Study On Segmentation And Feature Extraction Of Lip Region In Image

Feature recognition technology based on lip-movement is a new type of biometrics identification technology. The segmentation of the mouth area and the extraction of the features in an image are very critical to the feature recognition technology based on lip-movement. It is related to many fields such as the computer vision, the image processing, and the pattern recognition. At present, the research on the extraction and analysis of the lip in a color image in China is preliminary. For the practical application of this technology, there are still several key problems to be solved in related studies at the international level. Therefore, the segmentation of the mouth area and the extraction of the features have important research meanings and good potential application prospect.Locating the mouth directly in an image is difficult and has low accuracy. To solve this problem, a lip locating method based on the segmentation of the face area is proposed in this paper. The key of the method is to segment the face accurately. Because the clustering feature of the skin color in YCg Cr color space is better, this paper put forward a face segmentation method based on YCg Cr color space. Experimental results show that this proposed method based on YCg Cr color space has good detection effect in simple background and even if in complicated environment, most areas that are not face can be eliminated. The method has high detection accuracy.On the basis of the location rule of the mouth in the face and geometric proportion coefficient, this paper will locate the position of the mouth in the segmented face region and process the segmentation of mouth area. Experimental results show that this way can locate the mouth area accurately and segment the mouth area. On the basis of the segmentation result of the mouth, this paper will extract the features of the lip by constructing the mouth shape templates of outer lip and defining expressions describing the outer lip curves. What’s more, this paper will also use two parabolas to fit the outline of the lip. Experimental results show that this feature extraction method has small calculating amount and can extract the feature parameter well. It is the basis to the subsequent analysis and recognition.