Counterfeit Identification by Combining Printing Characteristic and Image Processing

• Main Authors: Bor-Jang Jean, 簡博章
• Other Authors: Ming-Chuan Wu
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/mnehxf

碩士 === 國立臺北科技大學 === 自動化科技研究所 === 101 === Since artificial pigment can’t be mixed into natural light like RGB, most commercial printings takes CMYK four colors process or overprint to print. However, the banknotes printing commonly adopts special and multiple printing formats, with different hue and patterns, which is quite different from the traditional printing, photocopying and printers. Based on the unique characters of banknote printing, this paper extracts the characteristic value through image processing, and then detects them with Bayesian classifier, so as to identify the counterfeit banknotes via a simple program. Different to the common currency detectors that require specialized light or backlights, the method proposed in this paper can conduct identification under general light environments. Besides, this paper can also allow the smart phones with close-up photographing and the function of identifying counterfeit banknotes through camera.