Facial Expression and Gesture Analysis for Emotionally-Rich Man-Machine Interaction - 3D Modeling and Animation: Synthesis and Analysis Techniques for the Human Body - page 180

shown (Bassili, 1979) that facial expressions can be more accurately recognized
from image sequences, than from single still images. Bassili's experiments used
point-light conditions, i.e., subjects viewed image sequences in which only white
dots on a darkened surface of the face were visible. Expressions were
recognized at above chance levels when based on image sequences, whereas
only happiness and sadness were recognized when based on still images.
Affective Gesture Analysis
The detection and interpretation of hand gestures has become an important part
of human computer interaction (MMI) in recent years (Wu & Huang, 2001).
Sometimes, a simple hand action, such as placing a person's hands over his ears,
can pass on the message that he has had enough of what he is hearing. This is
conveyed more expressively than with any other spoken phrase.
Gesture tracking and recognition
In general, human hand motion consists of the global hand motion and local finger
motion. Hand motion capturing deals with finding the global and local motion of
hand movements. Two types of cues are often used in the localization process:
color cues (Kjeldsen & Kender, 1996) and motion cues (Freeman & Weissman,
1995). Alternatively, the fusion of color, motion and other cues, like speech or
gaze, is used (Sharma, Huang & Pavlovic, 1996).
Hand localization is locating hand regions in image sequences. Skin color offers
an effective and efficient way to fulfill this goal. According to the representation
of color distribution in certain color spaces, current techniques of skin detection
can be classified into two general approaches: nonparametric (Kjeldsen &
Kender, 1996) and parametric (Wren, Azarbayejani, Darrel & Pentland, 1997).
To capture articulate hand motion in full degree of freedom, both global hand
motion and local finger motion should be determined from video sequences.
Different methods have been taken to approach this problem. One possible
method is the appearance-based approach, in which 2-D deformable hand-shape
templates are used to track a moving hand in 2-D (Darrell, Essa & Pentland,
1996). Another possible way is the 3-D model-based approach, which takes the
advantages of a priori knowledge built in the 3-D models.
Meaningful gestures could be represented by both temporal hand movements
and static hand postures. Hand postures express certain concepts through hand
configurations, while temporal hand gestures represent certain actions by hand