摘要：

Some MATCH-ON-CARD implementations emit only a few unique matcher scores. This precludes fine-grained setting of the operating threshold. This disadvantages those implementations because overly conservative FMR values give rise to elevated FNMR. The late 2008 implementations produce many more unique matcher scores than in late 2007. This increase allows users to more finely target false match rates. Some cards indicated a capability to accept no more than 55 minutiae, and this necessitates removal of minutiae from either or both of the reference and verification templates prior to transmission to the card in 16.5 % of genuine comparisons. The use of one implementation's minutia quality values for removal is superior to using radial distance alone. MINEX II attained unprecedented transparency in its execution: the evaluation plan was published during its development with industry, and version-controlled open-source software was released for both conformance and conversion of INCITS 378 and ISO/IEC 19794-2 COMPACT CARD templates a, and for invocation of ISO/IEC 7816 MATCH-ON-CARD operations. a These standards are available from

展开