Location of title and author regions in document images based on the Delaunay triangulation

Yi Xiao; Hong Yan

doi:10.1016/j.imavis.2003.11.002

Author(s)

Yi Xiao
Hong Yan

Detail(s)

Original language	English
Pages (from-to)	319-329
Journal / Publication	Image and Vision Computing
Volume	22
Issue number	4
Publication status	Published - 1 Apr 2004

Link(s)

DOI	DOI https://doi.org/10.1016/j.imavis.2003.11.002 Final Published version
	Check@CityULib
Link to Scopus	https://www.scopus.com/record/display.uri?eid=2-s2.0-0442296221&origin=recordpage
Permanent Link	https://scholars.cityu.edu.hk/en/publications/publication(11df3348-d237-41f4-a4b6-d497c7878740).html

Abstract

Automatic title and author location can be a crucial step in journal document image processing systems. This paper presents a Delaunay triangulation-based method for identification of title and author areas in a technical document image. The positions and alignments of small text line regions are measured by different triangle groups and the character stroke widths are calculated from the constrained Delaunay triangulation. The rules defining spatial features and font attributes of the title and author region are applied to single line text regions to extract the title and author regions. Our experiment results show that the proposed method is effective. © 2003 Elsevier B.V. All rights reserved.