On image analysis by orthogonal moments
Miroslaw Pawlak, Xiu Xiu Liao · 2003
Examines the reconstruction properties of orthogonal moment descriptors. An asymptotic expansion for the global reconstruction error is established. This reveals mutual relationships between a number of moments, the image smoothness, sampling rate and noise model characteristics. The problem of a data-driven selection of an optimal number of moments is addressed.>