Signature verification using integrated classifiers

Abstract

This paper presents a new approach for off-line signature verification. The proposed system is based on global, grid, ink distribution and texture features. The Boosting algorithm is applied to train and integrate multiple classifiers, and the distance-based classifier used as the base classifier corresponding to each feature set. Adaptive threshold is associated with individuality. Experimental results show the system is insensitive to the order of base classifiers and gets a high verification ratio.