Published 12 issues per year
ISSN Print: 0040-2508
ISSN Online: 1943-6009
Indexed in
Expansions in Orthogonal Polynomials for Estimating the Multidimensional Probability Density Characteristics of Recognition System Output Signals
ABSTRACT
An analytical method for computing the recognition characteristics of a correlated normally distributed signal against a correlated normally distributed background has been developed. The method is based on representing the multidimensional probability density of the signal at the processing channel outputs by multidimensional Hermite and Laguerre polynomials series. A sufficient approximation accuracy of the probability density is provided, and analytical expressions for M-dimensional integrals by signal probability at the recognition system output are obtained. The quality of recognition by means of multielement images with allowance for the interchannel correlation and the differences between the law of signal distribution at processing channel outputs and the normal law is estimated. The results given in this article may be used in scientific and engineering computations.