An improved fast recognition algorithm for embedded characters

During the development of hand-held character recognition devices, the conflict of limited resources and high demand for real-time makes the traditional support vector machine(SVM) classification method fail to meet the requirements of recognition speed and recognition rate at the same time. A fast classification algorithm of support vector machine(FCSVM) is proposed. After the transformation on the full set of support vectors, a subset of support vectors instead of the full set of support vectors is used in classification. The speed of classification is much faster than that of conventional SVM under the condition that the precision of classification does not decline. The experimental results show that the improved fast classification algorithm can remarkably reduce the computation complexity and improve the classification speed, and the effects are more obvious in embedded systems.