Fingerprint Recognition Using Frequency Domain Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge is to develop a fingerprint recognition method that effectively operates with smaller fingerprint sensing areas, where the existing techniques face difficulties in accurately enrolling and recognizing fingerprints due to reduced sensing area sizes, necessitating innovative image processing and matching techniques.
Innovation Solution
The proposed method involves generating enrollment modified images by modifying fingerprint images in both spatial and frequency domains, extracting property information, and storing this information for matching with input fingerprint images, which includes determining similarity scores and translation, rotation, and scale information to accurately recognize fingerprints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If the fingerprint sensing area is reduced to downsize portable devices, then device portability is improved, but fingerprint recognition accuracy deteriorates
Solution Approach 1:
The patent transforms the fingerprint matching problem from spatial domain comparison to frequency domain analysis. By converting fingerprint images to frequency domain representations and comparing spectral characteristics, the system achieves accurate recognition even when spatial details are limited by small sensor area. This dimensional transformation allows extraction of discriminative features that are invariant to the reduced sensing area.
Solution Approach 2:
The patent modifies fingerprint images through various transformations including frequency domain conversion, filtering, and feature extraction. By changing the representation parameters of fingerprint data from raw spatial images to frequency-domain spectral features, the system maintains recognition accuracy despite reduced input data from smaller sensors.
2Volume of moving object
If only partial fingerprint data is captured due to small sensing area, then device portability is improved, but enrollment and recognition reliability deteriorates
Solution Approach 1:
The patent extracts essential frequency-domain features and spectral characteristics from partial fingerprint images captured by small sensors. By isolating and comparing key frequency components rather than attempting to process complete spatial images, the system achieves reliable recognition from incomplete data. This extraction approach focuses on discriminative features that remain detectable even in limited sensing areas.
Solution Approach 2:
The patent introduces frequency domain transformation as an intermediary process between fingerprint capture and recognition. This intermediate representation converts limited spatial information into a form where partial data can be effectively compared, serving as a bridge that enables reliable recognition despite incomplete input from small sensors.
3Device complexity
If conventional fingerprint matching techniques are used with small sensing areas, then system simplicity is maintained, but matching accuracy deteriorates
Solution Approach 1:
The patent replaces conventional spatial-domain image matching mechanisms with frequency-domain spectral analysis. Instead of directly comparing fingerprint images in spatial domain, the system transforms images to frequency domain and compares spectral characteristics, providing a more effective matching mechanism for small sensing areas while maintaining computational feasibility.
Data Source
AI summary
A fingerprint recognition method includes generating an enrollment modified image by modifying a fingerprint image corresponding to a fingerprint to be enrolled; extracting enrollment property information from the fingerprint image; generating mapping information that maps the enrollment modified image to the enrollment property information; and storing the enrollment modified image and the enrollment property information.