Research On The Image Processing System Of Currency Sorter

Currency Sorter is an optical, mechanical and electrical financial facility. For the need of the ATM's demand, Currency Sorter mainly classifies the new and used bills, face direction and denomination of the bills. Digital image recognition is a kind of recognition technique widely used in Currency Sorter. Aimed at the deficiency of the existent currency sorters, such as unefficient performance and unsatisfied result, the paper makes a thorough study on the key technology and brings out the parallel system architecture which based on the advanced chip DSC25. The parallel processing arithmetic that can improve the whole performance and the Image Processing arithmetic are researched. The paper consists of the following parts:1. The research of the image processing arithmetic. The paper summarizes the currency recognition technique in the world. Aimed at the feature of the bills, the paper brings out the classification arithmetic of the new and used bills based on the histogram of image and the block feature. Since the Artificial Neural Network has the property of self-organization, self-adaptation and fault tolerance, the paper uses the modified LVQ network to discriminate the new and used bills through the comparison of BP and LVQ network. On the basis of the actual state, we improve on the better one in the all homothetic arithmetic to realize other functions of the Currency sorter, such as denomination recognition, edition recognition and orientation recognition.2. The hardware and software system design of the Currency Sorter. For the allowance of the real time demand, the paper analyses the key factors which determine the system running velocity. Through comparison the hardware solution of the Currency Sorter, the parallel system architecture is proposed in this paper, which based on the advanced chip DSC25 (ARM/DSP dual kernel chip) to complete the digital image scanning and processing. This system includes the Image Processing system based on DSC25 and the control system based on MC9S12A128B. The parallel processing arithmetic that can improve the whole performance and the Image Processing arithmetic are researched on this platform. We present the detail software design including all function modules.3. The experiment result analysis. The classification arithmetic is verified on the real machine. The system running velocity is tested too. The factors which can influence the recognition rate are analyzed.