真偽幣之識別

Abstract

本論文應用影像處理技術，發展演算法以辨識各種錢幣之真偽，希望未來 能根據理想的演算法製作硬體，提供公用電話機、販賣機等使用，以避免 收到電動 玩具輔幣進而增加收益。此演算法主要分為使用特徵值分析錢 幣之正反面辨識和使用週期比對之邊緣辨識兩大類。本文包含演算法和實 驗及結果，且針對低錯誤率之演算法也提供硬體未來發展的方向。 This thesis applies the image processing technique to develop algorithms for recognition of the real coin. I hope we can implement hardware based on the ideal algorithms for public phones and vending machines to avoid to get the tokens from the TV Games in order to make better profit.Those algorithms are categorized in two parts. The first one uses the characteristic values to analyses the head or tail of the coins. The second one uses the cycle contrast to recognize the rims of the coins. This thesis includes all algorithms forcoin recognition, the detailsand the result of the experiment. Besides the algorithms mention above, we also focuses on the low error rate algorithm. Itmay become the trend for the future development of hardware.