Method for unconstrained text detection in natural scene image
Text detection in natural scene images is an important prerequisite for many content‐based multimedia understanding applications. The authors present a simple and effective text detection method in natural scene image. Firstly, MSERs are extracted by the V‐MSER algorithm from channels of G, H, S, O1...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Wiley
2017-10-01
|
Series: | IET Computer Vision |
Subjects: | |
Online Access: | https://doi.org/10.1049/iet-cvi.2016.0452 |
_version_ | 1797684972637650944 |
---|---|
author | Zhandong Liu Yong Li Xiangwei Qi Yong Yang Mei Nian Haijun Zhang Reziwanguli Xiamixiding |
author_facet | Zhandong Liu Yong Li Xiangwei Qi Yong Yang Mei Nian Haijun Zhang Reziwanguli Xiamixiding |
author_sort | Zhandong Liu |
collection | DOAJ |
description | Text detection in natural scene images is an important prerequisite for many content‐based multimedia understanding applications. The authors present a simple and effective text detection method in natural scene image. Firstly, MSERs are extracted by the V‐MSER algorithm from channels of G, H, S, O1, and O2, as component candidates. Since text is composed of character candidates, the authors design an MRF model to exploit the relationship between characters. Secondly, in order to filter out non‐text components, they design a set of two‐layers filtering scheme: most of the non‐text components can be filtered by the first layer of the filtering scheme; the second layer filtering scheme is an AdaBoost classifier, which is trained by the features of compactness, horizontal variance and vertical variance, and aspect ratio. Then, only four simple features are adopted to generate component pairs. Finally, according to the orientation similarity of the component pairs, component pairs which have roughly the same orientation are merged into text lines. The proposed method is evaluated on two public datasets: ICDAR 2011 and MSRA‐TD500. It achieves 82.94 and 75% F‐measure, respectively. Especially, the experimental results, on their URMQ_LHASA‐TD220 dataset which contains 220 images for multi‐orientation and multi‐language text lines evaluation, show that the proposed method is general for detecting scene text lines in different languages. |
first_indexed | 2024-03-12T00:37:38Z |
format | Article |
id | doaj.art-d80635eaeb4f4f0e96606180b0d4f63c |
institution | Directory Open Access Journal |
issn | 1751-9632 1751-9640 |
language | English |
last_indexed | 2024-03-12T00:37:38Z |
publishDate | 2017-10-01 |
publisher | Wiley |
record_format | Article |
series | IET Computer Vision |
spelling | doaj.art-d80635eaeb4f4f0e96606180b0d4f63c2023-09-15T09:32:59ZengWileyIET Computer Vision1751-96321751-96402017-10-0111759660410.1049/iet-cvi.2016.0452Method for unconstrained text detection in natural scene imageZhandong Liu0Yong Li1Xiangwei Qi2Yong Yang3Mei Nian4Haijun Zhang5Reziwanguli Xiamixiding6School of Computer Science and TechnologyXinjiang Normal UniversityNew Medical Road 102Urumqi830054People's Republic of ChinaSchool of Computer Science and TechnologyXinjiang Normal UniversityNew Medical Road 102Urumqi830054People's Republic of ChinaSchool of Computer Science and TechnologyXinjiang Normal UniversityNew Medical Road 102Urumqi830054People's Republic of ChinaSchool of Computer Science and TechnologyXinjiang Normal UniversityNew Medical Road 102Urumqi830054People's Republic of ChinaSchool of Computer Science and TechnologyXinjiang Normal UniversityNew Medical Road 102Urumqi830054People's Republic of ChinaSchool of Computer Science and TechnologyXinjiang Normal UniversityNew Medical Road 102Urumqi830054People's Republic of ChinaSchool of Computer Science and TechnologyXinjiang Normal UniversityNew Medical Road 102Urumqi830054People's Republic of ChinaText detection in natural scene images is an important prerequisite for many content‐based multimedia understanding applications. The authors present a simple and effective text detection method in natural scene image. Firstly, MSERs are extracted by the V‐MSER algorithm from channels of G, H, S, O1, and O2, as component candidates. Since text is composed of character candidates, the authors design an MRF model to exploit the relationship between characters. Secondly, in order to filter out non‐text components, they design a set of two‐layers filtering scheme: most of the non‐text components can be filtered by the first layer of the filtering scheme; the second layer filtering scheme is an AdaBoost classifier, which is trained by the features of compactness, horizontal variance and vertical variance, and aspect ratio. Then, only four simple features are adopted to generate component pairs. Finally, according to the orientation similarity of the component pairs, component pairs which have roughly the same orientation are merged into text lines. The proposed method is evaluated on two public datasets: ICDAR 2011 and MSRA‐TD500. It achieves 82.94 and 75% F‐measure, respectively. Especially, the experimental results, on their URMQ_LHASA‐TD220 dataset which contains 220 images for multi‐orientation and multi‐language text lines evaluation, show that the proposed method is general for detecting scene text lines in different languages.https://doi.org/10.1049/iet-cvi.2016.0452unconstrained text detectionnatural scene imagecontent-based multimedia understanding applicationsV-MSER algorithmcharacter candidatesMRF model |
spellingShingle | Zhandong Liu Yong Li Xiangwei Qi Yong Yang Mei Nian Haijun Zhang Reziwanguli Xiamixiding Method for unconstrained text detection in natural scene image IET Computer Vision unconstrained text detection natural scene image content-based multimedia understanding applications V-MSER algorithm character candidates MRF model |
title | Method for unconstrained text detection in natural scene image |
title_full | Method for unconstrained text detection in natural scene image |
title_fullStr | Method for unconstrained text detection in natural scene image |
title_full_unstemmed | Method for unconstrained text detection in natural scene image |
title_short | Method for unconstrained text detection in natural scene image |
title_sort | method for unconstrained text detection in natural scene image |
topic | unconstrained text detection natural scene image content-based multimedia understanding applications V-MSER algorithm character candidates MRF model |
url | https://doi.org/10.1049/iet-cvi.2016.0452 |
work_keys_str_mv | AT zhandongliu methodforunconstrainedtextdetectioninnaturalsceneimage AT yongli methodforunconstrainedtextdetectioninnaturalsceneimage AT xiangweiqi methodforunconstrainedtextdetectioninnaturalsceneimage AT yongyang methodforunconstrainedtextdetectioninnaturalsceneimage AT meinian methodforunconstrainedtextdetectioninnaturalsceneimage AT haijunzhang methodforunconstrainedtextdetectioninnaturalsceneimage AT reziwangulixiamixiding methodforunconstrainedtextdetectioninnaturalsceneimage |