About Face: Matching Unfamiliar Faces Across Rotations of View and Lighting

Area of Science:

• Cognitive Psychology
• Computer Vision
• Neuroscience

Background:

• Facial identity recognition is susceptible to image variations like lighting and viewpoint.
• Understanding how these factors interact is crucial for developing robust face recognition systems and theories of human perception.

Purpose of the Study:

• To investigate the combined effects of lighting and viewpoint on unfamiliar face matching.
• To determine the relative contributions of viewpoint and lighting across different facial axes (yaw and pitch).

Main Methods:

• Utilized two experiments employing 3D head models for precise control over lighting and viewpoint.
• Factorially analyzed the impact of variations in yaw and pitch angles, and lighting direction on face matching performance.

Main Results:

• Viewpoint significantly affected face matching along the yaw axis, with minimal lighting influence.
• Along the pitch axis, both viewpoint and lighting influenced matching, with an interaction observed: lighting effects were prominent in front and below views.
• Results were consistent whether head or camera position varied, suggesting perception is not solely image-dependent.

Conclusions:

• Face perception relies on integrated shape and surface information, not just 2D image properties.
• Implicit knowledge of upright faces and natural (top-down) lighting conditions guides face recognition.
• Findings contribute to understanding the complex interplay of factors in human face perception.