New method of banknote feature extraction
Xianglong Tang · Journal of Communications · 2010
Combined with the merits of Contourlet transform and fuzzy logic,a new method of banknote image feature extraction was proposed. Contourlet coefficients were extracted in different scales and directions by using Contourlet transform on the banknote image. The extracted coefficients were considered as linguistic variables. Then it calculates the firing strength on the fuzzy feature plane and normalizes them as feature vector. Coarse classification was conducted by banknote geometric features. Then the neural network was constructed for the recognition of banknote and the scheme of reject class was introduced. The experimental results show that this method achieves a high recognition rate and fast recognition speed. The proposed method was applied in the resource constrained of embedded system.