Please use this identifier to cite or link to this item:
The imaging recognition of spurious coins by artificial neural network
|Authors: ||Yi-Cih Jheng|
|Contributors: ||NTOU:Department of Marine Engineering|
spurious coin;Artificial neural network;Pattern match
|Issue Date: ||2011-06-30T08:33:30Z
|Abstract: ||一般鑑定偽硬幣的方式，多是依靠經驗法則去檢驗錢幣之外觀，但是在現今的精密製造技術下要仿製偽幣並不困難，只要選用相同金屬成份材質，並將外徑、厚度控制在標準範圍內，這樣仿造出來的偽幣，若只單靠人的肉眼去判斷是相當容易造成誤判的，因此以影像辨識研究來發展偽硬幣的辨識技術，就是本文之主題。本論文係建立一偽硬幣之鑑定系統，該系統將採用類神經網路理論(Artificial neural network theorem)來進行硬幣真偽之鑑定。鑑定的方式，本文將採用圖樣比對(Pattern match)的方式，將兩枚錢幣取像作運算，計算出兩者之相似度及誤差量，並將誤差較大的地方當成特徵，做為類神經網路和模糊判斷模式之鑑定依據，並由類神經網路決策出硬幣之真偽。|
Traditionally, the method of recognizing spurious coins depends on the experience of antiquarian expert by examining the feature of coin. However, it is not difficult to fake the spurious coins by the technique of modern manufacture technology under the conditions of the same material, diameter and thickness as genuine coins. Hence, correctly to detect out the spurious coins can not only be dependent on the eyes of human being. To improve this situation, a more accurate method that can avoid the faults of artificial judgment is proposed by this paper. In this paper, a recognizing system of artificial neural network and pattern comparing model is used to fulfill the work of detecting spurious coins. Pattern comparing model is used to comparing the discrepancies of spurious and genuine coins. In this procedures, the similarity (called error amount) is calculated as the input data for the artificial neural network. The imaging for each coin is divided 37 regions for comparing. Back propagation neural network is trained to predict the genuine and sham of coins. The results are shown that a good judgment is obtained by this paper.
|Appears in Collections:||[輪機工程學系] 博碩士論文|
Files in This Item:
All items in NTOUR are protected by copyright, with all rights reserved.