Word level script identification for scanned document images

Huanfeng Ma; David Doermann

doi:10.1117/12.530538

Word level script identification for scanned document images

Huanfeng Ma
, David Doermann

Department of Computer Science and Engineering

University of Maryland, College Park

Research output: Contribution to journal › Conference article › peer-review

16 Scopus citations

Abstract

In this paper, we compare the performance of three classifiers used to identify the script of words in scanned document images. In both training and testing, a Gabor filter is applied and 16 channels of features are extracted. Three classifiers (Support Vector Machines (SVM), Gaussian Mixture Model (GMM) and k-Nearest-Neighbor (k-NN)) are used to identify different scripts at the word level (glyphs separated by white space). These three classifiers are applied to a variety of bilingual dictionaries and their performance is compared. Experimental results show the capability of Gabor filter to capture script features and the effectiveness of these three classifiers for script identification at the word level.

Original language	English
Pages (from-to)	124-135
Number of pages	12
Journal	Proceedings of SPIE - The International Society for Optical Engineering
Volume	5296
DOIs	https://doi.org/10.1117/12.530538
State	Published - 2004
Event	Document Recognition and Retrieval XI - San Jose, CA, United States Duration: Jan 21 2004 → Jan 22 2004

Keywords

Gabor Filter
Gaussian Mixture Model (GMM)
K-Nearest-Neighbor (k-NN)
Script Identification
Support Vector Machines (SVM)

Access to Document

10.1117/12.530538

Cite this

@article{3818e931db634d3587199600e2447140,

title = "Word level script identification for scanned document images",

abstract = "In this paper, we compare the performance of three classifiers used to identify the script of words in scanned document images. In both training and testing, a Gabor filter is applied and 16 channels of features are extracted. Three classifiers (Support Vector Machines (SVM), Gaussian Mixture Model (GMM) and k-Nearest-Neighbor (k-NN)) are used to identify different scripts at the word level (glyphs separated by white space). These three classifiers are applied to a variety of bilingual dictionaries and their performance is compared. Experimental results show the capability of Gabor filter to capture script features and the effectiveness of these three classifiers for script identification at the word level.",

keywords = "Gabor Filter, Gaussian Mixture Model (GMM), K-Nearest-Neighbor (k-NN), Script Identification, Support Vector Machines (SVM)",

author = "Huanfeng Ma and David Doermann",

year = "2004",

doi = "10.1117/12.530538",

language = "English",

volume = "5296",

pages = "124--135",

journal = "Proceedings of SPIE - The International Society for Optical Engineering",

issn = "0277-786X",

publisher = "SPIE",

note = "Document Recognition and Retrieval XI ; Conference date: 21-01-2004 Through 22-01-2004",

}

TY - JOUR

T1 - Word level script identification for scanned document images

AU - Ma, Huanfeng

AU - Doermann, David

PY - 2004

Y1 - 2004

N2 - In this paper, we compare the performance of three classifiers used to identify the script of words in scanned document images. In both training and testing, a Gabor filter is applied and 16 channels of features are extracted. Three classifiers (Support Vector Machines (SVM), Gaussian Mixture Model (GMM) and k-Nearest-Neighbor (k-NN)) are used to identify different scripts at the word level (glyphs separated by white space). These three classifiers are applied to a variety of bilingual dictionaries and their performance is compared. Experimental results show the capability of Gabor filter to capture script features and the effectiveness of these three classifiers for script identification at the word level.

AB - In this paper, we compare the performance of three classifiers used to identify the script of words in scanned document images. In both training and testing, a Gabor filter is applied and 16 channels of features are extracted. Three classifiers (Support Vector Machines (SVM), Gaussian Mixture Model (GMM) and k-Nearest-Neighbor (k-NN)) are used to identify different scripts at the word level (glyphs separated by white space). These three classifiers are applied to a variety of bilingual dictionaries and their performance is compared. Experimental results show the capability of Gabor filter to capture script features and the effectiveness of these three classifiers for script identification at the word level.

KW - Gabor Filter

KW - Gaussian Mixture Model (GMM)

KW - K-Nearest-Neighbor (k-NN)

KW - Script Identification

KW - Support Vector Machines (SVM)

UR - https://www.scopus.com/pages/publications/8844250730

U2 - 10.1117/12.530538

DO - 10.1117/12.530538

M3 - Conference article

AN - SCOPUS:8844250730

SN - 0277-786X

VL - 5296

SP - 124

EP - 135

JO - Proceedings of SPIE - The International Society for Optical Engineering

JF - Proceedings of SPIE - The International Society for Optical Engineering

T2 - Document Recognition and Retrieval XI

Y2 - 21 January 2004 through 22 January 2004

ER -

Word level script identification for scanned document images

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this