Fingerprint indexing based on Minutia Cylinder-Code

Area of Science:

• Biometrics
• Computer Science
• Pattern Recognition

Background:

• Fingerprint identification systems require efficient indexing for large databases.
• Existing methods often rely on complex feature sets, impacting performance.

Purpose of the Study:

• To develop a novel hash-based indexing method for accelerated fingerprint identification.
• To improve the efficiency and accuracy of large-scale fingerprint matching.

Main Methods:

• A Locality-Sensitive Hashing (LSH) scheme was designed using Minutiae Cylinder-Code (MCC).
• MCC maps minutiae-based representations to fixed-length, transformation-invariant binary vectors.
• A new search algorithm was developed based on a numerical approximation of MCC vector similarity.

Main Results:

• The proposed method was compared against 15 existing fingerprint indexing techniques.
• The new approach demonstrated superior performance across typical fingerprint indexing benchmarks.
• Outperformance was achieved despite utilizing a smaller feature set compared to other leading methods.

Conclusions:

• The novel hash-based indexing method significantly enhances fingerprint identification speed.
• The LSH scheme with MCC offers an effective and efficient solution for large-scale biometric databases.
• This approach represents a promising advancement in biometric security and identification.