJ.pdfAccepted Manuscript Version1.74 MB
Adobe PDF
© 2018, Springer Science+Business Media, LLC, part of Springer Nature. Road sign detection plays a critical role in automatic driver assistance systems. Road signs possess a number of unique visual qualities in images due to their specific colors and symmetric shapes. In this paper, road signs are detected by a two-level hierarchical framework that considers both color and shape of the signs. To address the problem of low image contrast, we propose a new color visual saliency segmentation algorithm, which uses the ratios of enhanced and normalized color values to capture color information. To improve computation efficiency and reduce false alarm rate, we modify the fast radial symmetry transform (RST) algorithm, and propose to use an edge pairwise voting scheme to group feature points based on their underlying symmetry in the candidate regions. Experimental results on several benchmarking datasets demonstrate the superiority of our method over the state-of-the-arts on both efficiency and robustness.
Please use this identifier to cite or link to this item: