Traffic Signs Recognition Based on Affine Invariant Hu's Moment Features

Abstract:

Traffic signs effected by shooting environment and natural environment , and varying degrees of geometric distortion. This work introduced a new method to extract the traffic signs' Hu moment features that based on affine invariant. First , according to the shape coordinates x,y of traffic sign is independent of each other before affine transformation . Then getting traffic signs only have rotating effect by ICA transformation . Finally , recognizing traffic sign by compare the Hu's moment feature. Results show this method can greatly improve the feature extraction accuracy of Hu moments and traffic sign recognition efficiency

• Periodical: Applied Mechanics and Materials (Volumes 321-324)
• Pages: 945-949
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.321-324.945
• Online since: June 2013
• Authors: Min Liu, Jian Xu Mao
• Keywords: Affine Transformation, Hu's Moment Features, Independent Component Analysis (ICA), Traffic Signs Recognition

References

[1] Dandan Zuo, Xin Liu, Shuangdong Zhu: submitted to Journal of Ningbo University (NSEE) , vol.24(2011), p.42 (In Chinese)

[2] Merve Can Kuş, Muhittin Gökmen, Şima Etaner-Uyar: C. International Symposium on Computer and Information Sciences,(2008), p.1

[3] Zhenghe Song, Bo Zhao, Zhongxiang Zhu, Meng Wang and Enrong Mao.C. 2008 2nd International Conference on Future Generation Communication and Networking. Vol.1(2008), p.387

DOI: 10.1109/fgcn.2008.16

[4] Min Shi, Haifeng Wu, Hasan Fleyeh. C. 1st international Congress on Image and Signal Processing, Vol. 4(2008), p.516

[5] A Hyvärinen. J. IEEE Transactions on Neural Networks, Vol.10(1999), p.626

[6] A Hyvärinen and E. Oja. J. Neural Networks, Vol.13 (2000), p.411

[7] Zhuqing Yang, Yong Li, Dewen Hu. J.Acta Automatica Sinica, Vol. 28(2002) , p.762