Face and Ear Recognition system using Novel Techniques for Human Authentication

Abstract

Biometrics plays a vital role in the object identification, mainly for human recognition newlineand detection for security purpose. In the current, biometric human recognition scenario, novel ideas are evolving to solve the errors in face and ear recognition system. Lot of techniques and systems are blooming to overcome the problems present in the face and ear recognition. New techniques are proposed in this work to identify people and the notable objectives are to recognise persons face, ear image perfectly, depends on their pose, size varied images, and to detect the face picture present in an image. Identifying a persons face is the basic job, in the practical situation several parameters includes image size, resolution, pose variation and several atmospheric disturbances like dust, rain, sweat etc are affecting the recognition system. To protect a persons personal data or to authenticate a person, his/her Name, Password, and PIN (etc) are used. Chances are there to duplicate, copy, the data and reuse it, without the person knowledge. So it is necessary to change password frequently. But the biometrics paved the way to use the biometric traits such as face, fingerprint, ear, iris, and palm (etc) as an authentication tool. It is not easy to duplicate or replace the biometric traits. Biometrics acts as the reasonable identification authority for identifying a human. During recognition, non similar poses of a same person s images gives false image matching and gives false output. Similarly, it is difficult to do comparison between different sizes of images, so good methods are necessary to produce the correct biometric data and for solving the problems in it. Two basic problems called, Pose variation and size variation are discussed and solutions are proposed. In this proposed work Novel Spatial domain techniques like Gray averaging, Location averaging, Max-Min Comparison, and Intensitys position estimation are combined with frequency domain technique includes Discrete Cosine Transform, Fast Fourier Transform.