Sweat pores-based (level 3) novel fingerprint quality estimation

Poor-quality images mostly result in spurious or missing features, which further degrade the overall performance of the recognition systems. This work augments the fingerprint quality with respect to one of the level 3 micro features, i.e., sweat pores. The paper proposes sweat pores-based an effective fingerprint quality estimation scheme, which is a multi-factor quality assessment model comprises five quality measures estimating their respective quality scores. The scores are then fused into a single quality score by a fusion engine based on the ‘Weighted Sum rule’. The proposed scheme is experimented with publicly available fingerprint datasets, which were scanned with Cross Match Verifier 300 scanner at 500 dpi (pores are quite visible in these samples, as of now, no 1000ppi datasets are publicly available). The experimental results show that this arrangement could correctly estimate and assign grades (good/acceptable/poor) to nearly 93.33% images in the dataset. This paper also presents a novel technique for identifying the strong pores for high reproducibility. The corresponding result is shown for the same set of images and is found as 91.93%. This method is also being tried with higher resolution images.