Letters: A hierarchical k-means clustering based fingerprint quality classification

摘要: This paper presents a novel technique that employs hierarchical k-means clustering for quality based classification of fingerprints subsequent improvement in fingerprint matching results. A set statistical and frequency features have been calculated from image. algorithm has utilized to classify the image into one four classes, i.e. good, dry, normal or wet. An objective method also proposed evaluate performance classification. It shown through experimental results minutiae matcher improves when is incorporated stage. The false accept rate reject are 1.8 on FVC 2002 db1 database without utilizing information. False reduced 0.79 whereas at threshold value utilized. significant system shows effectiveness fingerprints.