Unsupervised categorization of heterogeneous text images based on fractals

This paper deals about text extraction from heterogeneous documents for categorizing documents and indexing tasks. The purpose of this work is to find similar text regions basing on their fonts. First text regions are extracted, and then font matching is performed using fractal descriptors. Experiments are done for both maps and ancient documents.

• Type: Conference
• Year: 2008
• Where: ICPR
• Authors: Badreddine Khelifi, Nizar Zaghden, Adel M. Alimi, Rémy Mullot
• Keywords: Computer Vision, Heterogeneous Documents, ICPR 2008, Similar Text Regions, Text Regions