Liveness Detection Using Multi-Organ Feature Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current liveness detection methods in identity authentication are prone to inaccuracies due to the ease of simulating face movements using photos or videos, which can mislead systems into determining non-living entities as living, compromising security.
Innovation Solution
A liveness detection method that acquires real-time image frames, recognizes and updates feature value sets for organ regions like eyes, mouth, and nose, and performs liveness detection based on data features from these sets, using extremum conditions and combination sets to differentiate between genuine and simulated actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If basic face actions (nodding/shaking head) are used for liveness detection, then the ease of operation is improved, but the reliability deteriorates because photos or videos can simulate these actions
Solution Approach 1:
The patent divides the face into multiple organ regions (eyes, eyebrows, mouth, nose) and tracks their movements independently. By segmenting the liveness detection into multiple organ-level analyses rather than treating the face as a whole, the system can detect subtle inconsistencies in simulated actions while maintaining simple user operations.
Solution Approach 2:
The patent extends detection from 2D image analysis to 3D spatial tracking by establishing world coordinate systems and tracking organ positions in three-dimensional space. This dimensional enhancement allows the system to detect depth information and movement patterns that are difficult to simulate in flat photos or videos, thereby improving reliability without complicating user interaction.
2Measurement precision
If face key-point positioning and face tracking are used to verify user operations, then the measurement precision is improved, but the device complexity increases
Solution Approach 1:
The patent develops a unified face tracking system that simultaneously performs multiple functions: organ region recognition, key-point positioning, movement trajectory tracking, and liveness determination. By creating a multi-functional system rather than separate specialized modules, the patent achieves high measurement precision while controlling overall system complexity through shared computational infrastructure.
Solution Approach 2:
The patent pre-establishes world coordinate systems and organ region models before actual liveness detection occurs. By preparing tracking frameworks, coordinate transformations, and organ position references in advance, the system achieves precise real-time measurement during detection without requiring complex computational processing during the actual verification, thus balancing precision and complexity.
3Measurement precision
If multiple organ regions are tracked with feature value sets and extremum conditions, then the liveness detection accuracy is improved, but the loss of time increases due to extensive data processing
Solution Approach 1:
The patent implements incremental updating of feature value sets, where only newly detected organ positions and movements are processed rather than re-analyzing all historical data. This partial processing approach maintains high detection accuracy through cumulative feature accumulation while significantly reducing computational time compared to complete re-evaluation at each detection step.
Solution Approach 2:
The patent establishes continuous tracking of organ regions across multiple video frames, maintaining persistent feature value sets that accumulate movement patterns over time. This continuous action allows the system to build robust liveness indicators through sustained observation rather than relying on single-frame analysis, improving accuracy while optimizing processing efficiency through temporal continuity.
Data Source
AI summary
Provided are a liveness detection method and apparatus, an electronic device and a storage medium. The method includes: in a case of satisfying a liveness detection starting condition, acquiring (S110) an image frame in a video in real time; recognizing (S120) at least two organ regions of a user in the image frame, and updating a feature value set corresponding to each recognized organ region according to a feature value calculated based on the recognized organ region corresponding to the feature value set; and performing (S130) a liveness detection on the user according to data features in a combination set formed by at least two feature value sets corresponding to the at least two organ regions and extremum conditions respectively corresponding to the at least two feature value sets.