Optimum projectors for error-trapping and interpolation in multidimensional signals

The problems of locating and correcting erroneous or missing data in multidimensional signals are considered. Although locating errors is basically a nonlinear problem, an extension (which is primarily linear) of the error-trapping algorithm of finite-field cyclic codes to discrete-time real or complex signals, is described. The role of projectors in error-trapping and interpolation and the design of optimum projectors for multidimensional signals are discussed. The design of encoders which result in well-conditioned estimation operators is discussed. Matlab examples are presented.<>