A novel approach for the Arabic handwriting recognition is presented. The use of a Planar Hidden Markov Model (PHMM) has permitted to split the Arabic script into five homogeneous horizontal regions. Each region was described by a 1D-HMM. This modeling is based on different levels of segmentation: horizontal, natural and vertical. Both holistic and analytical approaches have been tested for the description of the median band of the Arabic writing. We show finally that a hybrid approach conducted to the improvement of the whole system performances.

Route Mtdenine 601 1 Gab& (Tunisia)Sameh.Masmoudi@,isetgb.rnu.tn AbstractOff-line recognition of handwritten words is a difficult task due to the high variability and uncertainty of human writing. The majority of the recent systems have some constraints such as the limitation of the size of the lexicon to deal with. However, the recognition of city names involves the use of large vocabulav. Therefore, a segmentation stage is necessary to reduce the complex& of the problem. This paper presents the segmentation stage relative to a Planar HMM-based model (PHMM) for off-line recognition of Arabic cursive Tunisian city names. we discuss the diflerent segmentation steps and the variety of problems encountered when performing them. We especially focused on the median zone whose model depend highly on natural and vertical segmentation results.

No abstract