Fast thinning algorithm based on improved SOG (SOG*)

We propose improved self-organized graph (improved SOG: SOG ) thinning method, which maintains excellent thinning results of self-organized graph (SOG) built from self-organizing features map and improves the performance of the existing SOG by using a new incremental learning method of Kohonen features map. In the experiments, this method shows the thinning results to be equal to those of SOG but the time complexity O((logM)/sup 3/) to be superior to that of SOG. Therefore, the proposed method is useful for the feature extraction from digits and characters in the preprocessing step.