Training model for pattern recognition, especially Hidden-Markov model, involves defining displacement between related training patterns by correlating patterns so that correlation is maximal

摘要

The method involves defining a displacement value representing the displacement between at least two related training patterns (11,13) by correlating the two training patterns so that the correlation is maximal. The training patterns are shifted with respect to each other so that corresponding regions of the patterns have approximately the same length relative to a common reference point. AN Independent claim is also included for the following: a pattern recognition arrangement for implementing the method.