Palm Liveness Detection Using Image Variations and ROI Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional biometric liveness detection systems face challenges in distinguishing between real and fake biometric information, particularly due to the large differences among fake biometric types, and are prone to unreliable outcomes caused by varying imaging conditions, which affect the quality and size of detected palm regions, leading to inefficient liveness detection.
Innovation Solution
The system generates variations of input palm images under different conditions and uses specific liveness detectors for each type of fake biometric, combining individual detection results to enhance robustness and accuracy, and adjusts region of interest (ROI) dimensions without padding with zero pixels to maintain contextual information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single liveness detector is used to distinguish real and fake biometric information, then the device complexity is low, but the reliability of liveness detection deteriorates due to large differences among fake biometric types and varying imaging conditions
Solution Approach 1:
The patent divides the liveness detection system into multiple specialized detectors, each trained to detect specific types of fake biometric information (e.g., photo-based fakes, video-based fakes, 3D mold-based fakes). This segmentation allows each detector to focus on particular欺骗 patterns, improving overall detection reliability while maintaining manageable complexity through modular design
Solution Approach 2:
The system varies detection parameters by generating multiple versions of the input biometric image under different conditions (e.g., different lighting, angles, or preprocessing transformations) and processes each version with appropriate detectors. This parameter variation enhances the system's ability to detect fakes under diverse imaging conditions while maintaining robust performance
2Measurement precision
If imaging conditions are not controlled, then the ease of operation is high (user can place palm freely), but the measurement precision of palm region deteriorates due to varying image quality and size
Solution Approach 1:
The system dynamically adjusts to varying imaging conditions by implementing adaptive palm region detection that can identify and extract the palm region regardless of its position, size, or orientation in the captured image. This dynamic adaptation maintains measurement precision while allowing users to place their palms freely without strict positioning requirements
3Manufacturing precision
If ROI dimensions are adjusted by padding with zero pixels, then the manufacturing precision of ROI dimensions is improved, but the loss of information increases due to removal of contextual information
Solution Approach 1:
Instead of uniformly padding the ROI with zero pixels, the system applies local quality preservation by selectively retaining or enhancing specific regions of the palm image based on their informational value. This approach maintains precise ROI dimensions while preserving critical contextual information such as palm lines, creases, and surrounding skin texture that are essential for accurate liveness detection
Data Source
AI summary
The present teaching relates to method, system, medium, and implementations for detecting liveness. An input image is received with visual information claimed to represent a palm of a person. One or more variations of the input image are generated based on one or more models specifying conditions to generate the variations. The liveness of the visual information is detected based on the input image and the one or more variations thereof.