A New Kind of Anti-fake Label Application Based on SIFT Algorithm

Abstract

The traditional anti-fake label is easy to fabricated because it has not a unique feature. To satisfy the trace-to-source requirement of nowadays, a new printed label was proposed which combined the advantages of unique QR code and physically unique texture feature of traditional texture label. Adopting SIFT algorithm, the texture feature of this kind of label was extracted and stored in database, and then labels were matched by using minimum Euclidean-distance method. Experimental Results showed that a good identification and matching accuracy can be obtained even labels’ image were deformed. Finally, a new kind of anti-fake label system were completely conformed by texture paper and QR code, SIFT texture feature recognizing module, texture matching module, database and mobile phone App. In this paper, the emphasis was focused on the middle two parts.