The Ethnic Facial Research Based On Algebraic Characteristics

The reseach of ethnic facial characteristics not only belongs to anthropology, it also belongs to the research category of face recognition technology. Anthropological research suggests that the facial features between the different ethnic groups have some differences, what'more, the same nation in different regions have different characteristics, so the study can explain the various ethnic facial features inherent laws. In this paper, we used a method based on algebraic features ethnic facial characteristics. I mainly used principal component analysis and Haar-like feature extraction of facial features in the national application.This paper applies the relevant image processing technology, building the national facial image database were pretreated in the image database is divided into training set and test set images image. In ChapterⅣ,the way of principal component analysis on the processed Uygur, Zhuang and Tibetan facial feature image extraction and algebra. The larger eigenvalue contribution rate of the corresponding feature vector is slected to construct training images and the characteristics of the unknown face image,then using k- nearest neighbor classification on national facial images for testing. The results show that the research which combined of principal component analysis and the k-nearest neighbor classification of facial features is feasible.In chapterⅤwe construct the four Haar-like feature extraction templates, using them to extract the ethnicical face image algebra feature extraction, the final feature vector used to test the angle relations. From the experimental results, relative to the principal component analysis, based on Haar-like feature extraction algorithms get better results, this paper Haar-like feature extraction method is effective and feasible, can be used to analyze the human face of their national identity.