現行的紙製貨幣「鈔票」,常有以下問題,首先,在不斷的重覆使用下,造成磨損、皺摺及破損等,破壞了鈔票的完整性,嚴重影響這張鈔票的續用性;二者,常被不良集團偽造,為了檢驗這些鈔票,已有不少品鈔與驗鈔的方法與論文被提出。本研究使用ARM11嵌入式系統配合整鈔機機構發展出更快速可靠的整鈔辨識系統,此系統包含鈔票的面向、面額與OVD脫落以及特徵真偽判斷。以CIS配合機構的動作在鈔票經過時掃描鈔票的影像,從影像中擷取鈔票的部份作辨識,並將辨識之結果在LCD顯示螢幕上以QT所撰寫之互動式人機介面告知使用者,並搭配SPI傳輸各類參數來與外部機台溝通。此系統加入了影像辨識功能以彌補市面上傳統點驗鈔機僅以感測器作為整鈔時的依據,減少不適用鈔流入市面的機會,並方便鈔票做整理排序。此系統在每分鐘360張的速率下面額辨識率為97%正反向辨識率為99%,OVD脫落辨識為91%。
Banknote is a paper currency used at present and it has some problems in use. First, when banknote has been used repeatedly, abrasion, wrinkle and damage will undermine its integrity, which may affect the use continuation; second, counterfeits appear in our life frequently. This study developed a rapid and reliable banknote sorting system base on embedded system ARM11. This system includes banknote denomination recognition, face direction identification and OVD fall-off recognition. In this study we added image recognition features to make up traditional detectors based on sensor for banknote sorting. This banknote identification system has achieved a detection success rate of 97% in denomination and 99% in upside/down identification, and OVD fall-off recognition has detection rate of 91%.