—This paper proposes an efficient real time driving sign " stop " detection method using template matching, genetic algorithms and neural network. Stop signs are usually installed on junctions without traffic lights to warn drivers. However, many accidents still occur at these locations because either the driver does not notice the signs or just ignore them. Therefore, to reduce such accidents, we propose an automatic stop sign detection method to aid the drivers and also contribute to future automatic driving system such as the Intelligent Traffic System (ITS). Our method efficiently extracts the sign region´s candidate regions, performs template matching using genetic algorithms and verifies the presence of the road sign using neural networks. Although, we face various problems including camera shake and complicated scenes, our method produces an encouraging accuracy of about 96%.
[1]
Ivan Laptev,et al.
On Space-Time Interest Points
,
2003,
Proceedings Ninth IEEE International Conference on Computer Vision.
[2]
Hironobu Fujiyoshi,et al.
Traffic Sign Recognition Using SIFT Features
,
2009
.
[3]
S. Miyata,et al.
Feature Extraction and Recognition for Road Sign Using Dynamic Image Processing
,
2008,
2008 3rd International Conference on Innovative Computing Information and Control.
[4]
Hironobu Fujiyoshi,et al.
Improvement in the Accuracy of Matching by Different Feature Subspaces in Traffic Sign Recognition
,
2009
.
[5]
Deguchi Daisuke,et al.
Cascaded Traffic Sign Detector Using Generative Learning Considering Color Variance
,
2008
.