The Method to Evaluate a Similarity of Two Dimensional Data Using Critical Point Graph

According to increase in use frequency of computer simulation, the system is needed that classifies a large amount of these results into database and retrieves it. To classifies and retrieves simulation data, it is necessary to evaluate the similarity between characters of data. It has been reported that a method using "critical point graph (CPG)" as an index of data is effective in evaluating. However, a past method using CPG has a question of not corresponding to affine transformation of data. In this paper, we propose the evaluation method corresponding to affine transformation using CPG in two dimensional data. In proposal method, we normalize datasets by using principal component analysis. Actually using the proposal method, we evaluated the similarity between several data. From the result, one can safety state that this method is corresponding to affine transformation