SIGNATURE FORGERY DETECTION USING DEEP LEARNING

Abstract

Signature forgery detection is a critical aspectof security, providing a reliable method for verifying authenticity. This process analyses features such as stroke patterns, length, continuity, and thickness. Variations in these characteristics, even for signatures from the same individual, make forgery detection a challenging task. The system processes signature images by extracting key features using convolutional neural networks (CNNs). Preprocessing steps such as grayscale conversion and binarization are applied to enhance the clarity of features. The CNN model analyses the input signature and compares it against predefined criteria to determine its authenticity. Effective feature extraction and model training are vital to ensure accurate and reliable results. This method simplifies the verification process while maintaining precision and efficiency.