Series number recognition of paper currency based on RBF
Zhenlong Chen · Electronic Design Engineering · 2010
The series number recognition of paper currency is widely used in financial security domain.Regarding the practical features of series number of paper currency,a optical character recognition method was proposed,which was based on the radial basis function network.The camera was used to collect the image of paper currency.After basic preprocessing steps,characters are individually extracted and normalized from the binary image of region of series numbers.Then the 2-D distribution of the characters was taken as the feature vector,and the recognition succeeded with the help of the radial base function network.Experimental results present that with the simple neural network of two hidden layers,high recognition rate up to 95% is reached while the training time lies in 100ms order,which satisfies the optical character recognition(OCR) requirement of series number recognition.