Mouth Segmentation Based on Color Analysis
AUTHORS:V. Libal, L.J.M. Rothkrantz, M. Vlcek
KEYWORDS:
ABSTRACT:
This paper reports work done at mouth segmentation
and mouth features extraction problem. The work extends Juran and Krsek's
approach to mouth segmentation. Resulting algorithms are supposed to be
include into larger program as Automated Recognition of Facial Expressions
program (ARFE) or program of automated lip-reading.
Juran and Krsek introduce successful solution to
mouth segmentation based on finding an external lip contour using active
contour. Comments on their approach are given in chapter two. This discussion
is extended and led into more general notes on model of the mouth.
Methods for complete mouth segmentation are presented
in chapter three. A novel method for coarse mouth region location was used.
Juran and Krsek's algorithm of active contours or any edge detector can
be used for fine location of external lip contour. Simple methods for partitioning
the mouth region into its subsets are suggested in the end of third chapter.
In fourth chapter some of typical results of program
that is based on presented algorithms are shown together with comments
on them
Summary of meeting the requirements, revision of
notes from chapter two and some more proposals for future works are introduced
in concluding fifth chapter.
