本篇研究延續實驗室學長以渦電流感測器及統計圖樣分析之方法來辨識硬幣真偽,而利用紅外線感測器來改善系統辨識能力,整個系統包含一組模擬投幣機構、一組紅外線感測器、兩組渦電流感測器及硬幣辨識程式。程式撰寫是使用美商國家儀器公司(National Instrument)所出產的LabVIEW 8.2來撰寫,而機構設計是先以Pro-E軟體繪製,再加工製造出。然而硬幣之大小、厚度、材質及周圍輪廓粗糙度是影響感測器偵測結果的重要變因,也是判斷硬幣真偽的主要依據。對於遊樂場代幣(代幣4號)而言,在利用一組紅外線感測器及一組渦電流感測器進行辨識,其辨識能力可提升至95%以上。本篇也設計了一組模擬投幣機構,可針對多枚硬幣進行辨識,如當硬幣通過感測器,被程式判定為偽幣時,程式會給剔除機構一個訊號即時將偽幣剔除。
This paper uses the method of statistical pattern analysis which combines eddy -current sensors、infrared sensors and a coin launcher to identify whether the coin is true or not. The whole system includes a set of coin launcher, infrared sensors, two sets of eddy-current sensors and a recognition program for rapid coin identification. The coin learning and recognition programs are developed by NI LabVIEW software, and the mechanical design is drew by Pro-E software first, and manufacturing later. The size, thickness, material and roughness of the outline of coins are main variables to affect the result of sensing, and also the main accordance to identify the truth of coins. With a set of eddy-current sensors and a set of infrared sensors, the identification rate for counterfeit tokens(token 4) can be increased more than 95%. We design a set of coin launcher, while coins going through the sensors, the program will send a message to acceptation gate to eliminate counterfeit coins when they are distinguished as false.