Hybridization of 2D-3D images for human face recognition

Abstract

Now-a-days face recognition is more realistic biometric approach for biometric based system for human authentication purpose. It has been aimed by the researchers and scientists over some decades to provide more reliable and secure environment. Although face recognition techniques have gained significant level of success, it is still having some challenging tasks due to the presence of facial pose, expression as well as illumination variations. With the trends of decrease in the cost of cameras, increase in the technological aspects and availability of processing power, face recognition task has now gained most of the researchers' attention in handling this complex task of computer vision. The human face images can be acquired by different methodologies, such as: from video sequences, from various sensors like optical, thermal and 3D etc. The variations of face images have also motivated the researchers to design the intelligent system for feature estimation purpose. In this chapter, an overview of hybrid techniques with its application in the domain of 3D face registration and recognition is discussed. Authors have also proposed a new 3D face recognition scheme from 2D and 3D hybrid face images using two supervised classifiers. The authors have also reported the contribution of their research work by considering all the related and recent works with proposed methodology. The investigation is accomplished on Frav3D database and achieved maximum 95.17 % accurate face recognition rate.