Biometric Authentication for Online Examination

Abstract

There are many biometric authentication methods such as face, finger print, iris, hand, veins, keystroke, signature, voice based authentication .Among that face recognition based research has been chosen under cost and time consuming factors.The main objective of the face recognition research is to recognize a sample face from a set of given authenticated student faces in order to provide more security. In this project principle component analysis (Eigen face approach) is applied to recognize a student face that is a face under different lightening and emotional condition. For comparison and experimental analysis simple approach such as user name and password based authentication and finger print based authentication are used. The final result is analysed and the face recognition method produces best result. The present invention of face recognition based authentication involves two phases such as, face detection which is the primary process and face recognition which is an authenticating phase.
Face detection involves four main concepts. Firstly, face localization which separates parameter space and object space using Hough method and skin color information method[4]. The Second step is face normalization which extracts only the face by discarding all the surroundings. Third step is to locate facial characteristics using neural network .Finally the student face is extracted using Eigen face approach. After extracting the features of the face, all these features will be basically stored as a template that will be used for recognition. In recognition phase, the student face is captured and checked for authentication. Only if the face matches with the store template the student will be allowed for the examination. Like single face recognition multi face recognition system uses principle component analysis (PCA) technique[1]. To perform PCA five steps to be undertaken. The first step is subtracting the Mean of the data from each variable. The second step is calculating and forms a covariance Matrix. The third step is calculating Eigenvectors and Eigen values from the covariance Matrix. The fourth step is to choose a Feature Vector. Final step is multiply the transposed Feature Vectors by the transposed adjusted data.