Multimodal Biometric Fusion via Feature Subset Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multimodal biometric systems face difficulties in fusing palmprint and hand-shape features at the feature level due to incompatibility of feature spaces, and lack the use of feature subset algorithms to eliminate redundant information.
Innovation Solution
A multimodal biometric system that fuses palmprint and hand-shape features at the feature level using a feature subset algorithm to select and combine relevant features, reducing redundancy and improving data storage and processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If palmprint and hand-shape features are fused at the feature level, then recognition accuracy is improved, but feature space incompatibility makes the fusion process more difficult
Solution Approach 1:
The patent transforms the feature representation by converting both palmprint and hand-shape features into a unified feature space through parameter transformation. This allows features from different modalities to be compatible and fusible by changing their mathematical representation form, resolving the incompatibility issue while maintaining recognition accuracy.
Solution Approach 2:
The patent introduces an intermediary transformation layer that mediates between the original palmprint and hand-shape feature spaces. This intermediary mechanism converts features into a common representation format, enabling their fusion without direct combination of incompatible feature spaces.
2Reliability
If all extracted features are used for recognition, then comprehensive information is available, but redundant features increase computational complexity and storage requirements
Solution Approach 1:
The patent extracts only the essential and non-redundant features from the complete feature set through feature selection mechanisms. By taking out and retaining only the most discriminative features, the system maintains information completeness for reliable recognition while eliminating redundant features that would increase computational burden.
Solution Approach 2:
The patent applies partial action by using a subset of the most important features rather than all available features. This selective approach provides sufficient information for accurate recognition without the excessive computational complexity associated with processing the complete feature set.
3Reliability
If multiple biometric modalities are integrated, then system reliability is improved, but the difficulty of feature-level combination increases
Solution Approach 1:
The patent resolves the difficulty of combining features from multiple biometric modalities by transforming them into a unified parameter space. This parameter transformation enables compatible representation of palmprint, hand-shape, and other biometric features, making their fusion straightforward while maintaining the reliability benefits of multi-modal integration.
Data Source
AI summary
The present invention relates to a method and system for a multimodal biometric system utilizing a single image to generate hand shape and palmprint features. The invention utilizes a digital camera, and incorporates feature subset selection algorithms to eliminate redundant data. The inventions, through the use of feature algorithm, successfully fuses the hand shape features and palmprint features at the features level.