Optimization of Reference Signals for Character Recognition Systems

Abstract

The role of signal structure in a signal discrimination system is discussed. The optimality criterion for reference signals for detection in the case of white Gaussian independent noise is defined. The need for normalization of the reference signals is demonstrated. A geometric interpretation is presented. Optimum classes are obtained and several examples cited. A theoretical optimum class of signals is derived against which any set of signals developed within given constraints may be rated.