Traffic sign localization based on edge-color pair and feature filters

Publication Type:
Journal Article
Citation:
ICIC Express Letters, 2016, 10 (3), pp. 727 - 732
Issue Date:
2016-03-01
Full metadata record
Files in This Item:
Filename Description Size
JianGuoYaoPaper.pdfSubmitted Version206.73 kB
Adobe PDF
© 2016, ICIC International. Traffic sign localization is a key part of the traffic sign recognition system. A novel algorithm for traffic sign localization based on edge-color pair combined with feature filters is presented in this paper. Firstly, the color edge detection is taken by the method of distance measure of direction region in a color image. Secondly, the suspected regions of the traffic sign are localized roughly based on edge-color pair. Finally, a two-level feature filter is designed to localize the traffic sign accurately. The experiment on 200 traffic sign images that were taken under various conditions shows the extraction rate of 96.1%. The experimental results show that the algorithm in this paper can effectively eliminate the color fade effect and the analogue interference effect on traffc sign localization.
Please use this identifier to cite or link to this item: