Biometric Region Detection Using Multi-Stage ROI Classifiers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current biometric authentication methods face challenges in accurately detecting biometric regions, particularly in varying sizes and complexities, leading to inefficiencies and increased rates of misdetection.
Innovation Solution
A method involving multiple classifiers with different ROI sizes, each trained for specific stages, calculates similarity values to detect biometric regions by setting initial and subsequent ROIs of varying sizes, and determines candidate regions based on thresholds learned from these sizes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single fixed-size ROI is used for biometric detection, then the device complexity is low, but the measurement precision and detection accuracy deteriorate due to inability to adapt to varying biometric region sizes
Solution Approach 1:
The detection process is segmented into multiple stages, with each stage having a dedicated classifier trained on a specific ROI size. This segmentation allows each classifier to specialize in detecting biometric regions at particular scales, improving overall detection accuracy while maintaining manageable complexity through modular architecture
Solution Approach 2:
The system dynamically adapts to varying biometric region sizes by selecting and applying appropriate classifiers based on the detected scale. Rather than using a static fixed-size ROI, the system adjusts the detection approach according to the actual size of the biometric region, enabling accurate detection across different scales
2Measurement precision
If multiple classifiers with different ROI sizes are implemented, then the biometric region detection accuracy improves, but the processing time and computational complexity increase
Solution Approach 1:
Classifiers are pre-trained on specifically sized ROIs during the offline phase, so that during online detection, the system can directly apply these pre-configured classifiers without performing time-consuming training. This preliminary preparation reduces real-time processing requirements while maintaining high detection accuracy
Solution Approach 2:
The system applies only the necessary subset of classifiers required for the current detection task rather than exhaustively applying all possible classifiers. By selecting and applying only the relevant classifiers for the detected scale, the system achieves accurate detection while minimizing unnecessary computational overhead and processing time
Data Source
AI summary
A method of detecting a biometric region, the method including receiving an input image; setting a plurality of regions of interest (ROI) corresponding to a plurality classifiers, respectively, the plurality of classifiers corresponding to a plurality of multi-stages, respectively, the plurality of ROIs having different sizes with respect to each other, each ROI of the plurality of ROIs including a biometric region of the input image; calculating a plurality of similarity values for the plurality of classifiers, respectively, based on the ROIs corresponding to the plurality of classifiers from among the plurality of ROIs; and detecting a first biometric region for a user authentication based on at least one of the plurality of similarity values.