Abstract: On the basis of reviewing the current RMB-banknote crest number identification research literature, a real-time template generation algorithm to identify crest number in high-speed RMB-banknote moving environment is proposed. Firstly enough complete RMB images should be collected in currency detector equipped with CIS, then the area of crest number should be dug out and processed with tilt correction, binarization, morphology processing, character segmentation and highly normalization. Secondly corresponding template set should be generated to each character bitmap and then according to the principle of maximum difference, five character bitmap should be jetted out from the templates set, forming final template simplified set. Finally, the data and analysis of the real-time identification effect applied with this template are provided.
External IDs:dblp:conf/incos/ZhangW13b
Loading