Nowadays in biometric recognition, face recognition is the pioneering process, today all the important high alert areas are monitored by cameras, but still now there are few drawbacks in facial recognition due to several reasons. Pose variation is one of the major drawbacks, in this work methods are explained and practised to solve the pose variation problem. The recognition of person’s face using new recognition algorithms is the major research area going all around the world. In this proposed work an efficient feature extraction/reduction technique called Location averaging technique is combined with four object recognition techniques for facial recognition. Location averaging technique combines separately with Gaussian kernel, Eigen vector, Max-Min comparison and Haar function techniques, and produces effective results. Results are analysed using runtime, mismatching and accuracy for 170 images of two standard face databases. Accuracy percentage of about 99% is obtained through these novel recognition methods.

Keywords: Location averaging technique, Gaussian kernel, Eigen vector, Max-Min comparison, Haar function, Feature reduction