Two dimensional filtering-A state estimator approach

Abstract

This paper considers the problem of developing recursive filter structures capable of optimal filtering of two dimensional signals. The inability of recently proposed structures to achieve optimal performance is discussed and the required optimal filter structure is presented and shown to be nonrecursive in the normally defined sense. Asymptotic state estimators which retain a recursive structure are developed along with the conditions which must be satisfied for these estimators to operate in a minimum mean square error sense.