Snap

Handwritten Character Recognition using Adaptive Resonance Theory Network (ART - I)

The objective of the project work is to develop a system that recognizes handwritten characters and converts them into text, to be displayed in a text editor, as well as pronounces it. The methodology involves image acquisition followed by image processing, to extract individual characters. The extracted characters will be converted into a binary input vector. This input vector will be fed to the Adaptive Resonance Theory neural network (ART1). The ART1 neural network would be trained with unsupervised learning technique. ART1 neural network is more stable and plastic. It is advantageous because it enables the learning of a new pattern at any point of time even though it has not been trained earlier. The input pattern would be converted into printed characters and it would be pronounced.