


摘 "要: 紙幣新舊檢測是紙幣清分機中的一個重要功能,與傳統的基于紙幣反射圖像檢測紙幣新舊等級不同,采用了紙幣的透射圖像進行紙幣新舊檢測。根據紙幣的新舊程度人工劃分為3個等級,分別使用了基于灰度圖像整體亮度的新舊檢測法和傳統的模式識別的方法,采用紙幣透射圖像直方圖分布作為紙幣新舊特征,分別使用了KNN、SVM分類器進行分類,并且提出一種簡單的級聯分類器。實驗結果表明級聯分類器比KNN和SVM的單獨使用有更好的表現。
關鍵詞: 透射圖像; 模式識別; SVM; 級聯分類器
中圖分類號: TN919?34; TP391.41 " " " " " " " 文獻標識碼: A " " " " " " " " " 文章編號: 1004?373X(2015)06?0101?04
New and old banknotes′ classification based on transmission images
QI Lei, REN Ming?wu
(School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China)
Abstract: Classification of new or old banknotes is a important function in the banknote sorter. The transmission image is adopted in this paper to detect new or old banknotes. It is different from the traditional new or old banknote detection based on reflection images. According to the new or old degree of banknotes, all banknote samples are divided into three grades. The new or old banknote detection method based on gray level image brightness and pattern recognition method are used respectively. The histogram distribute of transmission image is used as feature of new or old banknotes. The KNN and SVM classifier are employed to carry out classification. A new cascade classifier which combines SVM and KNN is put forward. Test proves the new cascade classifier is more effective than SVM and KNN in detection of new or old banknotes.
Keywords: transmission image; pattern recognition; SVM; cascade classifier
0 "引 "言
紙幣清分機是一種集光機電磁于一體的高端金融現金處理設備,其主要功能是對紙幣進行面值、面向、新舊程度及紙幣序列號的識別。紙幣清分機挑選適合在市場中流通的紙幣、高效地代替了人工選擇可流通紙幣的工作。當紙幣流通時間過長時,表面會出現粗糙、模糊、磨損等現象。過舊的紙幣影響了紙幣的真偽鑒別、冠字號識別等,導致貨幣在市場中的流通帶來了不便。根據中國人民銀行發布的《不宜流通人民幣挑剔標準》,過舊的紙幣不宜流通,因此紙幣清分機中需要具有檢測紙幣新舊的功能來滿足這樣的需求。
早期國外曾根據紙幣在清分機中與傳送帶的摩擦聲音來判別新舊[1],但該技術已被數字圖像技術[2]所代替。目前,國內紙幣清分機產品多采用基于數字圖像技術做相關研究[3]。紙幣新舊識別成為紙幣清分算法的核心和難點,而且新舊等級分的越多,難度越大。……