University of Hull logo

Face matching across 2-D and 3-D representations

Young, A; Liu, C; Ward, James

Authors

A Young

C Liu

James Ward J.W.Ward@hull.ac.uk

Abstract

Research on face recognition has mainly relied on photographic or video stimuli where the 3-D volumetric information of a face for both training and test materials is limited to the depth cues given by the 2-D pictorial representation. In reality, however, observers can be required to match 3-D faces containing rich binocular depth cues against 2-D representations in photographs. The matching process in this case requires deriving correspondence from different types of depth information. To examine this transfer of information across 2-D - 3-D representations, participants were asked to perform identification tasks where only one of the training or test images contained stereo information. Recognition performance of this incongruent condition of dimensionality was compared with two congruent conditions where both images for training and test were shown with or without stereo information. It was found that recognition accuracy for congruent conditions was higher than for the incongruent condition in a yes - no recognition task and a sequential matching task. No difference was found between the results of the two congruent conditions. These results suggest that face recognition is little affected by the absence of stereo information, but can be compromised by the mismatch between the type of depth cues present in the face stimuli.

Journal Article Type Meeting Abstract
Publication Date Dec 31, 2004
Journal PERCEPTION
Print ISSN 0301-0066
Publisher SAGE Publications
Peer Reviewed Peer Reviewed
Volume 33
Issue ECVP Abstract Supplement
Pages 109 - 109
Institution Citation Young, A., Liu, C., & Ward, J. (2004). Face matching across 2-D and 3-D representations. Perception, 33(ECVP Abstract Supplement), (109 - 109). ISSN 0301-0066
Keywords face recognition; face matching; 2D; 3D; two dimensional; three dimensional
Downloadable Citations