非表示:
キーワード:
-
要旨:
When a person approaches you, there are multiple cues that could help you to identify that person, such as their face, hairstyle, clothes or walking pattern. Currently we are using animated
figures to explore the differential impact of these cues on person recognition. Here we mainly
concentrate on the face as a cue to identity and especially the influence of external and internal
facial features on recognition performance. Internal facial feature refer to the size and outline
of the eyes and mouth and their configuration. External facial features rather denote the shape
of the face or the hairstyle associated with a particular face. We use 3D graphics and computer
animation to explore the impact of such sources of information. Animation techniques
allow us to modify some features, while keeping other features constant, e.g. the body or walk
pattern. As stimuli we attached faces from the MPI face database to 3D body models from
POSER and animated them with a simple walk as if approaching the observer. In a learning
phase, observers studied two static targets that only differed in their facial features. Hairstyle
and clothing were kept constant. In a later recognition task, these two targets were moving
towards the observer, who had to identify them as quickly and accurately as possible. During
this recognition task, we varied hairstyle and specific clothing to test which kind of external
features impair recognition performance most. Our results show that for close-up recognition
observers seem to rely more on internal facial features, whereas they are highly influenced by
external factors, i.e. hairstyle, when a person is further away. These findings provide additional
evidence that external facial features are important for recognizing unfamiliar faces.
