Overview:

IIT 3D Face Scanner - original hardware-software system for human face 3D reconstruction and "exact frontal image" generation. 3D reconstruction of human face is performed by photogrammetric technology based on close-range stereo vision with IR structured light. Special module of automatic face features segmentation (nose, eyes, brows, etc.) is applied for textured 3D face model rotation estimation in order to generate the "exact frontal image" of captured face. This system can both significantly increase the efficiency of existing 2D face identification systems (due to more stable "exact frontal" face imagery source), and also be of use as a basis for creation of 3D face biometric systems.

Functions:

• 3D reconstruction of human face
• "Exact frontal" face image generation based on 3D face model and corresponding close-to-frontal image

Hardware recommended:

• 3 b/w video cameras (for IR stereo capturing)
• color video camera (for color face texture capturing)
• special source of IR structured light
• frame grabbers
• PC station

Output data:

• 2D data: Color "exact frontal" face image in BMP or JPEG format
• 3D data: 3D face model (cloud of points, triangulated set or regular grid) in CAD format
• Virtual 3D data: 3D face model with photorealistic texture and possibility for interactive 3D-visualization using special 3D Viewer software

Features:

• Accuracy:
  • 3D face measurement - 1 mm (for each point coordinate)
  • Accuracy of face image rotation to the frontal plane - 10 degree
• Processing time - not greater than 2 sec from image capturing to data output

Destination:

• Increasing of efficiency of 2D face recognition systems (due to usage of more stable "exact frontal" face imagery source)
• Face recognition using 3D models
• Hybrid 2D&3D face recognition - using 3D models and 2D frontal images as different evidence sources

Applications:

• passports, visas and other travel documents
• buildings, areas and borders securing
• anti-criminal and anti-terrorist police

Advantages:

• Increasing of efficiency of face recognition systems (due to possibility of generation of any required angular view of scanned human face
• More robustness and safety of 3D face recognition systems relative to clear 2D systems