A hierarchical k-means clustering based fingerprint quality classification

作者：Munir Muhammad Umer^*; Javed Muhammad Younus; Khan Shoab Ahmad

来源：Neurocomputing, 2012, 85: 62-67.

DOI：10.1016/j.neucom.2012.01.002

• Biometrics
• Fingerprint matching
• k-means clustering
• Quality classification

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

• 出版日期2012-5-15