A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION
Mihaela L. Gordan, Constantine L. Kotropoulos, Apostolos Georgakis, Ioannis · 2002
The problem of lip contour detection is critical in the lipreading systems based on contour processing. The typical contour detection strategy based on image segmentation in homogeneous regions fails in the case when the mouth images available for lipreading are lowcontrast gray level images. Most of the solutions adopted require manual marking of some contour points. Here we propose a new solution from the image segmentation class suitable to lip contour extraction, using a modified version of the fuzzy c-means image segmentation algorithm. The novelty of the solution proposed consists in the types of features used for segmentation, which include not only luminance information (as in the standard use of fuzzy c-means), but also spatial information about the pixels in the image. After a simple filtering of the outliers, the contours of the resulting segmented objects are extracted. The experimental results obtained are superior to the ones obtained by standard or other versions of geometrically constrained fuzzy c-means, or by gradient-based edge detection strategies, without the need of manual marking of any contour points. Thus we consider the strategy proposed promising for automatic lip contour extraction applications.