Indexing of Fingerprints for efficient retrieval

1
Indexing of Fingerprints for efficient retrieval

• Roshith Rajagopal
• Prathap Adusumilli

2
CLASSIFICATION

• Loop, whorl, arch, tented arc
• Classification insufficient
• 31.7 R, 33.8 L, 27.9 W,
• 3.7 arch, 2.9 TA
• Hence indexing
• 2Bir Bhanu

3
Past Work

• Continuous Classification Approach 1
• Indexing based Minutiae Triplets 2
• 1 A. Lumini, D. Maio and D. Maltoni _at_
  University of Bolonga.
• 2 Fingerprint Indexing on features of Minutiae
  Triplets

4
Continuous classification

• Performed at University of Bolonga
• Uses KL transfrom
• Fingerprint stored as a numerical vector
• The vectors are computed by reducing the
  dimensionality of the fingerprint directional
  images through the Karhunen-Loève transform

5
Triplet based Minutiae Method

• Uses triplets of minutiae points
• Generates triangles of minutiae points
• H(?min, ?max, ?,?,?,?)
• max and min angles of triangle formed
• ? triangle handedness (1 or -1)
• Type of minutiae Endpoint or bifurcation(1 or
  0)
• Minutiae Ridge ending?
• ? Length of max side
• Outputs Top N hypothesis

6
Indexing method based on Matching scores

• Matching scores between each pair of fingerprints
  in the database is maintained
• From Maeda, Matsushita and Sasakawa (2001)