Spatial linear predictive coding and its error matching for signal classification

Abstract

Mathematical analysis of the behavior of general dynamic systems based on linear prediction plays an essential role in many fields of science and engineering concerning the processing and representation of complex signals. This paper addresses the parameter estimation of the all-pole model of the linear predictive coding in the sense that the signal has both deterministic and random properties. Estimate of the model variance error is used as a basis for the derivation of a spatial distortion measure which can be used for matching spectral patterns.