Recognition of Facial Expression
based on Captured Facial Data and Structure of Muscle
Over the last few years, various media, for example digital camera, video and
motion capture system, is rapidly developed.
Thus, many workers research a human face technically for a
communication media and man-machine interface.
Human face is strongly related to us. Moreover, It has individual
,communicational, nonverbal and many various information.
By analysing human facial expression, we can build a ``facial expression database''
In this paper, we will introduce a method to extract rules and recongnition
of facial expression using motion caputerd data, in order to build
``facial expression database''.
Because motion captured data is 3D time series data, calculated cost.
However, facial parts don't move independently. There is a constraint of
facial muscle. Thus, to reduce the calculated costs, we will convert
facial data to stretch of muscle called ``stretching coefficient'' using
PCA(Principle Component Analysys).
And then, we will discover the rules from each stretching coefficient using
the FTDA(Fuzzy Transaction Datamining Algorithm). Finally, we will
recognize the similarity of six basic expressions.