Improving Identity Prediction in Signature-based Unimodal Systems Using Soft Biometrics

System optimisation, where even small individual system performance gains can often have significant impact on applicability and viability of biometric solutions, is an important practical issue. This paper analyses two different techniques for using soft biometric information (which is often already available or easily obtainable in many applications) improve identity prediction accuracy of signature-based tasks. It, is shown that such a strategy can improve performance of unimodal systems, supporting high usability profiles mid low-cost processing.