Research And Implementation Of Bank Drafts Identification Technology Based On Image Matching

With the establishment and development of the socialist market economy, more and more activities of using various bank notes for settlement happen in people’s lives. Meanwhile, all kinds of activities of counterfeiting notes and fraud are also increasing, which is a great threat to the safety of banks and customers. It is necessary to reduce the workload of testing and verifing bills and to reduce the mistake rate and miss rate. Therefore, it is important practical significance and valuable for application to use computers for identifing the authenticity of bank notes intelligently.Considering the fact that bank draft is the principle part of counterfeiting bills criminal activities, this paper mainly studies bank draft identification technology. The main idea of identification is:firstly, acquire the image of bank draft and store it in the computer when it is rendered; acquire the image of bank draft again and recall the reserved image when the holder comes to bank for transaction; the program compares and analyzes these two images for authentication. Due to the fact that the factors such as the location where the bank draft was put and the light might interfere in the result of identification, so the images of bank draft need pre-processing operations before the identification. Identification process is mainly composed of three parts:the overall initial matching; the amount of digital identification; seal recognition. The algorithms of the three parts were studied and implemented. The reserved image and the image to be identified were divided into several small areas respectively when carried on the overall initial matching. Calculated the similarity of corresponding small areas in two images by using histogram matching method and marked nonanalogous area in the original images. If there is dissimilarity in the digital area or seal area, it was required to identify the authenticity of the designated area. Both the amount of digital identification and seal recognition consisted of two stages which were rough matching and accurate matching. Those obvious forgery was identified by using the centroid position coordinates in rough matching of amount in figures. Accurate matching was based on edge information, which used the sum of distance between the edge point coordinates and the specified position for authentication. The rough matching of seal was based on the distance among the seals and identified false seals which was greater than the specified threshold. The accurate matching of seal was based on the combination of template matching and correlation matching. Before correlation matching, the image to be identified and reserved image were registrated by using SURF algorithm.On PC, the simulated Bank Draft Identification System was implemented under Matlab development environment. The Experimental result indecates that the algorithm has good real-time performance and accuracy, which is able to meet the actual demand of related bank work.