Elastic body spline technique for feature point generation and face modeling.

Area of Science:

• Computer graphics
• Image processing
• Computational physics

Background:

• Facial animation is crucial for realistic digital characters, especially with the MPEG-4 standard.
• Current mesh-based facial animation often uses elastic body spline (EBS) with uniform elastic properties.
• This uniform property assumption limits the realism of expressions.

Purpose of the Study:

• To improve the realism of facial animation by refining the elastic body spline (EBS) method.
• To develop an algorithm that accounts for varying elastic properties across different facial regions.
• To achieve more accurate deformation of facial mesh models for enhanced animation.

Main Methods:

• Partitioning facial images into distinct regions.
• Proposing an iterative algorithm to determine the unique elastic property for each facial region.
• Applying these regional properties to calculate EBS for mesh vertices.

Main Results:

• Successfully determined region-specific elastic properties for facial mesh deformation.
• The proposed iterative algorithm enables more precise control over facial mesh transformations.
• Demonstrated a method to achieve more realistic facial animations compared to uniform property approaches.

Conclusions:

• Accounting for regional variations in elastic properties significantly enhances facial animation realism.
• The developed iterative algorithm provides a robust method for calculating accurate EBS parameters.
• This research contributes to more lifelike digital human representations in computer graphics.