Skip to Main content Skip to Navigation
Conference papers

Multi-script Baseline Detection Using Perceptive Vision

Aurélie Lemaitre 1 Jean Camillerapp 1 Bertrand Coüasnon 1
1 IMADOC - Interprétation et Reconnaissance d’Images et de Documents
UR1 - Université de Rennes 1, INSA Rennes - Institut National des Sciences Appliquées - Rennes, CNRS - Centre National de la Recherche Scientifique : UMR6074
Abstract : Baseline detection is an important step for off-line handwriting recognition of many alphabets. However, it remains a difficult task in the case of varying and irregular handwriting, and sometimes requires script-specific adaptations. In this paper, we propose a new method for baseline detection that is script independent. Indeed, this method is based on the principle of the perceptive vision, which combines several points of view of the same word (from low to high resolution). We validate the genericity of our method by extracting the baseline on three scripts: Arabic, French, and Indian (Bangla), without any adaptation of parameter. We compare our results with other works on the Arabic IFN/ENIT database. Moreover, baseline extraction is often realized independently for each studied word. In this work, we also demonstrate how the use of knowledge on the initial context of the word (text line) significantly improves baseline extraction.
Complete list of metadatas

Cited literature [9 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-00545055
Contributor : Aurélie Lemaitre <>
Submitted on : Thursday, December 9, 2010 - 2:49:40 PM
Last modification on : Friday, November 16, 2018 - 1:21:50 AM
Document(s) archivé(s) le : Thursday, March 10, 2011 - 11:54:47 AM

File

IGS_final_v2.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00545055, version 1

Citation

Aurélie Lemaitre, Jean Camillerapp, Bertrand Coüasnon. Multi-script Baseline Detection Using Perceptive Vision. 14th Biennial Conference of the International Graphonomics Society (IGS 2009), Sep 2009, Dijon, France. ⟨hal-00545055⟩

Share

Metrics

Record views

228

Files downloads

89