Description of signature images and its application to theirclassification

摘要

A method for signature description based on the human way ofrecognizing objects is introduced. Like humans, this method firstrecognizes objects globally and then recognizes finer details. Thedescription is hierarchical: the signature is at the top level, and eachlower level gives more details about the higher one. Features areextracted in a relative way (e.g. the length is measured by its ratio tothe width) to make the method insensitive to the natural variations thatoccur in genuine signatures, and are mapped into a conceptualdescription. This conceptual description is then used to classify thesignature images