Arabic font recognition system applied to different text entity level analysis
Faten Kallel Jaiem, Fouad Slimane, Monji Kherallah · 2017
Arabic text recognition systems encounter many difficulties in processing multi-font and multi-size text images. Some Arabic font families introduce complex variability such as overlaps and ligatures. In this case, developing a font recognition phase for text recognition has become a necessity. In this paper, we presented a new version of the steerable pyramids font recognition system. We also compared and evaluated this system with two font recognition systems based on different methods. We adapted all the systems to a common measurement protocols, and evaluated them with text block, text line and word images. The results are very interesting in the different text entity levels and show the superiority of the steerable pyramid method over the text blocks image with the high resolution and comparable recognition rates with other text entities.