Offline MODI Character Recognition Using Complex Moments

Abstract

The varying writing style and critical representation of characters in Indian script makes Handwritten Optical Character (HOCR) challenging and has attracted researchers to contribute in this domain. 'MODI' Script had cursive type of writings in Devanagari, Marathi where large amount of historical documents were available and need to be digitally explored. The principal objective of this research work is to describe efficiency of Zernike Complex moments and Zernike moments with different Zoning patterns for offline recognition of handwritten 'MODI' characters. Every character was divided in six zoning patterns with 37 zones. Geometrical shapes were used to create zoning patterns. The work was resulted in 94.92% correct recognition rate was achieved by using Zernike moments and 94.78% by using Zernike complex moments with integrated approach for heterogeneous zones.