Research, Face Recognition Method Based On Rough Set Theory

The face recognition is a main direction of biology feature recognition technology.Contrast to other biology features,face recognition have some superiorities for example initiative ,conveniency.There are wide application value in the research of this field.It can be used in many aspacts,for example identity discriminition, information safety,man-machine interaction,etc.The face recognition system include:face detection and location,face extraction and chosen,face recogniton. In the first part,this paper adopts a method used for face location which takes the advantage of the pixel distribution features of face images.While locating the face inner outline,the distribution features of both binary image of the face and grey-level image of face are introduced.Then the face area images are standardized.Its convenient for the standardized face images to be uses in face recognition.It get ready for the subsequent processiong to face images,for example face feature extraction and chosen.In the second and third part,this paper chooses 11 feature points,forms a feature vector include 22 features.Then on the basis of rough sets theory,the face images in the face image base is acted as a knowledge expression system.Through exercising part face images in the face basis,a attribution table can be formed.By use of attribution reduction method in the rough sets theory,a minimum face image base can be formed.Finally,through the minimum oucliud distance rule,the face images in the test set can be recognized.Because the traditional method is that the features have been used in recognition directly after they have been extracted.But there are some problems such as too many features,too much computation,too complicated system.In this paper,rough sets theory have been used in face recognition.It reducts those non-needed features which is redundancy,so the system has been simplified.Tested by experiments,this method has feasibility.