Offline signature verification is a challenging and important form of biometric identification. Other biometric measures don’t have variability as that of signatures which poses difficult problem in verification of signatures. In this paper, we explore a novel approach for verification of signatures based on curve matching using shape descriptor and Euclidian distance.
In our approach, the measurement of similarities are proceeded by 1)finding correspondences between signatures, we attach shape descriptor (shape context) with Euclidian distance between the sample points of one signature and the sample point of other signature for better results, 2)we estimate aligning transforms by using this correspondences between signatures, 3) classify the signatures using linear discriminant analysis and measures of shape dissimilarity between signatures based on shape context distance, bending energy, registration residual, anisotropic scaling.