Offline Signature Verification Using Contour Tracing

The paper proposed a centroid based method for offline signature verification. The signatures are acquired by the individuals on sheet of paper within boxes of pre-determined size. These sheets are then scanned to produce reference signature images. After pre-processing, the global and local features of the signature images are extracted. The authenticity of the signer is determined by comparing the features of the input signature and features of the stored reference signatures. The similarity between the features of input signature and reference signatures is calculated and compared with a threshold. If this similarity value exceeds the threshold, the input signature is authenticated as original signature otherwise they are considered as a forgery. The results are produced focusing on two vital parameters False Acceptance Rate (FAR) and False Rejection Rate (FRR).