3D Facial Model Projection for Recognition Across Poses and Expressions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition technologies struggle with accurately identifying faces in varying poses and expressions due to the limitations of using two-dimensional images, leading to reduced recognition rates.
Innovation Solution
A system that generates a three-dimensional facial model from multiple two-dimensional images, allowing for deformation based on pose and expression, and projects this model onto a two-dimensional plane for comparison with input images, enhancing recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If two-dimensional images are used for facial recognition, then the recognition process is simple and fast, but the recognition accuracy decreases when faces are in varying poses and expressions
Solution Approach 1:
The patent transforms two-dimensional facial images into three-dimensional facial models by adding depth information and spatial relationships. This dimensionality change allows the system to represent facial features in 3D space, enabling accurate recognition regardless of the face's pose or expression in the original 2D image. The 3D model captures geometric relationships that are lost in flat 2D representations.
Solution Approach 2:
The system changes the parameter representation of facial data from two-dimensional coordinates to three-dimensional spatial parameters. By representing facial features with x, y, z coordinates and orientation angles instead of only x, y positions, the system gains the ability to model and recognize faces in various poses and expressions while maintaining recognition accuracy.
2Adaptability or versatility
If three-dimensional models are generated and transformed, then adaptability to varying poses and expressions is improved, but the system complexity increases
Solution Approach 1:
The system performs preliminary action by pre-generating three-dimensional facial models from two-dimensional images during a registration phase. These 3D models are stored and prepared in advance, so that during actual recognition, the system only needs to transform and compare pre-existing models rather than creating 3D representations in real-time. This reduces computational complexity during the recognition phase.
Solution Approach 2:
The patent creates simplified copies of facial data in three-dimensional form. Instead of working with complex raw 2D images directly, the system generates streamlined 3D models that capture essential facial geometry. These models serve as simplified representations that are easier to manipulate and compare, reducing overall system complexity while maintaining recognition capability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention concerns a method and apparatus for facial recognition. At least one example embodiment discloses a facial recognition apparatus configured to detect a facial feature point from a 2D input image to be recognized, adjust a stored three-dimensional (3D) facial model based on the detected facial feature point, generate a 2D projection image from the adjusted 3D facial model, and perform facial recognition by comparing the 2D input image and the 2D projection image.