Factors affecting the accuracy of automatic signature verification
Kamlesh Kumari, Vimal Kumar Shrivastava · International Conference on Computing for Sustainable Global Development · 2016
People are comfortable with pen and paper for authentication in legal transactions. Forensic Handwriting experts (FHE) use graphometric and graphology concepts to verify handwritten signature. Handwritten signature is a special case of handwriting. Appropriate selection of effective factors of identification is very important for the design and development of an effective and efficient Automatic Signature Verification. For ASV, various feature extraction techniques and knowledge models have been used in research. In this paper, we show how efficiently the image processing based feature extraction techniques capture the properties of handwritten signature features that could be described algorithmically. Each feature has its own significance and the contribution of discriminative feature affects the accuracy. Local Binary Pattern, Histogram of Gradient, Gray Level Co-occurrence Level Matrix and SpeedUp Robust Feature based feature extraction techniques are useful for verification of handwritten signature image. Parameters used with knowledge model which affect the accuracy of ASV are also described.