A Segmentation and Recognition Method of RMB Series Number Based on Laplacian Transformation and BP Neural Networks

Abstract

Serial number is an important component to paper money. It is an important identifier to every piece of paper money. This paper presents an effective method on segmentation and recognition of RMB serial number. This methodology can be classified as three stages: preprocessing stage, Serial number segmentation stage and serial number recognition stage. The original serial number image is greylized and normalized in preprocessing stage. Each character in the serial number is segmented by region growth algorithm in segmentation stage and every segmented character is recognized by using neural networks in character recognition stage. Experimental result shows the method discussed in this paper is fast and accurate.