A Segmentation Text/Background Method for Degraded Ancient Arabic Manuscript

Before any process of the textual content of a given document image can be performed, the text should be separated from the background of the image. Different methods have previously been developed for this purpose. None have been proved perfect especially for degraded and very poor quality gray-scale document images. In this paper, we review some known binarization algorithms showing their incapacity to separate text from background. We exploit then a local comparison technique which proves efficient in previous works especially to preserve stroke connectivity. To improve the performance of this binarization technique, we have implemented a pre-processing phase which aims to reduce noise and to increase contrast between foreground and background in the image. The achieved results are very encouraging.