3D Face Model Pose Adjustment for Recognition Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Face recognition systems face difficulties in matching face images collected at unconstrained poses, such as side views, with databases containing only frontal view images, leading to challenges in accurate recognition.
Innovation Solution
A method involving the detection of 2D landmarks from a 2D face image, positioning 3D landmarks in a 3D face model, iteratively adjusting the pose and shape of the 3D model, and projecting it back to adjust the 2D image pose, while also mapping background preserving textures to improve matching accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If face images are collected under unconstrained circumstances, then the system can capture more diverse real-world scenarios, but the matching accuracy deteriorates due to pose variations
Solution Approach 1:
The patent introduces a 3D face model as an intermediary between the captured 2D face image and the database matching process. The 3D model serves as a mediator that can be adjusted to different poses, allowing the system to handle unconstrained circumstances while maintaining matching accuracy by comparing adjusted images against the database
Solution Approach 2:
The system changes geometric parameters of the 3D face model (rotation angles, scaling factors) to adjust the model's pose. By modifying these parameters iteratively, the system can transform a frontal view model into various poses matching the captured image, thereby maintaining high matching accuracy across diverse scenarios
2Measurement precision
If a 3D face model is used to adjust pose, then matching accuracy is improved, but the computational complexity increases
Solution Approach 1:
The system performs preliminary actions by pre-establishing a 3D face model with known geometric structure before the actual matching process. This pre-computed model serves as a foundation that reduces the complexity of real-time pose adjustment, as the basic 3D geometry is already determined and only parameter adjustment is needed
Solution Approach 2:
The system employs an iterative feedback mechanism where the adjusted 2D image from the 3D model is compared with the captured image, and the geometric parameters are refined based on the matching error. This feedback loop gradually improves alignment while converging to an optimal solution, balancing accuracy with computational efficiency
Data Source
AI summary
A method and an apparatus for adjusting a pose in a face image are provided. The method of adjusting a pose in a face image involves detecting two-dimensional (2D) landmarks from a 2D face image, positioning three-dimensional (3D) landmarks in a 3D face model by determining an initial pose of the 3D face model based on the 2D landmarks, updating the 3D landmarks by iteratively adjusting a pose and a shape of the 3D face model, and adjusting a pose in the 2D face image based on the updated 3D landmarks.