A Research Of Face Recognition Method Through Geometric Feature

With the development of technology and internet,the security of information is increasingly threatened.How to identify a person's identity accurately and protect information security effectively is a key issue to be addressed in today's society.Biometric technology has greater security,reliability and validity because of using the inherent characteristics of the human body,and has received more and more attention.Face recognition,as one of the most friendly biometric authentication technology,are bound to become a hot topic in the field of multi-disciplinary.Face recognition is a process of identifying a face image by analyzing face images and extracting the facial feature information.It has been widely used in identity recognition,entrance guard system and video surveillance.The method of human face recognition based on geometric feature has an advantage of simple,intuitive and small memory requirements in the field of face recognition.Based on the face detection,combined with the prior knowledge of geometrical relationship among face organs' structure and distribution,this paper uses corner detection and gray projection method to extract the facial characteristics of major organs,expressing human face through a set of geometric feature vectors.The feature vectors are matched for recognition.The main works in this paper are following as:Through the research of current face detection algorithms,the Adaboost algorithm based on the Haar characteristics and integral figure is used for face detection in this paper.This algorithm can efficiently detect a human face.During the process of face recognition,in order to extract facial features,the localization of human eyes,nose and mouth is necessary,and among them the positioning of eyes is the most important.This paper proposes a precisely eyes localization method.It first applies the corner detection to roughly locate eye area.Then according to the distribution and transformation character of grey value in eyes area,the integral projection and differential projection are combined to locate eyes accurately.Then select the appropriate parameters through a lot of experiments.The experiment results show that this algorithm has a high accuracy.During the process of face recognition,in order to match facial features,this paper selects some feature points according to the located eyes,nose and mouth.Then construct feature vectors using them and carries on the normalization calculation on them.Then calculate the similarity of feature vectors and makes faces matching by it.