Discrete Cohen's class of distributions

An alias-free discrete-Cohen's class of distributions for finite discrete signals is defined. Based on this definition, the authors derive desirable distribution properties and the corresponding kernel constraints that show strong similarity with the continuous case. They also present: (a) transformations of kernels to different domains; (b) sampling schemes for discretizing continuous kernels; and (c) a new kernel constraint for reversibility of the discrete Cohen's class.<<ETX>>