Efficient Two Level Decision Regions for the Recognition of Kannada Numerals Invariantof Font Style and Size

Research areas:

• Volume 2,Issue 3, Nov. 2013

Year:

2013

Type of Publication:

Article

Keywords:

Segmentation, Pixel Density, Template Matching, Conflict Resolution

Authors:

Ajit Danti; Sahana H. R.

Journal:

IJAIM

Volume:

2

Number:

3

Pages:

62-65

Month:

November

ISSN:

2320-5221

Abstract:

Optical Character Recognition (OCR) is one of the important fields in image processing. Kannada is one such south Indian language which is also the official language of Karnataka state. In this paper two level decision regions are used to recognize Kannada numerals. The algorithm proposes conflict resolution technique to resolve conflicts while conflicting features are encountered. The combination of feature extraction method and conflict resolution method is applied to recognize Kannada numerals. Numerals are collected from the standard Nudi 4.0 and Baraha 7.0 word processor software. Some scanned images also taken for testing. Proposed algorithm efficiency and the overall the classification accuracy is found to be 96.95%.

Full text: IJAIM_195_Final.pdf