A multireference joint transform correlation technique for rotation-in
variant recognition as well as estimation of the rotational distortion
of input scene target(s) is proposed. Reference patterns correspondin
g to different ranges of distortions are synthesized from a set of tra
ining patterns. Different types of training set selection strategies a
re discussed. (C) 1998 Society of Photo-Optical Instrumentation Engine
ers. [S0091-3286(98)01201-X].