Abstract
Optical Character Recognition (OCR) of systems of characters having a strong cursive nature - such as Arabic - necessitates robust segmentation algorithms. This paper presents a sequential algorithm based on the input-time tracing principle and only depends on the connectivity properties of the acquired text in the image binary domain. Therefore, it is different from the known algorithms - that have been devised for the segmentation of typewritten Arabic text - in that no adjustable parameters are needed for successful segmentation.