Face segmentation using mathematical morphology on single faces

The face segmentation is an important step for face recognition because it is not only locates the face but also get a correct segmentation for a fuller region possible. In this paper an application of the viscous opening along the skin color for segmentation and obtaining the limit of facial contour is presented. The proposal consists of two stages: the first one use the classification of pixels based on skin color given an input image to find the region of interest; in the second stage, the viscous opening is used to generate two markers of the resulting image from the previous stage, with the objective of maintaining only the facial area, avoiding the neck area.