Implementation of Orthogonal Procrustes Regression Model for Face Recognition with Pose Variations

Abstract

A linear regression-based technique may be a hot topic in face recognition community. Recently, sparse
representation and cooperative representation-based classifiers for face recognition are planned and attracted nice
attention. However, most of the existing regression analysis-based ways are sensitive to cause variations. During this
paper, we introduce the orthogonal procrustes problem (OPP) as a model to handle cause variations existed in second
face pictures. OPP seeks associate optimum linear transformation between 2 pictures with completely different poses
therefore on create the remodeled image most closely fits the opposite one. We have a tendency to integrate OPP into the
regression model and propose the orthogonal procrustes regression (OPR) model. To deal with the matter that the linear
transformation isn't appropriate for handling extremely non-linear cause variation, we have a tendency to any adopt a
progressive strategy and propose the stacked OPR. As a sensible framework, OPR will handle face alignment, cause
correction, and face illustration at the same time.