Three-dimensional facial digitization using advanced digital image correlation.

Area of Science:

• Biomedical Engineering
• Computer Vision
• Medical Imaging

Background:

• Accurate 3D facial reconstruction is crucial for various scientific and engineering applications.
• Existing methods often rely on active lighting or artificial patterns, increasing complexity and cost.
• A non-intrusive, pattern-free approach is desirable for practical 3D facial digitization.

Purpose of the Study:

• To present an effective technique for acquiring 3D digital images of the human face.
• To achieve accurate 3D facial digitization without active lighting or artificial patterns.
• To demonstrate the practicability and cost-effectiveness of the proposed method.

Main Methods:

• The technique combines binocular stereo imaging and digital image correlation.
• Key steps include precise camera calibration using a pinhole model and bundle adjustment.
• Reliable pixel-to-pixel image matching is achieved by leveraging natural facial features like pores and freckles, employing feature-matching, iterative optimization, and reliability-guided computation.

Main Results:

• The proposed technique successfully acquires 3D digital images of the human face.
• Experiments validated the accuracy and effectiveness of the 3D digitization process.
• The method demonstrated robustness in image matching by utilizing inherent skin features.

Conclusions:

• The developed technique offers a simple and affordable solution for 3D facial imaging.
• Its reliance on natural skin features eliminates the need for external patterns or lighting.
• The approach holds significant potential for practical applications in scientific and engineering fields.