3D Face Liveness Detection via Head Rotation Phase Map
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing liveness detection systems face challenges in efficiently verifying live faces using single camera setups, particularly due to high resource requirements, data transfer limitations, and the need for high-resolution data collection and training.
Innovation Solution
A method involving 3D face shape verification, where a user is instructed to rotate their head through specific angles, capturing images at each position, transforming yaw and pitch angles into a graph in polar coordinates, and using a machine learning classifier to determine if the head is a live 3D face or a different object.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the resolution of the input camera is increased to improve liveness detection accuracy, then the measurement precision is improved, but the resource requirements and data transfer needs increase
Solution Approach 1:
The patent extracts only the essential features needed for liveness detection by transforming head rotation angles into a compressed phase map representation. Instead of transmitting entire high-resolution video frames, the system extracts angular position data and converts it into a condensed phase map that retains the critical 3D facial shape information while dramatically reducing data volume.
Solution Approach 2:
The patent changes the parameter representation from raw pixel data to angular parameters (yaw and pitch angles). By measuring head rotation angles and transforming them into a phase map, the system achieves high measurement precision for liveness detection while working with compact angular parameters rather than large image matrices.
2Measurement precision
If the input resolution is increased to capture more facial detail, then the measurement precision is improved, but the computational resources required for classification increase
Solution Approach 1:
The patent extracts only the critical angular information needed for liveness detection. By focusing on head rotation angles (yaw and pitch) rather than processing entire high-resolution images, the system achieves accurate facial feature detection with minimal computational overhead.
Solution Approach 2:
The patent transforms the problem from 2D image analysis to angular parameter analysis. By representing facial geometry through head rotation angles in a phase map, the system reduces the dimensionality of the data that needs to be processed, thereby lowering computational resource requirements while maintaining detection accuracy.
3Reliability
If video frames are collected and transmitted to server for classification, then the liveness detection capability is improved, but the data transfer requirements and lossless compression needs increase
Solution Approach 1:
The patent extracts angular parameters directly from video frames and transmits only these compressed representations to the server. By sending head rotation angle data in phase map format rather than complete video frames, the system maintains reliable liveness detection while eliminating the need for lossless compression of large video datasets.
Solution Approach 2:
The patent transforms spatial video data into angular parameter space. This dimensional transformation allows the system to represent complex facial 3D shape information compactly through phase maps, ensuring reliable server-side classification without requiring lossless compression of the original video data.
4Measurement precision
If more training data is collected to improve classification model accuracy, then the measurement precision is improved, but the quantity of data required increases
Solution Approach 1:
The patent changes the training data from full-resolution images to angular parameter representations. By training the classification model on phase maps derived from head rotation angles, the system achieves high classification accuracy using compact parameter-based datasets rather than voluminous image collections.
Data Source
AI summary
Method of 3D face verification, includes verifying that user is positioned in standard orientation so that head rotation angles are small; capturing image of user's head at first position; generating PIN and transmitting it to a user device; upon receipt of the PIN, instructing the user to rotate his head to a first position indicated by the PIN and a corresponding graphic; repeating instructions to the user to rotate his head for all remaining values of the PIN, and capturing images of the head at positions corresponding to the remaining values of the PIN; transforming yaw and pitch angles of the user's head in the captured images into a polar coordinates graph; transforming the graph into a phase map; sending the phase map into machine learning classifier; securely sending and storing personal information to the server; determining, using the machine learning classifier, whether the head is a live 3D face.