Mouth Localization

After having found a pair of possible eyes which satisfies the geometrical constraints imposed by the face, it is necessary to test for the presence of the mouth. This will be used not only to check the validity of the eyes but will more importantly localize the face further so that a more precise definition of its coordinates is obtained. We choose to locate the mouth after having located the eyes because it has a non-skin tone texture and stands out more clearly than the skin-covered nose. Furthermore, its position is more stable than the nose since it lies consistently between the two eyes despite rotations of the face. Thus, the next most reliable step in the hierarchy is mouth detection.

Unlike the eye region, the mouth does not have a blob-like, circular outline. When it is closed or slightly open, the mouth is a thin, elongated structure or a limb. Thus, it may not be detected by the symmetry transform's interest map. We propose the use of a limb extraction stage as outlined by Kelly [20] to detect the mouth. The limb extraction process begins with the computation of the points of symmetry as displayed in Equation . This is an intermediate computation that was used to compute the interest map. Furthermore, only annuli of radius r=1 to r=6 are used, since the mouth's vertical thickness is slightly smaller than the vertical thickness of the eye region (from the top of the eyebrows to the bottom of the eye orbit). We only compute dark symmetry since the lips and the interior of the mouth are darker than the surrounding skin.

The result of the symmetry computation is an image at each value of r (r=1 to r=6) with each point containing 8 magnitudes for each symmetry orientation. Thus, a symmetry magnitude is computed for each point p for each r from r=1 to r=6 and for each $\psi$ from $\psi$=1 to $\psi$=8 (8 symmetry orientations as in Figure ). The resulting symmetry points are output as in Figure .