An efficient and accurate text localization algorithm in compressed mobile phone image domain

Adrián Canedo-Rodríguez; Jung Hyoun Kim; John Kelly; Jung Hee Kim; Soo Hyung Kim; Yolanda Bianco-Fernández; Sai Kiran Veeramachaneni

An efficient and accurate text localization algorithm in compressed mobile phone image domain

Adrián Canedo-Rodríguez
, Jung Hyoun Kim
, John Kelly
, Jung Hee Kim
, Soo Hyung Kim
, Yolanda Bianco-Fernández
, Sai Kiran Veeramachaneni

Electrical and Computer Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

In this paper, we propose an algorithm to locate textual regions in JPEG mobile phone images by using the information provided by the Discrete Cosine Transform (DCT) coefficients. Given the 8×8 DCT blocks of the JPEG input image, this methodology delimitates the borders of the main text region by using a very efficient process with just 8 DCT coefficients from each block. The result is an algorithm with which we achieved an accuracy of more than 97% under very different real world conditions, suitable of being implemented in low computational resource devices that require short processing time.

Original language	English
Title of host publication	2010 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2010
Volume	2
State	Published - 2010

Cite this

@inproceedings{3460fe3c08f242a9b9193d55f4459aa0,

title = "An efficient and accurate text localization algorithm in compressed mobile phone image domain",

abstract = "In this paper, we propose an algorithm to locate textual regions in JPEG mobile phone images by using the information provided by the Discrete Cosine Transform (DCT) coefficients. Given the 8{\texttimes}8 DCT blocks of the JPEG input image, this methodology delimitates the borders of the main text region by using a very efficient process with just 8 DCT coefficients from each block. The result is an algorithm with which we achieved an accuracy of more than 97\% under very different real world conditions, suitable of being implemented in low computational resource devices that require short processing time.",

author = "Adri{\'a}n Canedo-Rodr{\'i}guez and Kim, \{Jung Hyoun\} and John Kelly and Kim, \{Jung Hee\} and Kim, \{Soo Hyung\} and Yolanda Bianco-Fern{\'a}ndez and Veeramachaneni, \{Sai Kiran\}",

year = "2010",

language = "English",

volume = "2",

booktitle = "2010 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2010",

}

TY - GEN

T1 - An efficient and accurate text localization algorithm in compressed mobile phone image domain

AU - Canedo-Rodríguez, Adrián

AU - Kim, Jung Hyoun

AU - Kelly, John

AU - Kim, Jung Hee

AU - Kim, Soo Hyung

AU - Bianco-Fernández, Yolanda

AU - Veeramachaneni, Sai Kiran

PY - 2010

Y1 - 2010

N2 - In this paper, we propose an algorithm to locate textual regions in JPEG mobile phone images by using the information provided by the Discrete Cosine Transform (DCT) coefficients. Given the 8×8 DCT blocks of the JPEG input image, this methodology delimitates the borders of the main text region by using a very efficient process with just 8 DCT coefficients from each block. The result is an algorithm with which we achieved an accuracy of more than 97% under very different real world conditions, suitable of being implemented in low computational resource devices that require short processing time.

AB - In this paper, we propose an algorithm to locate textual regions in JPEG mobile phone images by using the information provided by the Discrete Cosine Transform (DCT) coefficients. Given the 8×8 DCT blocks of the JPEG input image, this methodology delimitates the borders of the main text region by using a very efficient process with just 8 DCT coefficients from each block. The result is an algorithm with which we achieved an accuracy of more than 97% under very different real world conditions, suitable of being implemented in low computational resource devices that require short processing time.

M3 - Conference contribution

VL - 2

BT - 2010 International Conference on Image Processing, Computer Vision, and Pattern Recognition, IPCV 2010

ER -

An efficient and accurate text localization algorithm in compressed mobile phone image domain

Abstract

Fingerprint

Cite this