Fingerprint Recognition Using Frequency Domain Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fingerprint recognition methods on mobile communication terminals face inaccuracies due to reliance on minutiae matching, which is affected by rotation, distortion, and the limited size of the sensor, resulting in decreased recognition accuracy.
Innovation Solution
A method and apparatus for fingerprint recognition that acquires slice images, converts them into block images, eliminates high-frequency components using a low-pass filter, and forms a fingerprint pattern template, allowing for ridge flow direction adjustment and hashing to improve matching accuracy, unaffected by rotation and distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If minutiae matching method is used for fingerprint recognition, then the recognition process is simple, but recognition accuracy decreases due to rotation and distortion
Solution Approach 1:
The patent transforms the fingerprint image from spatial domain to frequency domain using 2D DCT (Discrete Cosine Transform). This parameter change in the representation domain allows the system to capture global pattern information while being invariant to rotation and translation, thereby improving recognition accuracy without significantly increasing system complexity
Solution Approach 2:
The patent replaces the traditional minutiae-based mechanical matching approach with a frequency-domain pattern matching approach. Instead of detecting and matching specific minutiae points (ending, bifurcation), the system uses DCT coefficients to represent the overall fingerprint pattern, making the recognition process more robust to distortion
2Area of moving object
If a small-sized bar-shaped sensor is used, then the device size is minimized, but the entire fingerprint image cannot be acquired
Solution Approach 1:
The patent extracts the essential frequency domain characteristics (DCT coefficients) from the partial fingerprint image captured by the small sensor. By transforming to frequency domain, the system can capture global pattern information even from limited spatial coverage, effectively extracting the most important recognition features without requiring the entire fingerprint image
Solution Approach 2:
The patent transitions from spatial domain representation to frequency domain representation, adding a new dimension for analyzing fingerprint patterns. This dimensional change allows the system to utilize global pattern information from partial images that would be insufficient in the spatial domain
3Area of moving object
If conventional sliding method is used to acquire fingerprint images, then the sensor size is minimized, but recognition accuracy decreases due to rotation and distortion
Solution Approach 1:
The patent applies 2D DCT transformation to convert the fingerprint image from spatial domain to frequency domain. This parameter change makes the representation invariant to rotation and translation, thereby improving recognition reliability while maintaining the use of compact sensors with sliding acquisition method
Solution Approach 2:
The patent creates a frequency domain copy (DCT coefficient representation) of the spatial fingerprint image. This copied representation in frequency domain preserves the essential pattern information while being immune to rotation and distortion, improving reliability of recognition
Data Source
AI summary
According to one embodiment of the present invention, provided is a method for fingerprint recognition and authentication, comprising: a step for sequentially acquiring a plurality of slice images including a fingerprint pattern via a fingerprint sensor; a slice dividing unit for dividing the slice images into block images; a step for converting spatial domain information of the block images into frequency domain information, and eliminating information of a high-frequency component by using a low-pass filter; and a step for forming a fingerprint pattern template by matching the block images, from which the information of the high-frequency component has been eliminated.