Thinning algorithms for Arabic OCR
Mohamed Mounir Tellache, M.A. Sid-Ahmed, B. Abaza · 2002
The development of effective thinning algorithms for the purpose of development of Arabic OCR (optical character recognition) is considered. Two parallel thinning algorithms are presented. One is based on local operations to detect edge points, end-points, and break-points. The other is a matching algorithm in which a set of eight templates and two images (the current image and a working image) are used in the processing. Both algorithms preserve very well the shape of the original image and yield results that can be effectively used in Arabic OCR algorithms.>