Off-line Chinese Signature Verification based on Optimal Matching of Projection Profiles

An optimal scheme for off-line Chinese signature verification is described in the paper. By using the projection profiles of signatures in different directions, the scheme obtained the objective function of signature verification. Then it employed dynamic programming for optimal matching and carried out training and verification. Otherwise, the technique of selecting the reference signature sample and decision threshold in verification was proposed in order to reduce random disturbance. The experimental results show that the least error rate can be obtained by using the combination of two different directions of projection profiles in gray images.