Recognition of Handwritten Hindi Text Using Middle Region of the Words

Recognition of Handwritten Hindi Text Using Middle Region of the Words

Naresh Kumar Garg (Giani Zail Singh Punjab Technical University Campus, Bathinda, India), Lakhwinder Kaur (Department of Computer Engineering, Punjabi University, Patiala, India) and M. K. Jindal (Panjab University Regional Centre, Muktsar, India)
Copyright: © 2015 |Pages: 10
DOI: 10.4018/IJSI.2015100105
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

Offline handwritten Hindi text recognition is a very tedious task. In this paper, a novel method using middle region of the words for recognition of handwritten Hindi text is proposed. A segmentation based approach is used for recognition. Although many efforts have been made to recognize isolated characters and words, a little work has been done to recognize the offline handwritten Hindi text by segmenting the sentences into lines and lines into words. The uniqueness of this approach lies in the fact that many of the commonly used words can be recognized by matching all the characters in the middle zone even by ignoring the upper modifiers, lower modifiers and half characters. Another advantage is that it is not word specific i.e. any number of words can be added to the list to be recognized. Topological features are used for recognition of characters and efforts are made to correctly extract the features. Results obtained with the proposed technique are very challenging.
Article Preview

A good survey about OCR is given by (Mori et al., 1992). The performance of any classifier depends upon the quality of features fed into it. A very good survey about recognition of Devanagari script is given by (Jayadevan, 2011). It is mentioned in this paper that a lot of research has been done in the past in the recognition of printed text and isolated characters of handwritten Devanagari text, but only few research reports are available on recognition of handwritten text. Work on recognition of printed Devanagari text is explained by (Bansal, 1999). A good survey about feature extraction is given by (Arica and Vural, 2001). (Trier et al., 1996) present an interesting survey of feature extraction method for off-line recognition of segmented characters. The authors describe important aspects that must be considered before selecting a specific feature extraction method.

To the best of author’s knowledge, no commercial OCR for handwritten Hindi text is available, yet. The structural and statistical features are very useful for character recognition (Heutte et al., 1998). The recognition of handwritten numerals is explained in (Shridhar and Badreldin, 1986). The literature of handwritten character recognition shows us basically three main classes of feature extraction methods:

  • 1.

    Structural feature extraction

  • 2.

    Statistical feature extraction (Global transformations and Series Expansion)

  • 3.

    Pixel based features

Structural features may be defined in terms of character strokes, character holes, or other character attributes such as concavities and convexities, end points and junctions, intersection with straight lines etc. Structural features can be classified as local features or global features. Structural or topological feature extraction techniques examine the geometry of the character, e.g., stroke direction, convexities, junction points etc. In fact, topological feature are very useful for handwritten character recognition.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 5: 4 Issues (2017)
Volume 4: 4 Issues (2016)
Volume 3: 4 Issues (2015)
Volume 2: 4 Issues (2014)
Volume 1: 4 Issues (2013)
View Complete Journal Contents Listing