Palm Liveness Detection via Segmented Fake Detectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional liveness detection systems in biometric authentication face challenges in distinguishing between real and fake palm biometrics due to variations in imaging conditions and the use of fixed ROI shapes, leading to unreliable outcomes and loss of semantic information.
Innovation Solution
The system employs a method where individual fake palm detectors are used to generate decisions on specific types of fake palms, and these decisions are combined to derive a liveness detection decision, while also generating variations of palm images under different conditions to improve robustness and using natural image content to define ROIs without padding with zero pixels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single binary liveness detection model is used to distinguish real palms from all types of fake palms, then the device complexity is reduced, but the detection reliability deteriorates because the differences among different fake palm types are larger than the differences between real and fake palms
Solution Approach 1:
The patent divides the liveness detection task into multiple specialized detectors, each trained to detect a specific type of fake palm (printed, photo-based, video-based, 3D mold-based). This segmentation allows each detector to focus on distinguishing subtle differences between real palms and its specific fake type, thereby improving overall detection reliability while maintaining manageable system complexity through modular architecture
2Ease of manufacture
If fixed shape ROI regions are used for liveness detection, then the processing is simplified, but semantic information is lost due to padding with zero pixels when palm regions vary in size and shape
Solution Approach 1:
The patent implements dynamic ROI regions that adapt to the actual shape and size of detected palm areas in each image. Instead of using fixed rectangular ROIs that require zero-padding, the system generates contour-based ROIs that precisely fit the palm boundary, thereby preserving all semantic information while maintaining processing efficiency through adaptive region definition
3Loss of time
If liveness detection is performed on a single acquired palm image, then the processing time is reduced, but the detection reliability deteriorates due to variations in imaging conditions such as lighting and distance
Solution Approach 1:
The patent generates multiple variations of the acquired palm image by applying different transformations (geometric transformations, color space conversions, noise additions) before liveness detection. This preliminary action creates a set of diverse test images that account for various imaging conditions, allowing the system to achieve reliable and consistent detection results without significantly increasing processing time
Data Source
AI summary
The present teaching relates to method, system, medium, and implementations for detecting liveness. When an image is received with visual information claimed to represent a palm of a person, a region of interests (ROI) in the image that corresponds to the palm is identified. Each of a plurality of fake palm detectors individually generates an individual decision on whether the ROI corresponds to a specific type of fake palm that the fake palm detector is to detect. Such individual decisions from the plurality of fake palm detectors are combined to derive a liveness detection decision with respect to the ROI.