Unsupervised classification of handwritten Farsi numerals using evolution strategies
Masoud Sabaei, Karim Faez · 2002
Moments and functions of moments have been utilized as pattern features in various applications to achieve invariant recognition of two-dimensional image patterns. This paper introduces an experimental evaluation of the effectiveness of utilizing orthogonal moments such as Zernike moments, pseudo Zernike moments, and Legendre moments in recognition of the handwritten Farsi numerals. We used evolution strategies (ESs) for clustering of handwritten Farsi numerals, so that the clusters are formed only based on the inherent properties of the pattern features. Considering the fact that the classification is unsupervised, the error rate is about 5% for moments of orders higher than 5. The pseudo Zernike moments of order of 5 have the best performance among all the moment invariants.