Recognizing the Denomination of a Note Using Wavelet transform

The invention as denomination recognition method using a wavelet transform, and more particularly, to the full, or by performing a pre-processing a predetermined number of times, such as edge detection in the bill image input to a more accurate denomination recognition portion without being affected by contamination of the banknote wavelet after migration to a method of executing the identified banknotes to a feature vector constructed by comparing the absolute magnitude of the wavelet coefficients and the reference value. The invention Wavelet the method comprising performing a pre-processing such as edge detection, the original image of the preprocessing a predetermined number of times with respect to an original image generated by scanning a bill conversion step, and each of the sub-generated in the wavelet transformation to obtaining a wavelet coefficient for the band, than with the individual sub-band comparing the absolute size and the reference value of the wavelet coefficients for each cell divided by a predetermined number of horizontal m, vertical n; and the horizontally m, vertical n dividing a reference value comprising the steps of: greater than or equal to extract the number of such wavelet coefficients, respectively, wherein the extraction step to reduce the size of the step for completion by the sub band constituting the characteristic vector of the vertical structure, the discriminating power increase and a feature vector characteristic vector and the step of applying a regular analysis, the degree of similarity between the vector of the feature vector based on a predetermined determination It is achieved by having the step of measuring and recognizing the denomination. Wavelet transform, subband, feature vectors, a regular analysis, denomination recognition, edge detection