Abstract:In this paper, we present an efficient fingerprint classification algorithm which is an essential component in many critical security application systems e.g. systems in the e-government and e-finance domains. Fingerprint identification is one of the most important security requirements in homeland security systems such as personnel screening and anti-money laundering. The problem of fingerprint identification involves searching (matching) the fingerprint of a person against each of the fingerprints of all registered persons. To enhance performance and reliability, a common approach is to reduce the search space by firstly classifying the fingerprints and then performing the search in the respective class. Jain et al. proposed a fingerprint classification algorithm based on a two-stage classifier, which uses a K-nearest neighbor classifier in its first stage. The fingerprint classification algorithm is based on the fingercode representation which is an encoding of fingerprints that has been demonstrated to be an effective fingerprint biometric scheme because of its ability to capture both local and global details in a fingerprint image. We enhance this approach by improving the efficiency of the K-nearest neighbor classifier for fingercode-based fingerprint classification. Our research firstly investigates the various fast search algorithms in vector quantization (VQ) and the potential application in fingerprint classification, and then proposes two efficient algorithms based on the pyramid-based search algorithms in VQ. Experimental results on DB1 of FVC 2004 demonstrate that our algorithms can outperform the full search algorithm and the original pyramid-based search algorithms in terms of computational efficiency without sacrificing accuracy.

Weight Competitive Coding for Finger-Knuckle-Print Verification

Reconstruction in Gabor Response Domain for Efficient Finger-knuckle-Print Verification

Suspecting Less and Doing Better: New Insights on Palmprint Identification for Faster and More Accurate Matching

Palmprint verification using consistent orientation coding

Comparative Competitive Coding for Personal Identification by Using Finger Vein and Finger Dorsal Texture Fusion

Multimodal Biometric Identification System Based on Finger Geometry, Knuckle Print and Palm Print

Fast Palmprint Verification using Derivative Filter Code

Efficient Fingercode Classification

A New Biometric System Based on Inner-Knuckle-print Recognition

Palmprint Verification Using Binary Orientation Co-Occurrence Vector

The Multiscale Competitive Code Via Sparse Representation For Palmprint Verification

Fast and Robust Personal Identification by Fusion of Finger Vein and Finger-Knuckle-Print Images

Finger Vein Verification Based on Neighbor Pattern Coding.

Joint Palmprint and Palmvein Verification by Dual Competitive Coding

Finger Vein Recognition by Generating Code

A Novel 2D Contactless Fingerprint Matching Method

An efficient algorithm for fingercode-based biometric identification

CompNet: Competitive Neural Network for Palmprint Recognition Using Learnable Gabor Kernels

Learning Compact Multifeature Codes for Palmprint Recognition From a Single Training Image per Palm

A unified distance measurement for orientation coding in palmprint verification

A Hybrid Finger Identification Pattern Using Polarized Depth-Weighted Binary Direction Coding