1999 Volume 61 Issue 6 Pages 51-56
We analyzed which type of feature allowed sufficient distinction and determined whether interspecies differences could be recognized using this feature. In this research, we considered the statistical analysis method for feature selection and then tested the validity of the distinction model using the selected feature. The statistical model was found to be appropriate for selecting the most suitable feature that was species dependent. A high correlation was achieved in terms of the precision of distinction overall. It is considered that the method using statistical analysis for feature selection, at the stage of constructing a distinction model, is a useful supplementary means.