Artificial Neural Network Based Sinhala Character Recognition

Sinhala is the main language spoken by the majority of the population of Sri Lanka. There is a clear need for an optical character recognition (OCR) system for the Sinhala language. However, the language contains very similar characters, which makes it very difficult to distinguish them except on feature analysis. The character recognition rates of previous systems proposed for Sinhala character recognition are low, and so further improvement is needed. Consequently, in this paper, we propose a new Sinhala character recognition method that uses character geometry features and artificial neural network (ANN). The results of experiments conducted using various documentary images of the Sinhala language indicate that the proposed method has better character recognition performance than conventional methods.