Fuzzy geometrical features for identifying distorted overlapping fingerprints

Author

Pal, Sankar K. ; Sarbadhikari, Suptendra Nath

Author_Institution

Machine Intelligence Unit, Indian Stat. Inst., Calcutta, India

Volume

3

fYear

1997

fDate

9-12 Sep 1997

Firstpage

1527

Abstract

Overlapping fingerprints are naturally abundant and pose great difficulty for proper identification. We demonstrate the effectiveness of fuzzy geometrical features for classifying distorted overlapping fingerprints directly from raw unprocessed images. The fuzzy geometrical features viz., length, height and index of area coverage are found to be the best for classifying these patterns when Bayes´, k-NN (with k=1, 3, 5) and MLP (multilayer perceptron) classifiers are used. The overall performance is best for MLP, followed by 1-NN

Keywords

Bayes methods; feature extraction; fingerprint identification; fuzzy systems; image classification; multilayer perceptrons; Bayes´ classifiers; classifying distorted overlapping fingerprints; distorted overlapping fingerprints; fuzzy geometrical features; height; identification; index of area coverage; k-NN classifiers; length; multilayer perceptron; nearest neighbor classifier; pattern recognition; performance; raw unprocessed images; Concurrent computing; Fingerprint recognition; Fuzzy sets; Image matching; Law enforcement; Machine intelligence; Neural networks; Pattern matching; Pattern recognition; Uncertainty;

fLanguage

English

Publisher

ieee

Conference_Titel

Information, Communications and Signal Processing, 1997. ICICS., Proceedings of 1997 International Conference on

Print_ISBN

0-7803-3676-3

Type

conf

DOI

10.1109/ICICS.1997.652249

Filename

652249