Biometric Authentication Using Joint Angle Model Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biometric authentication systems face challenges in accurately matching biometric features when the shape of a biometric instance, such as a palm, changes due to varying joint angles between the enrollment and authentication processes, leading to potential miscomparisons.
Innovation Solution
A biometric authentication device that generates model images by simulating different joint angles of a biometric model, captures a biometric image, determines similarities between the captured image and model images, selects the most similar model image, and corrects the image based on the corresponding joint angles to align with the enrollment data, thereby improving authentication accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If biometric authentication is performed using captured biometric images, then authentication can be performed with high accuracy, but authentication accuracy deteriorates when joint angles change between enrollment and authentication processes
Solution Approach 1:
The system pre-generates multiple model images corresponding to different joint angle combinations before authentication. These model images are stored in advance and used for comparison during authentication, eliminating the need to generate models in real-time and enabling quick selection of the best-matching model based on captured images.
Solution Approach 2:
The system changes the parameter of joint angles to generate multiple model images with different angle combinations. By varying joint angle parameters in the biometric model, the system creates a comprehensive set of reference images that cover possible pose variations, allowing accurate authentication regardless of the actual joint angles during capture.
2Measurement precision
If multiple model images are generated by changing joint angles, then authentication accuracy is improved, but device complexity increases
Solution Approach 1:
Instead of using complex real-time 3D rendering or multiple physical sensors, the system creates 2D image copies from a biometric model with varied joint angles. These copied images serve as reference models for authentication, simplifying the overall system while maintaining accuracy across different poses.
Solution Approach 2:
The system dynamically selects the most appropriate model image from the pre-generated set based on the captured biometric image characteristics. This dynamic selection process adapts to different joint angle configurations without requiring real-time model generation, balancing complexity and accuracy.
3Measurement precision
If model images are pre-generated for different joint angles, then matching accuracy is improved, but storage requirements increase
Solution Approach 1:
The biometric model is segmented into multiple configurations based on joint angle combinations. Each segment (model image) represents a specific pose configuration, and only necessary segments are stored and used during authentication. This segmentation allows efficient storage by organizing data into manageable, purpose-specific units.
Solution Approach 2:
A single biometric model serves multiple functions by generating different model images for various joint angles. This multi-functional approach allows one base model to create numerous reference images, reducing the need to store separate models for each pose and optimizing storage efficiency.
Data Source
AI summary
A biometric authentication device performing an authentication based on a similarity between a biometric image that is an object of comparing and an enrolled biometric image, includes: a storage configured to store a plurality of model images generated by changing a bending angle of a joint of a biometric model and correction information of each of the plurality of model images; a biometric sensor configured to capture a biometric image that is an object of comparing; and a processor configured to execute a process, the process including: determining similarities between the biometric image captured by the biometric sensor and the plurality of model images; selecting a model image based on the similarities; reading correction information corresponding to the model image that is selected, from the storage; and correcting one of the biometric image captured by the biometric sensor or the enrolled biometric image based on the correction information.