Abstract
This paper presents a dual classifier handprinted character recognition system that is implemented using Radial Basis Function (RBF) networks. Each classifier in the system extracts a different see of features from the input character and makes its own independent classification decision. The features used are the diagonal and partitioned radial projections, and the four-directional edge maps of the image. The system then combines these decisions before giving a final classification output. Several different methods of designing the combiner are examined. The proposed system is tested on a database of handprinted alphanumeric characters, and the results are found to be very promising.