Utilisation des points d'intérêt pour rechercher des mots imprimés ou manuscrits dans des documents anciens

Abstract : In certain document types, when the vocabulary is not too important, a recognition of the words by visual index can be a solution more effective than using an analytical approach based on the recognition of each letter. This approach is known as word spotting. In this paper we propose a new form for this approach adapted to the identifying headers printed form fields, detection of dashed lines, or transcription of handwritten proper nouns. We suggest to use a dense description by local descriptors for images and a parse for models. This article gives performances evaluations.
Document type :
Conference papers
Complete list of metadatas

Cited literature [12 references]  Display  Hide  Download

https://hal.inria.fr/hal-00750386
Contributor : Jean Camillerapp <>
Submitted on : Friday, November 9, 2012 - 4:48:59 PM
Last modification on : Thursday, November 15, 2018 - 11:57:51 AM
Long-term archiving on : Saturday, December 17, 2016 - 9:26:56 AM

File

cifed2012_submission_3.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-00750386, version 1

Citation

Jean Camillerapp. Utilisation des points d'intérêt pour rechercher des mots imprimés ou manuscrits dans des documents anciens. CIFED 2012 - colloque international sur l'écrit et le document, Mar 2012, Bordeaux, France. ⟨hal-00750386⟩

Share

Metrics

Record views

279

Files downloads

383