Recognition of retroreflective traffic signs by a vehicle camera system
¹Czech University of Life Sciences Prague (CULS), Faculty of Engineering, Department of Vehicles and Ground Transport, Kamýcká 129, CZ165 00 Prague, Czech Republic
²ŠKODA AUTO a.s., tř. Václava Klementa 869, Mladá Boleslav II, CZ293 01 Mladá Boleslav, Czech Republic
*Correspondence: khrapova@tf.czu.cz
Abstract:
The systems of traffic sign recognition are based on the evaluation of three components of every sign: shape, colour and pictogram. There are different factors that can have an influence on the efficiency of detection and recognition of these components. One of the most important factors is the quality of the retroreflective sign surface. Retroreflective sheeting improves the readability of colour and pictogram of traffic sign by increasing brightness of its background and/or legend elements. The aim of the paper is to provide a comprehensive survey of the efficiency of sign’s recognition by a modern vehicle camera system. The traffic sign sheeting was measured by the handled retroreflectometer. Then this measurement was repeated by the modern camera system used for recognition of traffic signs in the vehicle. The results of this paper present the analysis of the recognition efficiency of traffic signs and the overview of other factors that can have a significant impact on sign detection and recognition distance. The results can be used for creation a traffic sign database for learning-based recognition techniques to vehicle camera systems.
Key words:
ADTF, attributes of the sign, recognition distance, TSDR