Abstract Views :252 |
PDF Views:3
Authors
Affiliations
1 Department of CSE, Technocrats Institute
of Technology, Bhopal (M.P.), IN
2 Computer Science & Engineering Department, Technocrats Institute of Technology, University of RGPV, Bhopal (M.P.), IN
Source
Automation and Autonomous Systems, Vol 3, No 6 (2011), Pagination: 278-281
Abstract
In the recent years more research towards Indian handwritten character recognition system. Some problems are faced on the classification and recognition. India is the multilingual country, Marathi which is used in Maharashtra state. It is herein proposed the isolated handwritten Marathi numerals recognition system which uses multiple feature extraction methods. Ring data feature extraction method which not till used as per my knowledge for Marathi handwritten numerals recognition. Ring data features are specially, rotation invariant features are those features, which remain constant for rotated and unrotated images. Fourier Descriptor describes the shape in terms of its spacial frequency content. Regular geometrical moments are used for image transformation. The classifier, namely, Minimum distance classifier is used for classification and recognition purpose. This classifier is trained with 16 and 30 dimensional Fourier Descriptors (FD) and Ring data of training samples this recognition system is experimented with a 500 database samples.
Keywords
Fourier Descriptors, Marathi Handwritten Numerals, Minimum Distance Classifier, Regular Geometrical Moments, Ring Data Features.
