New discrete orthogonal moments for signal analysis

The paper addresses some computational aspects and applications of the polynomial unbiased finite impulse response functions originally derived by Shmaliy as a new class of a one-parameter family of discrete orthogonal polynomials. We present two new explicit formulas to compute these polynomials directly. They are based on the discrete Rodrigues' representation and hypergeometric form, respectively. These straightforward calculations lead us to propose another discrete signal representation in moment domain. Experimental results show a comparison between the proposed moments and discrete Chebyshev moments in terms of noise-free/noisy signal feature extraction and reconstruction error.