Pose-invariant Face Recognition via SIFT Feature Extraction and Manifold Projection with Hausdorff Distance Metric

Face recognition has found its usage in various domains like video surveillance and human computer interaction. Current face recognition technique is enslaved to unknown pose of the given face image. This paper proposes a novel approach to pose-invariant face recognition. In the training phase, the SIFT feature descriptors of the sample images are extracted, then an image manifold is constructed using Laplacian Eigenmaps based on Hausdorff distance metric to model the low-dimensional embeddings of the sample images. In recognition phase, the SIFT feature descriptors of the given face image are similarly extracted, and the image is embedded into the existed manifold based on Hausdorff distance metric, the recognition is finally achieved by a K-nearest-neighbor classifier in the low-dimensional subspace. Experimental results on multiple datasets demonstrate the superiority of the proposed approach to existing methods in recognition accuracy rate.