Fingerprint Authentication Using Frequency Domain Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fingerprint authentication in mobile devices is unstable due to the miniaturization of fingerprint sensors, leading to variations in image acquisition conditions such as position and angle, which affects the collation information generation.
Innovation Solution
An information processing method that determines a base point and reference direction from the fingerprint image, acquires samples along a circular path, and calculates frequency components using the Yule-Walker method without a window function to generate collation information unaffected by acquisition conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Volume of moving object
If the fingerprint sensor is miniaturized to fit mobile devices, then the device size is reduced, but the image size becomes smaller leading to unstable authentication
Solution Approach 1:
The patent transforms the fingerprint image from spatial domain to frequency domain using Fourier transform, changing the representation parameters from pixel coordinates to frequency components. This parameter transformation makes the collation information insensitive to positional and angular variations caused by miniaturization, thereby maintaining authentication stability despite smaller sensor size.
Solution Approach 2:
The patent replaces traditional mechanical/image-based comparison methods with frequency domain analysis. Instead of comparing spatial patterns directly, the system uses frequency spectrum conversion to extract invariant characteristics, substituting the mechanical image matching approach with a mathematical transformation-based method that is robust to acquisition condition variations.
2Ease of operation
If the user touches the fingerprint sensor with the finger of the hand holding the mobile device, then the input operation becomes more convenient, but the image acquisition conditions (position and angle) become unstable
Solution Approach 1:
By converting to frequency domain representation, the patent creates collation information whose parameters (frequency components) are invariant to positional and angular shifts. This allows the system to maintain consistent authentication performance even when image acquisition conditions vary due to convenient but imprecise user operation.
Solution Approach 2:
The patent extracts the essential frequency domain characteristics from the fingerprint image, separating the invariant authentication features from the variable acquisition conditions. By taking out only the frequency spectrum information and discarding the spatial position and orientation dependencies, the system achieves robustness against operation-induced variations.
3Reliability
If frequency spectrum conversion is applied to generate collation information, then the authentication becomes less affected by disturbance, but the processing complexity increases
Solution Approach 1:
The patent replaces complex spatial pattern matching algorithms with Fourier transform-based frequency domain analysis. This substitution simplifies the processing by using well-established mathematical transforms that efficiently extract invariant features, reducing the overall system complexity while maintaining high disturbance resistance.
Data Source
AI summary
A processor of an information processing device acquires an image, determines base point from the image, and acquires first position information of the base point. The processor determines a reference direction indicating characteristics of color information of a section of the image around the base point. The processor acquires a sample for each of a plurality of reference points acquired sequentially in accordance with a predetermined condition from a starting point determined on the basis of the base point and the reference direction. The processor calculates as frequency information, frequency components of changes in the color information with respect to the second position information for the plurality of samples, using a linear prediction coefficient calculated using a Yule-Walker method without applying a window function. The processor causes the memory to store information associating the frequency information, the first position information and the reference direction, as collation information.