Recognition of Tifinaghe characters using neural network

The present work is a contribution in the field of printed Tifinaghe characters recognition. The input pattern is subject to some special treatments. The recognition system consists of three phases: pre-processing, features extraction and recognition. In the pre-processing phase, we applied four operations: digitization, noise reduction, skew correction and segmentation. Then the features extraction and recognition algorithms are applied. The features extraction phase uses the Walsh transformation, which gives a verbose description about the character image. Finally, the neural network is used in the recognition phase. The algorithm proved to be fairy efficient and reliable for simple documents.