Handwritten Document Offline Text Line Segmentation

Authors:
C. Weliwitage;A. L. Harvey;A. B. Jennings
Affiliations:
RMIT;RMIT;RMIT
Venue:
DICTA '05 Proceedings of the Digital Image Computing on Techniques and Applications
Year:
2005

Citing 0
Cited 5

Text line extraction from multi-skewed handwritten documents

Pattern Recognition
Text line detection in handwritten documents

Pattern Recognition
Text line segmentation in handwritten documents using Mumford-Shah model

Pattern Recognition
Handwritten Chinese text line segmentation by clustering with distance metric learning

Pattern Recognition
Text line and word segmentation of handwritten documents

Pattern Recognition

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, a method of text segmentation into lines of text known as cut text minimization (CTM) is described. Results applying the CTM method to the NIST data base examples of fifty two word handwritten paragraphs of the American Constitutions are given. The method uses a modified projection method to obtain starting points. Then an optimisation technique is applied which varies the cutting angle and start location to minimize the text pixels cut while tracking between the text lines. Also the method attempts to track around projecting ascenders or descenders by a line following technique. A comparison with the projections method is given. From the results, it is evident that the method is successful on quite distorted documents, and can correctly cut the text block into text lines with minimal incorrect partitioning of data into adjacent lines even when text lines have varying slope and there are penetrations into the space of adjacent lines. The CTM method does not assume text lines have a constant slope.