Text lines and PAWs segmentation of handwritten Arabic document by two hybrid methods

We present in this paper two hybrid segmentation methods of handwritten Arabic script. Both methods allow the segmentation of Arabic document in text lines and the segmentation of the text line into Pieces of Arabic Words (PAWs) respectively. The particularity of these two methods is that both use a combination with Mathematical Morphology (MM). The first method uses the (MM) in the first hand and the algorithm of construction of the Outer Isothetic Cover of a digital object (OIC) named MM-OIC in the second hand. The second method combines Hough Transform (HT) and MM called HT-MM. The two proposed methods are evaluated to the three databases : IFN/ENIT-database, BSB and KSU online databases. The impact of the MM on the two proposed methods is then discussed.