Discrimination selective optical weighted correlations

We have defined nonlinear correlations based on a superposition of simple linear correlation between binary images in order to control the discrimination capability in pattern recognition. The facility of optical system to perform linear correlation is the basic motivation. Moreover the binary information required allows the use of fast spatial light modulators working on binary regime. Particularly, in this paper we define a weighted nonlinear correlation named WSONG correlation, based on a correlation matrix representation. The procedure uses different diagonals for different recognition purposes. We have defined an amplitude of discrimination parameter as a function of the number of diagonals of the correlation matrix. The greater the tolerance, the higher number of correlation matrix diagonals involved in the process.