High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Pose Invariant Recognition using Sift Based Feature Extraction and MRF

Author(s):

Sushama Kumari R B , Narayanaguru College of Engineering; D.Gnanajebadas, Narayanaguru College of Engineering

Keywords:

Eigen Faces, Fisher Faces, Morphable Model, Active Appearance Model, Light Fields

Abstract

Face recognition has now achieved high performance under controlled image conditions. One of the main advantages of face recognition as compared to other biometric identification techniques is that it does not require the cooperation of participants. The important challenges in face recognition are to recognize faces across different poses, expressions and illuminations. In this paper, face recognition is performed by considering the pose variations. In this paper, an efficient method for the reconstruction of frontal views from nonfrontal face images using Markov Random Field is presented. In this approach the input face image is divided in to overlapping patches and a set of local warps are estimated corresponding to each patch inorder to synthesize the patches in frontal view. A set of warps corresponding to each patch is synthesized by aligning it with the images from the database of frontal images. Any face recognition techniques can be used for the recognition of reconstructed frontal face images. Feature Extraction is performed using SIFT algorithm. Scale Invariant Feature Transform is an efficient algorithm for feature extraction since it is independent of scale and orientation variations. The main advantages of the proposed method are that it does not require any manually selected facial landmarks or head pose estimation.

Other Details

Paper ID: IJSRDV3I30039
Published in: Volume : 3, Issue : 3
Publication Date: 01/06/2015
Page(s): 1136-1138

Article Preview

Download Article