Handwritten numerical recognition with neural networks and information fusion
Jinli Cao, M. Shridhar, Majid Ahmadi · 2002
Recently, in the area of character recognition, the concept of combining multiple classifiers has been proposed as a promising direction for the development of robust recognition systems. In this paper, an evidence fusion technique, based on the notion of fuzzy integral is utilized to obtain a reliable, high accuracy handwritten character recognition system. Experiments with a large real world data set reveal the robustness of this system.