سال انتشار: ۱۳۸۶
محل انتشار: سومین کنفرانس بین المللی فناوری اطلاعات و دانش
تعداد صفحات: ۷
Hadi Grailu – Engineering Department, Tarbiat Modares University, Tehran, Iran
Mojtaba Lotfizad – Engineering Department, Tarbiat Modares University, Tehran, Iran
Hadi Sadoghi-Yazdi – Engineering Department, Sabzevar Tarbiat Moalem University of Sabzevar, Sabzevar, Iran
In this paper a lossless/lossy Enhanced Pattern Matching (EPM) method is proposed for Farsi/Arabic printed binary text image compressionthat uses the touching nature of characters in theses languages. Using this method the library size will decrease and higher compression rate can be achieved. The PM method has been improved to include Farsi/Arabic texts. For comparison of patterns a generalized distance function is introduced which calculates the amount of similarity of two patterns. Also an encoding/transmission order is introduced for low bitrate progressive compression and low visual distortion. The results show that the compression performance of proposed method is up to 6 times better than the conventional PM  for Farsi/Arabic text images.