Difference between revisions of "Publications/minetto.10.icip"
From LRDE
Line 5: | Line 5: | ||
| title = SnooperText: A Multiresolution System for Text Detection in Complex Visual Scenes |
| title = SnooperText: A Multiresolution System for Text Detection in Complex Visual Scenes |
||
| booktitle = Proceedings of the IEEE International Conference on Image Processing (ICIP) |
| booktitle = Proceedings of the IEEE International Conference on Image Processing (ICIP) |
||
− | | pages = |
+ | | pages = 3861 to 3864 |
| address = Hong Kong |
| address = Hong Kong |
||
| abstract = Text detection in natural images remains a very challenging task. For instance, in an urban context, the detection is very difficult due to large variations in terms of shape, size, color, orientation, and the image may be blurred or have irregular illumination, etc. In this paper, we describe a robust and accurate multiresolution approach to detect and classify text regions in such scenarios. Based on generation/validation paradigm, we first segment images to detect character regions with a multiresolution algorithm able to manage large character size variations. The segmented regions are then filtered out using shapebased classification, and neighboring characters are merged to generate text hypotheses. A validation step computes a region signature based on texture analysis to reject false positives. We evaluate our algorithm in two challenging databases, achieving very good results |
| abstract = Text detection in natural images remains a very challenging task. For instance, in an urban context, the detection is very difficult due to large variations in terms of shape, size, color, orientation, and the image may be blurred or have irregular illumination, etc. In this paper, we describe a robust and accurate multiresolution approach to detect and classify text regions in such scenarios. Based on generation/validation paradigm, we first segment images to detect character regions with a multiresolution algorithm able to manage large character size variations. The segmented regions are then filtered out using shapebased classification, and neighboring characters are merged to generate text hypotheses. A validation step computes a region signature based on texture analysis to reject false positives. We evaluate our algorithm in two challenging databases, achieving very good results |
Revision as of 18:57, 4 January 2018
- Authors
- Rodrigo Minetto, Nicolas Thome, Matthieu Cord, Jonathan Fabrizio, Beatriz Marcotegui
- Where
- Proceedings of the IEEE International Conference on Image Processing (ICIP)
- Place
- Hong Kong
- Type
- inproceedings
- Projects
- Olena
- Date
- 2010-09-01
Abstract
Text detection in natural images remains a very challenging task. For instance, in an urban context, the detection is very difficult due to large variations in terms of shape, size, color, orientation, and the image may be blurred or have irregular illumination, etc. In this paper, we describe a robust and accurate multiresolution approach to detect and classify text regions in such scenarios. Based on generation/validation paradigm, we first segment images to detect character regions with a multiresolution algorithm able to manage large character size variations. The segmented regions are then filtered out using shapebased classification, and neighboring characters are merged to generate text hypotheses. A validation step computes a region signature based on texture analysis to reject false positives. We evaluate our algorithm in two challenging databases, achieving very good results
Bibtex (lrde.bib)
@InProceedings{ minetto.10.icip, author = {Rodrigo Minetto and Nicolas Thome and Matthieu Cord and Jonathan Fabrizio and Beatriz Marcotegui}, title = {SnooperText: A Multiresolution System for Text Detection in Complex Visual Scenes}, booktitle = {Proceedings of the IEEE International Conference on Image Processing (ICIP)}, pages = {3861--3864}, year = 2010, address = {Hong Kong}, month = sep, abstract = {Text detection in natural images remains a very challenging task. For instance, in an urban context, the detection is very difficult due to large variations in terms of shape, size, color, orientation, and the image may be blurred or have irregular illumination, etc. In this paper, we describe a robust and accurate multiresolution approach to detect and classify text regions in such scenarios. Based on generation/validation paradigm, we first segment images to detect character regions with a multiresolution algorithm able to manage large character size variations. The segmented regions are then filtered out using shapebased classification, and neighboring characters are merged to generate text hypotheses. A validation step computes a region signature based on texture analysis to reject false positives. We evaluate our algorithm in two challenging databases, achieving very good results}, keywords = {Text detection, multiresolution, image segmentation, machine learning} }