Application of SVM in Embedded Character Recognition System

During the development of hand-held character recognition device, the conflict of limited resources and high demanding for real-time makes the traditional classification method fails to meet the requirements of both speed and recognition rate at the same time. Thus, Support Vector Machines (SVM) classification algorithm is applied to the Embedded Character Recognition System. The SVM theory for pattern recognition is introduced firstly. One against one method is used to solve the Multi-class classification problem, with Cross-validation to sort the optimal parameters. The algorithm is finally transplanted to the embedded platform based on ARM. Comparing with the RBF neural networks, the experiment shows that the use of SVM for character recognition brings faster speed and higher recognition rate and meets the system requirement perfectly.