A novel scene text detection algorithm based on convolutional neural network

Xiaohang Ren, Kai Chen, Xiaokang Yang, Yi Zhou, Jianhua He, Jun Sun

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Candidate text region extraction plays a critical role in convolutional neural network (CNN) based text detection from natural images. In this paper, we propose a CNN based scene text detection algorithm with a new text region extractor. The so called candidate text region extractor I-MSER is based on Maximally Stable Extremal Region (MSER), which can improve the independency and completeness of the extracted candidate text regions. Design of I-MSER is motivated by the observation that text MSERs have high similarity and are close to each other. The independency of candidate text regions obtained by I-MSER is guaranteed by selecting the most representative regions from a MSER tree which is generated according to the spatial overlapping relationship among the MSERs. A multi-layer CNN model is trained to score the confidence value of the extracted regions extracted by the I-MSER for text detection. The new text detection algorithm based on I-MSER is evaluated with wide-used ICDAR 2011 and 2013 datasets and shows improved detection performance compared to the existing algorithms.

Original languageEnglish
Title of host publicationVCIP 2016 - 30th Anniversary of Visual Communication and Image Processing
PublisherIEEE
Number of pages4
ISBN (Electronic)978-1-5090-5316-2
DOIs
Publication statusPublished - 4 Jan 2017
Event2016 IEEE Visual Communication and Image Processing - Chengdu, China
Duration: 27 Nov 201630 Nov 2016

Conference

Conference2016 IEEE Visual Communication and Image Processing
Abbreviated titleVCIP 2016
CountryChina
CityChengdu
Period27/11/1630/11/16

Fingerprint

Neural networks

Bibliographical note

-

Keywords

  • CNN
  • deep learning model
  • I-MSER
  • scene text detection
  • text region extraction

Cite this

Ren, X., Chen, K., Yang, X., Zhou, Y., He, J., & Sun, J. (2017). A novel scene text detection algorithm based on convolutional neural network. In VCIP 2016 - 30th Anniversary of Visual Communication and Image Processing IEEE. https://doi.org/10.1109/VCIP.2016.7805444
Ren, Xiaohang ; Chen, Kai ; Yang, Xiaokang ; Zhou, Yi ; He, Jianhua ; Sun, Jun. / A novel scene text detection algorithm based on convolutional neural network. VCIP 2016 - 30th Anniversary of Visual Communication and Image Processing. IEEE, 2017.
@inproceedings{4d5315de30e942ada29c7c4f0789ee65,
title = "A novel scene text detection algorithm based on convolutional neural network",
abstract = "Candidate text region extraction plays a critical role in convolutional neural network (CNN) based text detection from natural images. In this paper, we propose a CNN based scene text detection algorithm with a new text region extractor. The so called candidate text region extractor I-MSER is based on Maximally Stable Extremal Region (MSER), which can improve the independency and completeness of the extracted candidate text regions. Design of I-MSER is motivated by the observation that text MSERs have high similarity and are close to each other. The independency of candidate text regions obtained by I-MSER is guaranteed by selecting the most representative regions from a MSER tree which is generated according to the spatial overlapping relationship among the MSERs. A multi-layer CNN model is trained to score the confidence value of the extracted regions extracted by the I-MSER for text detection. The new text detection algorithm based on I-MSER is evaluated with wide-used ICDAR 2011 and 2013 datasets and shows improved detection performance compared to the existing algorithms.",
keywords = "CNN, deep learning model, I-MSER, scene text detection, text region extraction",
author = "Xiaohang Ren and Kai Chen and Xiaokang Yang and Yi Zhou and Jianhua He and Jun Sun",
note = "-",
year = "2017",
month = "1",
day = "4",
doi = "10.1109/VCIP.2016.7805444",
language = "English",
booktitle = "VCIP 2016 - 30th Anniversary of Visual Communication and Image Processing",
publisher = "IEEE",
address = "United States",

}

Ren, X, Chen, K, Yang, X, Zhou, Y, He, J & Sun, J 2017, A novel scene text detection algorithm based on convolutional neural network. in VCIP 2016 - 30th Anniversary of Visual Communication and Image Processing. IEEE, 2016 IEEE Visual Communication and Image Processing, Chengdu, China, 27/11/16. https://doi.org/10.1109/VCIP.2016.7805444

A novel scene text detection algorithm based on convolutional neural network. / Ren, Xiaohang; Chen, Kai; Yang, Xiaokang; Zhou, Yi; He, Jianhua; Sun, Jun.

VCIP 2016 - 30th Anniversary of Visual Communication and Image Processing. IEEE, 2017.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A novel scene text detection algorithm based on convolutional neural network

AU - Ren, Xiaohang

AU - Chen, Kai

AU - Yang, Xiaokang

AU - Zhou, Yi

AU - He, Jianhua

AU - Sun, Jun

N1 - -

PY - 2017/1/4

Y1 - 2017/1/4

N2 - Candidate text region extraction plays a critical role in convolutional neural network (CNN) based text detection from natural images. In this paper, we propose a CNN based scene text detection algorithm with a new text region extractor. The so called candidate text region extractor I-MSER is based on Maximally Stable Extremal Region (MSER), which can improve the independency and completeness of the extracted candidate text regions. Design of I-MSER is motivated by the observation that text MSERs have high similarity and are close to each other. The independency of candidate text regions obtained by I-MSER is guaranteed by selecting the most representative regions from a MSER tree which is generated according to the spatial overlapping relationship among the MSERs. A multi-layer CNN model is trained to score the confidence value of the extracted regions extracted by the I-MSER for text detection. The new text detection algorithm based on I-MSER is evaluated with wide-used ICDAR 2011 and 2013 datasets and shows improved detection performance compared to the existing algorithms.

AB - Candidate text region extraction plays a critical role in convolutional neural network (CNN) based text detection from natural images. In this paper, we propose a CNN based scene text detection algorithm with a new text region extractor. The so called candidate text region extractor I-MSER is based on Maximally Stable Extremal Region (MSER), which can improve the independency and completeness of the extracted candidate text regions. Design of I-MSER is motivated by the observation that text MSERs have high similarity and are close to each other. The independency of candidate text regions obtained by I-MSER is guaranteed by selecting the most representative regions from a MSER tree which is generated according to the spatial overlapping relationship among the MSERs. A multi-layer CNN model is trained to score the confidence value of the extracted regions extracted by the I-MSER for text detection. The new text detection algorithm based on I-MSER is evaluated with wide-used ICDAR 2011 and 2013 datasets and shows improved detection performance compared to the existing algorithms.

KW - CNN

KW - deep learning model

KW - I-MSER

KW - scene text detection

KW - text region extraction

UR - http://ieeexplore.ieee.org/document/7805444/

UR - http://www.scopus.com/inward/record.url?scp=85011051633&partnerID=8YFLogxK

U2 - 10.1109/VCIP.2016.7805444

DO - 10.1109/VCIP.2016.7805444

M3 - Conference contribution

BT - VCIP 2016 - 30th Anniversary of Visual Communication and Image Processing

PB - IEEE

ER -

Ren X, Chen K, Yang X, Zhou Y, He J, Sun J. A novel scene text detection algorithm based on convolutional neural network. In VCIP 2016 - 30th Anniversary of Visual Communication and Image Processing. IEEE. 2017 https://doi.org/10.1109/VCIP.2016.7805444