Chinese handwriting signature authentication using data mining technique

The data mining technique is applied to search stable feature set and build authentication rules of handwriting signature in this paper. Supervised by data mining technique, 10 stable features including maximum speed, maximum acceleration, the amount and the places of inflexions and etc have been selected from 61 original signature features. Taking the selected feature set as the input attribute, true or false signature sample clusters are trained and learned to build authentication rules supervised by data mining technique to test the validity of the selected feature set. The result of the test shows that the selected feature set is effective to identify handwriting signature and the average veracity of Chinese authentication is up to 92%. It is proved that data mining technique is an effective method to identify handwriting signature.