Biometric Authentication Device Using Non-Directional Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biometrics authentication systems face increased False Acceptance Rate (FAR) when physically separating palm print features from images fails, especially when melanin is abnormally deposited, leading to higher inclusion of palm print features in biological information.
Innovation Solution
A biometrics authentication device and method that extracts non-directional features from images using Gabor filtering and normalization, suppressing the influence of palm prints while emphasizing vein features, thereby reducing FAR.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physical separation methods (polarizing filter or plural-wavelength photographing) are used to separate palm print features, then the FAR is reduced, but the device complexity and manufacturing difficulty increase
Solution Approach 1:
The invention extracts only the vein features from the photographed image by suppressing palm print features through non-directional feature extraction and normalization processing, without using physical separation methods like polarizing filters or plural-wavelength photographing
Solution Approach 2:
The invention changes the parameter of feature extraction by using non-directional Gabor filtering followed by normalization processing, which transforms the feature extraction approach to suppress palm print features while emphasizing vein features
2Measurement precision
If physical separation methods are used to separate palm print features, then the authentication accuracy is improved, but the ease of operation and ease of manufacture deteriorate
Solution Approach 1:
The invention extracts only the vein features from the photographed image by suppressing palm print features through non-directional feature extraction and normalization processing, without using physical separation methods
Solution Approach 2:
The invention replaces mechanical/optical physical separation methods (polarizing filters, plural-wavelength photographing) with image processing techniques (non-directional Gabor filtering and normalization) to achieve the same goal of separating vein features from palm print features
3Device complexity
If palm print features are included in biological information, then the device complexity is reduced, but the FAR increases especially when melanin is abnormally deposited
Solution Approach 1:
The invention applies different processing characteristics to different features: non-directional Gabor filtering suppresses palm print features while normalization processing emphasizes vein features, creating local quality differences in feature extraction
Solution Approach 2:
The invention changes the parameter of feature extraction by using non-directional Gabor filtering followed by normalization processing, which transforms the feature extraction approach to suppress palm print features while emphasizing vein features
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution effectively prevents the increase in FAR even when physical separation of palm print features is not applied, improving authentication accuracy and reducing false positives by emphasizing the more diverse vein features.
Implementation Method 1
a non-directional feature is extracted from an image f of the specified ROI. In a case in which filtering S is performed on an image f
Implementation Method 2
a per-direction directional feature normalization processing unit 42 normalizes respective directional features ge
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A biometrics authentication device 1 is configured so as to include: a filter 41 that extracts a plurality of directional features from an input image; a per-direction directional feature normalization processing unit 42 that normalizes the plurality of directional features extracted by the filter 41; a non-directional feature generating unit 43 to 45 that generates a non-directional feature on the basis of the plurality of directional features output from the per-direction directional feature normalization processing unit 42; a matching processing unit 5 that determines a degree of similarity between the non-directional feature and a registered non-directional feature stored in a storing unit 7; and a determining unit 6 that determines identity by using the degree of similarity.