To search, Click below search items.


All Published Papers Search Service


Recognizing Arabic Handwriting Using Statistical Hierarchical Architecture


Ramadhan A. M. Alsaidi, Ayed R.A. Alanzi, Saleh R. A. Alenazi and Taufiq H. Ghilan


Vol. 20  No. 8  pp. 9-15


======= DOI: 10.22937/IJCSNS.2020.20.08.2 ======= Among Artificial Intelligence research, feature extraction of Arabic handwriting is still an important topic and interesting challenge. This work is?based on Smale’s framework which is proposed in his paper “Mathematics of the Neural Response”. According to his proposal, hierarchical architectures can be seen as two sequential and different steps or stages: creating templates and measurements of similarities respectively. The main task of this work is to enhance the accuracy of recognizing Arabic handwriting using Smale's framework. To achieve this goal, this paper presents a statistical developed hierarchical method to improve the extraction of features which have been proved to be more effective in the stated research of Arabic hand-writing recognition. This method introduces one criterion for the selection of informative template based on the arithmetic mean. On the other hand, it considers the Square Pearson Correlation Coefficient (SPCC) technique as a similarity measurement.


arithmetic?mean ? derived kernel ? neural response ? template - Squared?Pearson?correlation?Coefficient - IfN/ENIT databases.