Enhancing off-line biometric signature verification using a fingerprint assessment approach

A method designed for matching biometric fingerprint images is applied to the static/image-based ”off-line” human signature modality. Using a publically available signature dataset, the verification performance is compared against three existing static methods. Furthermore, verification is assessed using all four methods within a multi-classifier system. The results show the application of the fingerprint method leads to a comparable performance with existing methods and a significant improvement is achieved within a multi-classifier configuration.