Farsi Font and Font Size Recognition Based on Analyzing Binarization Effect on Small Components of Document Images
Yaghoub Pourasad, Azam Ghorbani · 2012
This work presents a new method for Farsi font and font size recognition. There are few papers about Farsi font recognition and in these researches only font of an image document is recognized. On the other hand, finding the font size of a text, in addition to its font, can be very useful in document image analysis. Therefore, we present an approach that recognizes the font, and finds the font size of a document image. The method is based on binarization of document and then analyzing the effect of binarization on the document, including the size and shape of dots and broken strokes, which are formed in binarization step to recognize text font and font size. To evaluate our system, a database including 10*49 text images of 7 different fonts and 7 different sizes are formed using paint software, and recognition rate of 95.7% is achieved. This method is applicable on some other languages such as Arabic and Urdu.