RECOGNITION OF TELUGU CHARACTERS USING NEURAL NETWORKS

摘要: The aim of the present work is to recognize printed and handwritten Telugu characters using artificial neural networks (ANNs). Earlier on recognition has been done conventional pattern techniques. We make an initial attempt here for with improving upon earlier methods which do not perform effectively in presence noise distortion characters. Hopfield model network working as associative memory chosen purposes initially. Due limitation capacity network, we propose a new scheme named Multiple Neural Network Associative Memory (MNNAM). storage overcome by combining multiple parallel. It also demonstrated that suitable recognizing noisy well written different “hands” variety styles. Detailed experiments have carried out several learning strategies results are reported. shown satisfactory possible proposed strategy. A detailed preprocessing from digitized documents described.