Personalized 3D Face Model Generation for Robust Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Face recognition technology is sensitive to variations in face pose, facial expression, occlusion, and changes in illumination, which can lead to inaccurate identification of users.
Innovation Solution
A face recognition method that generates a personalized 3D face model from a 2D input image, using deep neural networks to extract feature information from both 2D pixel color values and 3D shape information, thereby enhancing recognition accuracy across different conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional 2D face recognition is used, then the system is simple to implement, but recognition accuracy deteriorates under variations in face pose, expression, and lighting conditions
Solution Approach 1:
The patent transforms 2D face images into 3D face models by inferring depth information and geometric structures. This dimensional transition enables the system to capture spatial relationships and surface orientations that are invisible in 2D images, thereby improving recognition accuracy under pose and lighting variations while maintaining computational feasibility through structured 3D representations.
Solution Approach 2:
The patent extracts multiple parameters from 3D face models including surface normals, curvature values, and geometric features that characterize facial geometry. By transforming the face representation from simple 2D pixel values to multi-parameter 3D descriptions, the system achieves robustness against lighting and pose changes while maintaining manageable complexity through parameterized models.
2Reliability
If 3D face modeling is introduced to improve robustness, then recognition reliability improves, but computational complexity and processing time increase
Solution Approach 1:
The patent divides the face recognition process into distinct modules: 3D model generation, surface normal extraction, curvature calculation, and feature comparison. Each module handles a specific aspect of the recognition task, allowing for optimized computation at each stage and improving overall reliability through specialized processing while managing computational complexity through modular architecture.
Solution Approach 2:
The patent performs preliminary 3D face model generation and geometric feature extraction before the actual recognition comparison. By pre-computing surface normals, curvatures, and other geometric parameters and storing them as compact representations, the system reduces real-time computational burden while maintaining high recognition reliability through pre-processed reliable features.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Disclosed is a face recognition apparatus and method, wherein the face recognition apparatus is configured to generate a personalized three-dimensional (3D) face model based on a two-dimensional (2D) input image, acquire 3D shape information and a normalized 2D input image based on the personalized 3D face model, determine feature information based on the 3D shape information and color information of the normalized 2D input image, and perform face recognition based on the feature information.