Offline Signature Verification using Gaussian Weighting Based Tangent Angle

Offline signature verification is an important and critical task in biometrics as signature shape is varying with time. We are proposing a novel approach for offline signature verification using the histogram of Gaussian Weighting Based Tangent Angle (GWBTA). As signatures differentiated using shapes, so shape features are playing an important role in offline signature verification. Here the distribution of the pixels around the center pixel of the block is encoded to GWBTA. While computing GWBTA, the coordinates of the pixels are transformed to give more importance to the pixels which are nearer to the center pixels by using Gaussian weighting. The distances between histograms of GWBTA of signatures are utilized to classify the signature. The results show that the GWBTA gives more accurate results than existing tangent angle.