Road Markers Recognition Based on Shape Information

Road information understanding is a necessary task for intelligent vehicles and driver assistance systems. Previous research mostly focused on the detection and estimation of lane position. Other information provided by road markers was scarcely mentioned. We propose a shape-based road marker detection and recognition method in this paper. The 8-neighbor chain codes of close regions are abstracted from the plane image. Selected moment features and an improved minimal-error-rate classifier are utilized to recognized different lane markers and other road markers. The lengths and slopes of lane markers are fast calculated also using the moment features. The results show that the method can detect and recognize road markers effectively.

[1]  David G. Stork,et al.  Pattern Classification (2nd ed.) , 1999 .

[2]  David G. Stork,et al.  Pattern Classification , 1973 .

[3]  Mohan M. Trivedi,et al.  Video-based lane estimation and tracking for driver assistance: survey, system, and evaluation , 2006, IEEE Transactions on Intelligent Transportation Systems.

[4]  C. Hilario,et al.  Detection and classification of road lanes with a frequency analysis , 2005, IEEE Proceedings. Intelligent Vehicles Symposium, 2005..

[5]  Z. Houkes,et al.  A new method for fast computation of moments based on 8-neighbor chain code applied to 2-D object recognition , 1997, 1997 IEEE International Conference on Intelligent Processing Systems (Cat. No.97TH8335).

[6]  Ming-Kuei Hu,et al.  Visual pattern recognition by moment invariants , 1962, IRE Trans. Inf. Theory.

[7]  A. Watanabe,et al.  Lane detection for a steering assistance system , 2005, IEEE Proceedings. Intelligent Vehicles Symposium, 2005..

[8]  Yoshiki Kobayashi,et al.  Multitype lane markers recognition using local edge direction , 2002, Intelligent Vehicle Symposium, 2002. IEEE.

[9]  Horst Bunke,et al.  Simple and fast computation of moments , 1991, Pattern Recognit..

[10]  Shigeo Abe DrEng.  Pattern Classification , 2001, Springer London.