Fingerprint Image Detail Feature Extraction and MATLAB Implementation

This paper discusses fingerprint image feature extraction, which serves as the cornerstone of fingerprint recognition systems. Fingerprint matching typically operates through minutiae point comparison. During the feature extraction process, we obtain minutiae features - specifically endpoints and bifurcation points - from thinned fingerprint images. However, these extracted features frequently contain substantial false minutiae that both increase processing time and compromise matching accuracy. To address this challenge, our methodology implements dual false feature removal techniques: edge-based elimination and distance-based filtration. These algorithms effectively reduce false minutiae by approximately one-third while preserving genuine features for reliable fingerprint matching. We present a MATLAB-based implementation for fingerprint detail feature extraction, incorporating our false feature removal algorithm. The implementation leverages MATLAB's image processing toolbox for efficient ridge thinning operations and connectivity analysis. The algorithm employs 3x3 pixel neighborhood scanning to classify minutiae types, followed by Euclidean distance calculations between potential features to eliminate clustered false points. Edge-based removal utilizes boundary detection to discard features near image borders where thinning artifacts commonly occur. This approach demonstrates simplicity in code structure, rapid execution through vectorized operations, and high accuracy in minutiae detection. Overall, this method provides a robust fingerprint feature extraction solution suitable for widespread application in biometric recognition systems.