Neuro-Fuzzy approach in the recognition of Arabic Characters
N. Ben Amor, M. Zarai, Najoua Essoukri Ben Amara · 2006
In the present paper we propose a hybrid approach for Arabic optical characters recognition (AOCR). The algorithm is based on wavelet transform for features extraction and a neuro-fuzzy approach for classification. Five features are used to describe the different characters. We briefly present the two single methods which are based respectively on the use of logic fuzzy sets and multilayer neural networks. We also describe a third method combining both techniques. We give our perspectives and provide comments for future use