Publications/calarasanu.16.visapp

Authors

Stefania Calarasanu, Séverine Dubuisson, Jonathan Fabrizio

Where

Proceedings of the 11th International Conference on Computer Vision Theory and Applications (VISAPP)

Place

Rome, Italie

Type

inproceedings

Projects

Image"Image" is not in the list (Vaucanson, Spot, URBI, Olena, APMC, Tiger, Climb, Speaker ID, Transformers, Bison, ...) of allowed values for the "Related project" property.

Date

2016-02-01

Abstract

A frequent challenge for many Text Understanding Systems is to tackle the variety of text characteristics in born-digital and natural scene images to which current OCRs are not well adapted. For example, texts in perspective are frequently present in real-word images, but despite the ability of some detectors to accurately localize such text objects, the recognition stage fails most of the time. Indeed, most OCRs are not designed to handle text strings in perspective but rather expect horizontal texts in a parallel-frontal plane to provide a correct transcription. In this paper, we propose a rectification procedure that can correct highly distorted texts, subject to rotationshearing and perspective deformations. The method is based on an accurate estimation of the quadrangle bounding the deformed text in order to compute a homography to transform this quadrangle (and its content) into a horizontal rectangle. The rectification is validated on the dataset proposed during the ICDAR 2015 Competition on Scene Text Rectification.

Documents

• Paper

Bibtex (lrde.bib)

@InProceedings{	  calarasanu.16.visapp,
  author	= {Stefania Calarasanu and S\'everine Dubuisson and Jonathan
		  Fabrizio },
  title		= {Towards the rectification of highly distorted texts},
  booktitle	= {Proceedings of the 11th International Conference on
		  Computer Vision Theory and Applications (VISAPP)},
  address	= {Rome, Italie},
  month		= feb,
  year		= 2016,
  abstract	= {A frequent challenge for many Text Understanding Systems
		  is to tackle the variety of text characteristics in
		  born-digital and natural scene images to which current OCRs
		  are not well adapted. For example, texts in perspective are
		  frequently present in real-word images, but despite the
		  ability of some detectors to accurately localize such text
		  objects, the recognition stage fails most of the time.
		  Indeed, most OCRs are not designed to handle text strings
		  in perspective but rather expect horizontal texts in a
		  parallel-frontal plane to provide a correct transcription.
		  In this paper, we propose a rectification procedure that
		  can correct highly distorted texts, subject to rotation,
		  shearing and perspective deformations. The method is based
		  on an accurate estimation of the quadrangle bounding the
		  deformed text in order to compute a homography to transform
		  this quadrangle (and its content) into a horizontal
		  rectangle. The rectification is validated on the dataset
		  proposed during the ICDAR 2015 Competition on Scene Text
		  Rectification. }
}