Pattern recognition of 1D and 2D signals using normalization and normal transformation

Abstract

Different classification methods for 1D signals using the normalization, including normalization in terms of level and in terms of step and the normal transformation, have been proposed. Normal filtering is a variant of special matched filtering. The application of normalization in terms of level and normal transformation for 2D signals was also considered. The filter formation and classification algorithms for the considered variants were presented. An algorithm for direct formation of the matrix operator of normal transformation for 1D signal was described. Case studies were used to illustrate the application of above methods. The possibility of a variable scale of input signals was also taken into account.