Off-line Signature Verification System using Fusion of Novel Diagonal and Statistical Zone based Features
Mandeep Kaur Randhawa, Guru Nānak, Rahul Sharma · 2012
Abstract: The aim of off-line signature verification is to decide, whether a signature originates from a given signer based on the scanned image of the signature and a few images of the original signatures of the signer. Although the verification process can be thought to as a monolith component, it is recommended to divide it into loosely coupled phases (like pre-processing, feature extraction, feature matching and classification) allowing us to gain a better control over the precision of different components. This paper will cover all the important phases used in this field by focusing on feature extraction, the third phase in the process. To demonstrate this, fusion of features, named, diagonal and statistical features are extracted from 25 equal zones and are used as input patterns to test the performance of the model. For training and testing feed-forward back propagation neural network and MatLab 7 is used. Results are evaluated on our own database comprising of 400 signature images taken from 10 signers (40 signatures from each signer or person) over a different period of time. The proposed system has specificity as 89 % and sensitivity as 95%.