National Conference on Digital Image and Signal Processing |
Foundation of Computer Science USA |
NCDISP2016 - Number 1 |
August 2016 |
Authors: G. G. Rajput, Suryakant B. Ummapure, Panditkumar Patil |
4608aa0e-fccb-4827-afed-27fe8940f678 |
G. G. Rajput, Suryakant B. Ummapure, Panditkumar Patil . Separation of Touching or Overlapping Lines from Handwritten Document images using Histogram and Connected Component Analysis. National Conference on Digital Image and Signal Processing. NCDISP2016, 1 (August 2016), 15-19.
A generic approach for the separation of overlapping and touching lines within handwritten text document images is proposed in this paper. Presence of touching or skewed that arises due to ascenders or descenders and style of writer makes text line extraction a difficult task. The approach is based on histogram and connected component analysis. The proposed method is a three stage approach wherein non overlapping lines are extracted during the first stage and separation of oriented and touching lines occurs during second and third stages respectively. Average height of a text line computed using histogram profile forms the basis for text line segmentation. The proposed method has been evaluated on 120 handwritten documents written in English, Devanagari, Kannada, Telugu, and Malayalam scripts containing non-overlapping and overlapping or touching occurrences.