Hybrid approach to efficient text extraction in complex color images

Keechul Jung, Junghyun Han

Research output: Contribution to journalArticle

31 Citations (Scopus)

Abstract

Texture-based methods and connected component (CC) methods have been widely used for text localization. However, these two primary methods have their own strength and weakness. This paper proposes a hybrid approach of the two methods for text localization in complex images. An automatically constructed MLP-based texture classifier can increase the recall rates for complex images with much less user intervention and no explicit feature extraction. The CC-based filtering based on the geometry and shape information enhances the precision rates without affecting overall performance. Then, the time-consuming texture analysis for less relevant pixels is avoided by using CAMShift. Our experimentation shows that the proposed hybrid approach leads to not only robust but also efficient text localization.

Original languageEnglish
Pages (from-to)679-699
Number of pages21
JournalPattern Recognition Letters
Volume25
Issue number6
DOIs
Publication statusPublished - 2004 Apr 19

Fingerprint

Textures
Color
Feature extraction
Classifiers
Pixels
Geometry

Keywords

  • CAMShift
  • Connected component
  • Content-based image indexing
  • Mean shift
  • Multi-layer perceptron (MLP)
  • Text localization
  • Texture
  • X-Y recursive cut

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Hybrid approach to efficient text extraction in complex color images. / Jung, Keechul; Han, Junghyun.

In: Pattern Recognition Letters, Vol. 25, No. 6, 19.04.2004, p. 679-699.

Research output: Contribution to journalArticle

@article{00b89a70c0474ffcb79b3c7e07060071,
title = "Hybrid approach to efficient text extraction in complex color images",
abstract = "Texture-based methods and connected component (CC) methods have been widely used for text localization. However, these two primary methods have their own strength and weakness. This paper proposes a hybrid approach of the two methods for text localization in complex images. An automatically constructed MLP-based texture classifier can increase the recall rates for complex images with much less user intervention and no explicit feature extraction. The CC-based filtering based on the geometry and shape information enhances the precision rates without affecting overall performance. Then, the time-consuming texture analysis for less relevant pixels is avoided by using CAMShift. Our experimentation shows that the proposed hybrid approach leads to not only robust but also efficient text localization.",
keywords = "CAMShift, Connected component, Content-based image indexing, Mean shift, Multi-layer perceptron (MLP), Text localization, Texture, X-Y recursive cut",
author = "Keechul Jung and Junghyun Han",
year = "2004",
month = "4",
day = "19",
doi = "10.1016/j.patrec.2004.01.017",
language = "English",
volume = "25",
pages = "679--699",
journal = "Pattern Recognition Letters",
issn = "0167-8655",
publisher = "Elsevier",
number = "6",

}

TY - JOUR

T1 - Hybrid approach to efficient text extraction in complex color images

AU - Jung, Keechul

AU - Han, Junghyun

PY - 2004/4/19

Y1 - 2004/4/19

N2 - Texture-based methods and connected component (CC) methods have been widely used for text localization. However, these two primary methods have their own strength and weakness. This paper proposes a hybrid approach of the two methods for text localization in complex images. An automatically constructed MLP-based texture classifier can increase the recall rates for complex images with much less user intervention and no explicit feature extraction. The CC-based filtering based on the geometry and shape information enhances the precision rates without affecting overall performance. Then, the time-consuming texture analysis for less relevant pixels is avoided by using CAMShift. Our experimentation shows that the proposed hybrid approach leads to not only robust but also efficient text localization.

AB - Texture-based methods and connected component (CC) methods have been widely used for text localization. However, these two primary methods have their own strength and weakness. This paper proposes a hybrid approach of the two methods for text localization in complex images. An automatically constructed MLP-based texture classifier can increase the recall rates for complex images with much less user intervention and no explicit feature extraction. The CC-based filtering based on the geometry and shape information enhances the precision rates without affecting overall performance. Then, the time-consuming texture analysis for less relevant pixels is avoided by using CAMShift. Our experimentation shows that the proposed hybrid approach leads to not only robust but also efficient text localization.

KW - CAMShift

KW - Connected component

KW - Content-based image indexing

KW - Mean shift

KW - Multi-layer perceptron (MLP)

KW - Text localization

KW - Texture

KW - X-Y recursive cut

UR - http://www.scopus.com/inward/record.url?scp=1842759520&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=1842759520&partnerID=8YFLogxK

U2 - 10.1016/j.patrec.2004.01.017

DO - 10.1016/j.patrec.2004.01.017

M3 - Article

AN - SCOPUS:1842759520

VL - 25

SP - 679

EP - 699

JO - Pattern Recognition Letters

JF - Pattern Recognition Letters

SN - 0167-8655

IS - 6

ER -