Skip to Main content Skip to Navigation
Theses

Contributions a l'indexation et a la reconnaissance des manuscrits Syriaques

Petra Bilane 1
1 imagine - Extraction de Caractéristiques et Identification
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
Abstract : THIS THESIS IS DEDICATED TO THE COMPUTED EXPLORATION OF SYRIAC MANUSCRIPTS, IT IS THE FIRST STUDY OF THE SORT. SYRIAC IS A LANGUAGE THAT DEVELOPPED IN THE EASTERN REGION OF THE MEDITERRANEAN COAST, ABOUT TWENTY CENTURIES AGO, AND IS STILL IN PRACTICE TODAY. THE HISTORY AS WELL AS THE DEVELOPPMENT OF THE LANGUAGE ARE PRESENTED IN THE FIRST CHAPTER. SYRIAC IS WRITTEN FROM RIGHT TO LEFT WITH A DISTINCT FEATURE WHICH IS A TILT OF ABOUT 45° WHICH RENDERS CLASSICAL SIGNAL AND DOCUMENT ANALYSIS ALGORITHMS WHICH WERE DEVELOPPED FOR OTHER LANGAUGES RATHER USELESS. IN THE SECOND CHAPTER, AFTYER DESCRIBING AND EXTRACTING THE DOCUMENTS STRUCTURE, WE DEVELOPPED A WORD SEGMENTATION METHOD THAT TAKES THIS TILT INTO CONSIDERATION, THIS LEAD US TO ABOUT THIRTY STABLE SHAPES WHICH ARE VERTICAL LETTRES AND "N-GRAMMES" MADE OUT OF TILTED LETTERS. IN THE SECOND PART OF THIS THESIS, WE WERE INTERESTED IN THE CONTENT OF THE DOCUMENTS FOR INDEXATION PURPOSES. WE DEVELOPPED A WORD SPOTTING METHOD THAT ALLOWED US TO FIND ALL THE OCCURRENCES OF A WORD IN A DOCUMENT USING SEVERAL WORD QUERY APPROCHES (WORD SPOTTING, WORD RETRIEVAL). IT IS BASED OM SHAPE SIMILARITY EVALUATED AFTER A THOUROUGH ANALYSIS OF THE ORIENATIONS OF THE HANDWRITING. THE LAST CHAPTER CONSISTS OF A FIRST CONTRIBUTION TO ASSISTED TRANSCRIPTION OF SYRIAC MANUSCRIPTS WHICH RELIES ON THE ABOVE DESCRIBED SEGMENTATION. WE SHOWED THAT TRANSCRIPTION BASED ON INTERACTION, IS IN CONFLICT WITH THE TRADITIONNAL APPROACHES OF O. C. R RECOGNITION.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00499537
Contributor : Petra Bilane <>
Submitted on : Friday, July 9, 2010 - 7:33:38 PM
Last modification on : Friday, October 23, 2020 - 4:49:28 PM
Long-term archiving on: : Monday, October 11, 2010 - 10:09:23 AM

Identifiers

  • HAL Id : tel-00499537, version 1

Citation

Petra Bilane. Contributions a l'indexation et a la reconnaissance des manuscrits Syriaques. Interface homme-machine [cs.HC]. INSA de Lyon, 2010. Français. ⟨tel-00499537⟩

Share

Metrics

Record views

400

Files downloads

1787