Skeletonization thinning and-thickening algorithms and their applications to Arabic characters

Voir Plus Voir Moins

Afficher le Résumé

This thesis developptent of thinninq and thickening algorithpts for the of thinninq purpose or thickening .captera captured ' Arabic fonts. Skeletonization is a widely used preprocessinq technique in tllonearwiset J.ptaqe analysis. analysis of for example.alphanupteric printed or handprinted characters. Chromosoptes. etc. One of the methods for obtaininq the 'skeleton' or a 'l11edloal line' of a binary picture, consists in erasJ.nq points of the boundary of the object without modifyinq the topological propertloes of the picture until it cansists of thin lines. Various existinq thinninq alqorithms are presented and col11pared ta the alqorlothms presented in the thesis.Two new parallel thinninq alqorithl11s are develaped. One loS based detect edqe-paints. local operatians to on end-paints and break-points. other is The a matchinq alqorithm in which a set of eiqht templates and two imaqes (the current imaqe and the workinq i~aqe ) used in the \ are processing. algorithms ~aintain Both the connectivity, conserve the shape of the original imaqe and do not leave extraneaus pixels within a 3-by-3 window ta produce a pixel thin skeleton from connected thick imaqe. AlsQ a a thickeninq alqorithm is esented Experimental results show that these methods are Very effective for thinninq and thickeninq of Arabic fonts.