End-to-end online handwriting signature verification
Yalin Yin, Xiangdong Zhou · 2019
This paper describes an new method for online handwriting signatures verification. The algorithm is based on "Siamese" deep neural network. This network consists of two identical sub-networks joined at their outputs. During verification the two sub-networks extract features from two signatures, while the joining fully-connected network measures the distance between the two feature vectors to determine whether the signature is genuine. The most remarkable advantage of the system is that it can be trained end-to-end without any handcraft feature extraction except some necessary preprocessing. Experiments on the publicly dataset yielded the performance of 4.5% equal error rate (ERR).