Complex NSST in Text Localization

DOI

10.2991/ceis-16.2016.80How to use a DOI?

Keywords

text localization; complex wavelet transform; NSST; projection; relative phase

Abstract

Text localization is a crucial course for image processing. In this paper, an algorithm for text localization based on Complex Nonsubsampled Shearlet transform (NSST) is proposed. Complex NSST is implemented by a projection followed NSST. The Projection operation converts the input image into its analytic one, and NSST composes the real component and imaginary component into multiscale and multidirectional subbands. The obtained relative phase and magnitudes are combined to get the new subbands. Then Dynamic thresholding, Morphological operations, logical AND operation, Voting-decision and Vertical and Horizontal Projections are used on the new subbands to achieve the final text regions. Compared with common wavelet transform, the complex wavelet transform has the properties of shift invariance, limited redundancy and low aliasing. At the same time, the added phase information delivers more accurate representation for image. Experiments show that the algorithm can accurately locates the text regions in an image.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Jiu-Wen Zhang
AU  - Lu Zhang
AU  - Chun-Hui Liu
AU  - Min Dong
PY  - 2016/11
DA  - 2016/11
TI  - Complex NSST in Text Localization
BT  - Proceedings of the 2016 International Conference on Computer Engineering and Information Systems
PB  - Atlantis Press
SP  - 395
EP  - 399
SN  - 2352-538X
UR  - https://doi.org/10.2991/ceis-16.2016.80
DO  - 10.2991/ceis-16.2016.80
ID  - Zhang2016/11
ER  -