An Improved Signature Forgery Detection using Modified CNN in Siamese Network
S. Kevin Joe Harris, J. Anitha · 2023
Signature forgery is a major security threat in many important applications, such as banking and legal systems. In this work, an enhanced technique for detecting signature forgeries utilizing convolutional neural networks in Siamese Network is proposed. An improved Convolutional Neural Network architecture is built by incorporating additional features, including feature maps from intermediate layers and additional convolutional layers. The proposed approach is trained and tested with a publicly accessible collection of genuine and forged signature images. The experimental results show that the proposed methodology outperforms the various state-of-art approaches considered in the experiment and attains considerable gain in terms of accuracy, false rejection rate and false acceptance rate. The proposed approach also shows high robustness for various types of forgeries, including skilled forgeries and simple forgeries. The findings also have important implications for the development of more effective and efficient signature forgery detection systems.