A Comparative Study on Discrete Orthonormal Chebyshev Moments and Legendre Moments for Representation of Printed Characters.
Sarat Saharia, Prabin Kumar Bora, Dilip Kr. Saikia · 2004
Moment functions are widely used in image analysis as feature descriptors. Compared to geometric moments, orthogonal moments have become more popular in image analysis for their better representation capabilities. In comparison to continuous orthogonal moments discrete orthogonal moments provide a more accurate description of the image features. This paper compares the performance of discrete orthonormal Chebyshev moments with continuous orthogonal Legendre moments for representation of noisy printed characters of the Assamese alphabet. Experimental results show that at all noise levels Chebyshev moments outperforms Legendre moments. The performance of both the moments increases when higher order moments are considered but after some point the performance of Chebyshev moments decreases.