Classification of the Italian Lira using the LVQ method
T. Kosaka, S. Ōmatu · 2002
Bill money classification by transaction machines has become important to make progress in office automation. The paper presents a new method to classify Italian Lira using learning vector quantization (LVQ). The Italian Lira of of 8 kinds, 1,000, 2,000, 5,000, 10,000, 50,000 (new), 50,000 (old), 100,000 (new), 100,000(old) Lira with four directions A,B,C, and D are used where A and B mean the normal direction and the upside down direction and C and D mean the reverse version of A and B. The original image with 128/spl times/64 pixels is observed at the transaction machine in which rotation and shift are included. After correction of these effects, we select a suitable aria which shows the bill image and compressed image with 64/spl times/15 pixels to the neural networks. Although the neural network of the LVQ type could process any order of the dimension of the input data, the small size is better to achieve the fast convergence result. Thus, we have selected the above size of the image. Thirty-two bill images are in one set of the classification pattern of the experiment. The total number of data sets is 30 and 10 data sets are used for training of the network and the remaining 20 data sets are used to test the network. After training the neural network, 20 data sets are tested on how well the LVQ network could work. From the simulation results, the proposed method can offer suitable classification results for Italian Lira.