Study on offline signature has been conducted for several years. Skilled forgery verification is difficult to be verified because of the highest similarity between genuine and forgery signature. Based on previous research, it can be concluded that genuine offline signature is never similar, but it still has consistent features. Otherwise, skilled forgery tries to mimic genuine offline signature as similar as possible. It can be hipotized that if skilled forgery signature is matched to genuine signature, it should match on consistent parts (narrow variety) and unmatch on inconsistent features (wide variety). In this reseach, the offline signature verification is conducted by two steps. In first step, the comparison is conducted based on consistant features as most researcher done. In second step, an acceptance result of first verification will be reverification using inconsistant features in order to improve the verification accuracy in case skilled forgery. Based on the experiment, it can be concluded that this proposed method can improve the verification accuracy for certain condition or depend on writer signature characteristic. Therefore, this approach can be applied only if only the conformance characteristic of writer offline signature can be identified before second step of verivication can be done. At least, this result contribute to open mind that wide variety feature can be used in offline signature verification.
Oleh :
Binti Solihah
