Automatic extraction of eye and mouth fields from monochrome face image using fuzzy technique

Because a person's eyes and mouth play an important role in expressing his intention and feeling, many researches on automatic extraction of eye and mouth areas are being carried out. One common conventional method, which uses intensity information, needs a good light condition so that the facial images will not be affected by noise like shadows. In this paper, we present an extraction method which can extract eye and mouth areas even there is noise present in these areas. In the method, (1) Coarse field classification is performed with a membership function which shows the degree of similarity between two adjacent pixel values. (2) For every segmented field fine clustering is performed with fuzzy c-means to determine the area which includes eye or mouth. (3) Because the area including eye or mouth may also include noise like eyebrow or shadows, the average pixel value of each segmented field is adjusted by its distance to the center of guessed field of eye or mouth to remove the noise. An experiment using this method was performed and the satisfactory result was obtained.