Handwritten character recognition via direction string and nearest neighbor matching

This paper studies the current status of handwritten character recognition and two major problems for research. Then one of the problems of feature selection is studied. On the basis of existed research results, a new feature named direction string is put forward in this paper for handwritten character recognition. It uses stroke trend and can integrate the properties of both the traditional statistical features and structural features. A measure of distance between different direction strings is also proposed. Then a classifier for handwritten character recognition is implemented using nearest neighbor matching algorithm based on the proposed direction strings and their distances. The classifier can be used in the platform of Win32, iOS (iPad) and Android (smart phone). Experimental results have shown that the developed classifier could obtain good recognition accuracy.