Scene Text Recognition by using EE-MSER and Optical Character Recognition for Natural Images

Authors

  • Miss. Beena Dhamecha Department of Computer Engineering, R.K.University, Rajkot, Gujrat, India
  • Prof. Vaishali J. Kalariya Department of Computer Engineering, R.K.University, Rajkot, Gujrat, India

Keywords:

text detection; text recognition; complex background; edge-enhanced maximally stable extremal regions(EEMSER),Optical Character Recognition

Abstract

Text in Natural Images contains a valuable information about scene, convey the information about what is
actually depicted in the images. Text extraction is challenging task because of text might be in different style, different
font, illumination condition, poor quality, surface deformation, complex background etc. In this paper proposes a two
method one is EE-MSER for Text Detection and OCR for Text Recognition.In EE-MSER,by using geometric and stroke
width information the letter candidates are filtered to exclude non-text region. OCR method takes the detected text
portion and gives the recognized text. Experiments on ICDAR 2003 Robust Reading character dataset and CHAR74k
dataset shows that proposed technique perform well compared with other techniques.

Downloads

Published

2015-12-25

How to Cite

Miss. Beena Dhamecha, & Prof. Vaishali J. Kalariya. (2015). Scene Text Recognition by using EE-MSER and Optical Character Recognition for Natural Images. International Journal of Advance Engineering and Research Development (IJAERD), 2(12), 91–95. Retrieved from https://www.ijaerd.org/index.php/IJAERD/article/view/5268

Issue

Vol. 2 No. 12 (2015): Volume 2 Issue 12, December 2015

Section

Articles