Handwriting Moroccan regions recognition using Tifinagh character

The territorial organization of Morocco during administratives division of 2009 is based on 16 regions. In this work we will create a system of recognition of handwritten words (names of regions) using the Amazigh language is an official language by the Moroccan Royal Institute of Amazigh Culture (IRCAM) (2003a) [1] such as this language is slightly treated by researchers in pattern recognition field that is why we decided to study this language (El Kessab et al., 2013 [3]; El Kessab et al., 2014 [4]) that knowing the state make a decision to computerize the various public sectors by this language. In this context we propose a data set for handwritten Tifinagh regions composed of 1600 image (100 Image for each region). The dataset can be used in one hand to test the efficiency of the Tifinagh region recognition system in extraction of characteristics significatives and the correct identification of each region in classification phase in the other hand.


a b s t r a c t
The territorial organization of Morocco during administratives division of 2009 is based on 16 regions. In this work we will create a system of recognition of handwritten words (names of regions) using the Amazigh language is an official language by the Moroccan Royal Institute of Amazigh Culture (IRCAM) (2003a) [1] such as this language is slightly treated by researchers in pattern recognition field that is why we decided to study this language (El Kessab et al., 2013 [3]; El Kessab et al., 2014 [4]) that knowing the state make a decision to computerize the various public sectors by this language.
In this context we propose a data set for handwritten Tifinagh regions composed of 1600 image (100 Image for each region). The dataset can be used in one hand to test the efficiency of the Tifinagh region recognition system in extraction of characteristics significatives and the correct identification of each region in classification phase in the other hand.
& We chose a database word contains 1000 words written in marker and that represents the 16 region of Morocco.
Optical Character Recognition (OCR) can be applied on both cases printed or handwritten. In this work we use several efficient techniques in each of the three principal phases forming a the system of recognition which are firstly the pre-processing then secondly the features extraction then finally learning-classification several studies has been done for recognition of Handwritten Tifinagh regions recognition by using in the features extraction phase the square and triangular zoning method in one hand or in the learning-classification phase the support vectors machines (SVM) and the neural networks on the other hand.

Isolated Handwritten Tifinagh characters
Feature extraction (Square and Triangular Zoning) Support Vectors Machines (Classifier) Amazigh alphabet is considered as a national language since a new constitution of 2011 is a creative field [3][4][5][6][7] is very useful to create a system for Tifinagh hand writing words representing the regions.

Experimental design, materials and methods
For several years, on-line and off-line handwriting character recognition has been considered as a very dynamic field given that its applicability in many different domains such as bank check processing, automatic data entry and postal sorting, The postal automation, bank checks Table 1 The obtained recognition rates τ r and τ g by each hybrid method and each classifier.

Regions
Neural networks Support vectors machines All values of the recognition rate for each region τ r (given in %) and also those of the global rate recognition τ g of all 16 regions (given in %) which we have obtained in the table.
identification, automatic processing of administrative files, etc. In this work we have presented the steps of the recognition system in Fig. 1.
We chose a database word contains 1000 words written with marker and that represents sixteen region of Morocco Table 1.
The extraction steps were We ask 70 students (in Laboratory of Information Processing and Decision Support) to write the 16 region with Tifinagh characters (Fig. 3).
The direction of writing of this character is the left to right in horizontal lines. The characters are written in a way separated in the text (see Figs. 2 and 3). Each original region image has a size equal to 30 Â 30 pixels (Figs. 4-17). The number of the square zones in features extraction equal to 4, 6 and 9 zones.               The degree of the Polynomial (POL) kernel function is equal to 10 and their parameters a ¼b¼1 in classification phase with support vectors machines.
We realized a variation on the size of the zones in features extraction to find the best performing method.
To do this, we have chosen the values {5, 10, 15} of hidden layer neurons number.
The graphical representation to recognition rate of each region τ r is shown in Fig. 5.