Synthesis of quasi-optimum algorigthms for signal recognition against background of correlated noise employing discrete spectral transformations

A method is proposed for comparative estimation of the information capabilities of various discrete spectral transformations for recognizing signals in the presence of correlated noise. The approach can be used to construct a probability-recognition time relationship in which the individual points correspond to decision-making algorithms with the same operating structure, different only in the use in type of discrete transformation performed in advance. This relationship can be used to determine the need for using a transformation and to select said transformation to satisfy the restrictions regarding the complexity and speed of the recognition algorithms. Fourier and Walsh, as well as other computationally economical transformations, can be used.