Skew Estimation by Parts

Soma SHIRAISHI; Yaokai FENG; Seiichi UCHIDA

doi:10.1587/transinf.E96.D.1503

Skew Estimation by Parts

Soma SHIRAISHI, Yaokai FENG, Seiichi UCHIDA

Full Text Views

0

Cite this

Summary :

This paper proposes a new part-based approach for skew estimation of document images. The proposed method first estimates skew angles on rather small areas, which are the local parts of characters, and subsequently determines the global skew angle by aggregating those local estimations. A local skew estimation on a part of a skewed character is performed by finding an identical part from prepared upright character images and calculating the angular difference. Specifically, a keypoint detector (e.g. SURF) is used to determine the local parts of characters, and once the parts are described as feature vectors, a nearest neighbor search is conducted in the instance database to identify the parts. Finally, a local skew estimation is acquired by calculating the difference of the dominant angles of brightness gradient of the parts. After the local skew estimation, the global skew angle is estimated by the majority voting of those local estimations, disregarding some noisy estimations. Our experiments have shown that the proposed method is more robust to short and sparse text lines and non-text backgrounds in document images compared to conventional methods.

Publication: IEICE TRANSACTIONS on Information Vol.E96-D No.7 pp.1503-1512

Publication Date: 2013/07/01

Publicized

Online ISSN: 1745-1361

DOI: 10.1587/transinf.E96.D.1503

Type of Manuscript: PAPER

Category: Pattern Recognition

Authors

Soma SHIRAISHI
  Kyushu University
Yaokai FENG
  Kyushu University
Seiichi UCHIDA
  Kyushu University

Keyword

skew estimation, part based, keypoint detector

Cite this

Copy

Soma SHIRAISHI, Yaokai FENG, Seiichi UCHIDA, "Skew Estimation by Parts" in IEICE TRANSACTIONS on Information, vol. E96-D, no. 7, pp. 1503-1512, July 2013, doi: 10.1587/transinf.E96.D.1503.
Abstract: This paper proposes a new part-based approach for skew estimation of document images. The proposed method first estimates skew angles on rather small areas, which are the local parts of characters, and subsequently determines the global skew angle by aggregating those local estimations. A local skew estimation on a part of a skewed character is performed by finding an identical part from prepared upright character images and calculating the angular difference. Specifically, a keypoint detector (e.g. SURF) is used to determine the local parts of characters, and once the parts are described as feature vectors, a nearest neighbor search is conducted in the instance database to identify the parts. Finally, a local skew estimation is acquired by calculating the difference of the dominant angles of brightness gradient of the parts. After the local skew estimation, the global skew angle is estimated by the majority voting of those local estimations, disregarding some noisy estimations. Our experiments have shown that the proposed method is more robust to short and sparse text lines and non-text backgrounds in document images compared to conventional methods.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.E96.D.1503/_p

Copy

@ARTICLE{e96-d_7_1503,
author={Soma SHIRAISHI, Yaokai FENG, Seiichi UCHIDA, },
journal={IEICE TRANSACTIONS on Information},
title={Skew Estimation by Parts},
year={2013},
volume={E96-D},
number={7},
pages={1503-1512},
abstract={This paper proposes a new part-based approach for skew estimation of document images. The proposed method first estimates skew angles on rather small areas, which are the local parts of characters, and subsequently determines the global skew angle by aggregating those local estimations. A local skew estimation on a part of a skewed character is performed by finding an identical part from prepared upright character images and calculating the angular difference. Specifically, a keypoint detector (e.g. SURF) is used to determine the local parts of characters, and once the parts are described as feature vectors, a nearest neighbor search is conducted in the instance database to identify the parts. Finally, a local skew estimation is acquired by calculating the difference of the dominant angles of brightness gradient of the parts. After the local skew estimation, the global skew angle is estimated by the majority voting of those local estimations, disregarding some noisy estimations. Our experiments have shown that the proposed method is more robust to short and sparse text lines and non-text backgrounds in document images compared to conventional methods.},
keywords={},
doi={10.1587/transinf.E96.D.1503},
ISSN={1745-1361},
month={July},}

Copy

TY - JOUR
TI - Skew Estimation by Parts
T2 - IEICE TRANSACTIONS on Information
SP - 1503
EP - 1512
AU - Soma SHIRAISHI
AU - Yaokai FENG
AU - Seiichi UCHIDA
PY - 2013
DO - 10.1587/transinf.E96.D.1503
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E96-D
IS - 7
JA - IEICE TRANSACTIONS on Information
Y1 - July 2013
AB - This paper proposes a new part-based approach for skew estimation of document images. The proposed method first estimates skew angles on rather small areas, which are the local parts of characters, and subsequently determines the global skew angle by aggregating those local estimations. A local skew estimation on a part of a skewed character is performed by finding an identical part from prepared upright character images and calculating the angular difference. Specifically, a keypoint detector (e.g. SURF) is used to determine the local parts of characters, and once the parts are described as feature vectors, a nearest neighbor search is conducted in the instance database to identify the parts. Finally, a local skew estimation is acquired by calculating the difference of the dominant angles of brightness gradient of the parts. After the local skew estimation, the global skew angle is estimated by the majority voting of those local estimations, disregarding some noisy estimations. Our experiments have shown that the proposed method is more robust to short and sparse text lines and non-text backgrounds in document images compared to conventional methods.
ER -