Off-line signature verification focused on local stroke direction

摘要

This paper proposes an off-line signature verification method in which we extract some "characteristic part", and we judge from the feature of stroke direction in the extracted range. In this method, we search relative point by minimizing Energy pan-function to match signatures for entry or signature for entry with signature for the judge. We define the "characteristic part" as the part in that stroke direction is similar and complex. And we use genetic algorithm to extract the "characteristic part". An experiment to evaluate the effectiveness of the proposed method achieved the 96% rate of verification.