ニューロ多国籍紙幣識別における多次元ガウス密度関数による目的外紙幣の排除法 A Rejecting Method of Non-objective Currencies using Multi-Dimensional Gaussian Probability Density Function for Neuro-Multi National Currency Recognition
We have proposed neural network (NN) to develop new types of the paper currency recognition machines. We have showed various effectiveness of NN recognition method by experimental results. However, there are two themes on the paper currency recognition problem. One is the correct recognition of the objective currency. Another is the rejection of non-objective currencies. Conventionally, we use the template matching method as a post processor to reject the non-objective currencies. However, even objective currency is sometimes rejected by this conventional method owing to its terrible damage. In this paper, to solve this problem, we propose a rejecting method of the non-objective currencies by multi-dimensional Gaussian Probability Density Function (PDF). First, we estimate the PDF of currency vector as the multi-dimensional Gaussian PDF. Second, we get the function value and judge the evaluated currency as the objective one or not by this value. Here, we show the effectiveness of the proposed method using multiple national currencies. Furthermore, we implement the proposed method to the experimental system and discuss the effectiveness on a real time system.
- 電気学会論文誌. D, 産業応用部門誌 = The transactions of the Institute of Electrical Engineers of Japan. D, A publication of Industry Applications Society
電気学会論文誌. D, 産業応用部門誌 = The transactions of the Institute of Electrical Engineers of Japan. D, A publication of Industry Applications Society 118(12), 1361-1369, 1998-12-01
The Institute of Electrical Engineers of Japan