Signal and Parameter Estimation

Abstract

Methods to estimate signal or parameter values are frequently needed in multimedia signal coding and analysis. In this chapter, a more general introduction into principles of linear and nonlinear estimation is given. Some examples and procedures of estimation were already explained in the context of linear prediction and motion estimation. This chapter is intended to provide a deeper understanding of the underlying methods and the related optimization procedures. This is partially done for examples from signal restoration, which can straightforwardly be extended to prediction and interpolation of incomplete signals; most of the methods introduced are not restricted to the estimation of signals, but likewise applicable to the estimation of parameters. This is in particular true for cases where parameters are used for optimum modeling, mapping or similarity comparison between signals. The most general applicability is given for nonlinear estimation methods, which do not rely on linear filter models. Finally, methods are discussed which are capable to reject unreliable data in estimation procedures.