Source-linked AI summary
Fingerprint Recognition Using Minutia Score Matching
Ravi. J, K. B. Raja, Venugopal. K. R
TL;DR
Fingerprint recognition requires reliable matching of permanent physiological patterns, but fingerprint images must be processed and compared using discriminative minutiae. The paper applies Fingerprint Recognition using Minutia Score Matching (FRMSM), including binarization, thinning, noise removal, and thresholded minutiae matching. Compared with FRFNN, FRMSM reports the same zero FNMR and a lower FMR of 0.026 versus 0.23.
Problem
Fingerprint authentication requires matching permanent fingerprint patterns despite the need to process images and extract discriminative minutiae.
Method
FRMSM preprocesses fingerprint images through binarization, ridge thinning, and noise removal, then matches minutiae between input and template images using a normalized score.
Results
False Matching Ratio was 0.026 for FRMSM versus 0.23 for FRFNN, while False Non Matching Ratio was zero for both methods.
Takeaways & Limitations
FRMSM produced better False Matching Ratio values than the existing FRFNN method in the reported fingerprint-recognition comparison.
Abstract
from arXiv · showhide
The popular Biometric used to authenticate a person is Fingerprint which is unique and permanent throughout a person's life. A minutia matching is widely used for fingerprint recognition and can be classified as ridge ending and ridge bifurcation. In this paper we projected Fingerprint Recognition using Minutia Score Matching method (FRMSM). For Fingerprint thinning, the Block Filter is used, which scans the image at the boundary to preserves the quality of the image and extract the minutiae from the thinned image. The false matching ratio is better compared to the existing algorithm.
II. RELATED WORK
Prior work explored fingerprint identification through ridge detection, enhancement, and processing of scanned or inked impressions.
- A gray-level watershed method was proposed to identify ridges in directly scanned or inked fingerprint impressions.
ISSN: 0975-5462
Related studies investigated fingerprint enhancement, sensing conditions, orientation fields, graph-based matching, classification, and minutiae extraction. These approaches used diverse image-processing and matching strategies, including diffusion, local operators, graph algorithms, and filtered normalization.
- Oriented diffusion enhanced ridge patterns by smoothing along ridge flow while preserving individual ridges and valleys.
- Local operators including Sobel, Roberts, Prewitt, Canny, and LoG were evaluated for edge detection before fingerprint segmentation.
- Force level significantly affected image-quality scores across optical and capacitance sensors, while capacitance-sensor minutiae counts showed no significant force-level difference.
- Graph-based and hybrid methods addressed fingerprint matching, orientation estimation, principal-curve extraction, and minutiae-based recognition.
- Fingerprint classification studies used direction fields, Poincare-index singularity searches, and directional images to characterize ridge structure.
- A minutiae-extraction pipeline normalized 64 x 64 regions, enhanced ridge contrast with a Gabor filter, and scanned the processed image for transitions.
III. MODEL
The FRMSM model defines fingerprint features and error measures, then compares minutiae from input and template images using thresholded matching scores. Its score is normalized between 0 and 1, where the endpoints represent perfect matching and complete mismatch.
- The model defines termination and bifurcation as ridge endings and splits, alongside binarization, thinning, and their associated angles.Thinning reduces each ridge to one pixel in width.
- False Matching Ratio measures the probability of granting access to an imposter during imposter attempts.An imposter match is recorded when the matching score exceeds the established threshold.
- False Non Matching Ratio measures the probability of denying access to an approved user.
- Enrollee attempts use one-to-one matching between each input image and its corresponding template, recording a false non-match below the threshold.
- The matching score compares input and template minutiae, using template and input minutiae counts, and ranges from 0 to 1.A score of 1 indicates a perfect match, while 0 indicates complete mismatch.
B. Model
FRMSM preprocesses fingerprint images, extracts termination and bifurcation minutiae, and matches template and input minutiae using spatial and angular information.
- FRMSM matches a test fingerprint with a template database using minutia matching scores.
- Pre-processing: The input fingerprint is binarized by thresholding grayscale intensities into binary pixel values.
- Pre-processing: Block Filter thinning reduces ridge lines to one-pixel width while preserving minutiae location and orientation.Boundary pixels are preserved by assigning white values to the outer rows and columns, with dilation and erosion used for thinning.
- Minutiae extraction: Crossing Number classifies minutiae as termination, normal ridge, or bifurcation when its value is 1, 2, or at least 3.
- Minutiae matching: The matching matrix stores each minutia’s row index, column index, orientation angle, and type for template-input comparison.Template and input points are converted to polar coordinates relative to reference points, while orientation differences are retained.
IV. ALGORITHM
The FRMSM algorithm preprocesses a test fingerprint, extracts its minutiae, and compares it with a template fingerprint for recognition.
- The algorithm defines three objectives: preprocess the test fingerprint, extract minutiae points, and match the test image with the database.
- FRMSM compares an input test fingerprint image with a template fingerprint image for recognition.
V. PERFORMANCE ANALYSIS AND RESULTS
Performance analysis uses fingerprint databases containing left-loop, right-loop, whorl, and arch patterns. FRMSM is compared with FRFNN using FNMR and FMR.
- The evaluation uses fingerprint database images representing left-loop, right-loop, whorl, and arch patterns.
- 0.026 FMR is reported for FRMSM versus 0.23 for FRFNN, while both methods have zero FNMR.
VI. CONCLUSION
The paper presents fingerprint matching using FRMSM, with preprocessing that includes binarization, ridge thinning, and noise removal. FRMSM reportedly achieves better false matching rate values than the existing method.
- FRMSM is presented for matching fingerprint minutia points.
- The preprocessing pipeline includes image binarization, ridge thinning, and noise removal.
- FRMSM gives better false matching rate values than the existing method.