Anggraini, Keumala and Handayani, Lestari (2016) Arabic Character Recognition Using Learning Vector Quantization. In: International Conference on Science and Technology for Sustainability, 30 November 2016, Pekanbaru.

Preview

Text Icostech-LestariH 2016.pdf Download (1MB) | Preview

Abstract

Arabic character recognition should be research again. Arabic character have 28 characters with 4 different positions in sentence, then Arabic character has 28-100 characters. The method is used for Arabic character recognition is learning vector quantization neural network. It is because, the learning vector quantization could classify input in category defined on training network. The objective of this study is to testing LVQ method in Arabic characters recognition. The experiment conducted using all types’ position of character in sentence, there are isolated, begin, middle, and end. The testing data of Arabic character passed preprocessing phase to get vector number that was the size of matrix is used as input for learning vector quantization. The size of matrix was 8x10 for isolated, middle; end and 7x12 for begin. The success accuracy rate for isolated was 76, 43%, begin was 65, 45%, middle was 62, 73%, and end was 80%. The success accuracy percentage for all Arabic character was 72%.

Actions (login required)

View Item