3D Facial Landmark Matching on Mobile ID Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for biometric facial recognition using two-dimensional images are complex, costly, and prone to errors due to the difficulty in capturing distortion-free images, making personal identification inefficient and unreliable.
Innovation Solution
A method and mobile device that utilize a personalized ID document with stored three-dimensional facial feature data, enabling the capture and comparison of current three-dimensional facial models with reference values to identify individuals securely and efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If two-dimensional facial images are captured using conventional cameras, then the identification process can be performed, but the images often contain distortions that reduce recognition accuracy
Solution Approach 1:
The patent transitions from two-dimensional facial image capture to three-dimensional facial model acquisition. By using multiple cameras or structured light to capture depth information, the system creates 3D models that are inherently more resistant to lighting variations, pose changes, and perspective distortions, thereby improving both measurement precision and reliability of facial recognition
Solution Approach 2:
The patent creates a digital 3D copy of the facial geometry that can be stored and compared without degradation. This digital twin approach allows the reference facial model to be preserved exactly as captured, eliminating the quality degradation that occurs with repeated use of analog or 2D digital images
2Measurement precision
If strict image acquisition parameters are enforced to ensure quality, then recognition accuracy improves, but the complexity and time required for image capture increases
Solution Approach 1:
The patent replaces complex mechanical image acquisition constraints with computational 3D modeling. Instead of requiring precise control of camera position, lighting angles, and subject pose through mechanical means, the system uses algorithms to reconstruct 3D facial models from multiple views, substituting mechanical precision requirements with computational processing
Solution Approach 2:
The patent changes the fundamental parameters of facial data representation from 2D pixel values to 3D geometric coordinates. This parameter transformation allows the system to work with intrinsic facial geometry rather than extrinsic image qualities affected by capture conditions, reducing the need for strict acquisition parameter control
3Reliability
If multiple image acquisition parameters are controlled to ensure suitable images, then the identification reliability improves, but the time required for the identification process increases
Solution Approach 1:
The patent performs preliminary 3D model construction and stores reference facial models in advance. During actual identification, the system compares captured images against pre-prepared 3D reference models rather than processing images through multiple quality checks and adjustments at the moment of identification, significantly reducing processing time while maintaining high reliability
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for identifying a person based on a biometric facial feature using a mobile device, wherein an ID document (3) is provided with a memory (2) for electronic data, and wherein, for personalizing the ID document (3) for a person to whom the ID document (3) is assigned, at least one of the following data sets is stored in the memory (2): reference value data, which indicates a reference value for a geometric relationship for at least two facial recognition landmarks in a three-dimensional reference model for the person's face; landmark data, which indicates facial recognition landmarks in the three-dimensional reference model and from which the reference value data can be determined;and model data representing the three-dimensional reference model from which the reference value data can be determined. Furthermore, the method comprises the following: reading at least one of the data sets from the memory (2) of the ID document (3) and providing the reference value data to at least one processor of a mobile device (5); capturing current image data of the person's face with an image capture device (7) of the mobile device (5); determining a current three-dimensional model of the person's face using the current image data;Determining comparison data from the current three-dimensional model of the person's face, wherein the comparison data indicates a comparison value for the geometric parameter for the at least two facial recognition landmarks determined in the current three-dimensional model, and providing the comparison data to the at least one processor (6) of the mobile device (5); and identifying the person when, using the at least one processor (6) of the mobile device (5), it is determined that the comparison value is equal to the reference value within a specified error tolerance. Furthermore, a mobile device (5) is designed to identify a person based on a biometric facial feature.