Offline signature verification system using grid and tree based feature extraction

摘要

The objective of this paper is to process the hand written signature and verify it. For verifying the signatures of a particular person, we have taken n samples of Genuine Signature, signed by that person on a piece of paper. Further we scanned the paper containing the set of signatures. Now we have extracted each of the genuine signatures of the person and stored it in separate file of the format .bmp. The extraction of the signatures in the last step has been in minimum area to provide accurate area of the signing of signature. We could have matched the signature of each person with the other signature but usually it is almost impossible to produce exactly the same set of signatures. We would verify the signatures on the following parameters allowing a percentage of error in it. Permissible boundary, Hand pressure, Euclidian distance, Center of cylinder generated from minimum spanning tree, Delaunay triangulation of the signature, Angle between base line and center of gravity.