New feature extraction approach for bank note classification using Quaternion Wavelets

Article type: Research Article

Authors: Gai, Shan | Yang, Guowei | Zhang, Sheng

Affiliations: Department of Information Engineering, University of Nanchang Hangkong, Jiangxi Nanchang, P.R. China

Note: [] Corresponding author. Shan Gai, Department of Information Engineering, University of Nanchang Hangkong, Jiangxi Nanchang, 330063, P.R. China. Tel./Fax: +86 0791 15180197463; E-mail: gaishan886@163.com

Abstract: In order to improve the performance of the banknote image classification, new feature extraction method based on quaternion wavelet transform (QWT) is proposed in this article. The QWT yields one shift-invariant magnitude and three phases by the quaternion algebra. The statistical characteristics such as mean, standard deviation and entropy are used as feature vector in the banknote image classification. The experimental results show that the proposed method by the QWT obtains better classification results than the conventional methods.

Keywords: Feature extraction, quaternion algebra, quaternion wavelet transform, banknote classification

DOI: 10.3233/IFS-120675

Journal: Journal of Intelligent & Fuzzy Systems, vol. 25, no. 3, pp. 685-694, 2013