Author(s)

Zihan Zeng

Corresponding Author

Zihan Zeng

ABSTRACT

Handwriting identification is an excellent way to identify identities, so people have always used handwritten signatures as their unique features. But with the diversification of counterfeiting methods, people are beginning to need more advanced methods and techniques to verify signatures. This paper proposes a new feature extraction method, combined with a convolutional neural network, to improve the accuracy of signature verification to nearly 80% under the condition of minimizing the amount of computation. This research laid the foundation for further improvement of accuracy and provided a theoretical basis for the establishment of a complete signature verification system.

KEYWORDS

Artificial Intelligence, Deep Learning, Computer Vision, Signature