Bayesian face recognition using deformable intensity surfaces

Robust Face Recognition through Local Graph Matching

“…Recognition Rates (EBGM) -Wisskott et al [4] 95.5% (LDA + PCA) -Etemad et al [2] 96.2% (BIC) -Moghaddam et al [3] 94.8% (Boosted Haar) -Jones et al [13] 94.0% (LBP) -Timo et al [15] 97% Our method 98.4%…”

“…Since our graphs have three nodes, the most straightforward way for doing this is to apply a triangulation technique 2 . This has the advantage of selecting very small number of graphs which can be used as the representatives of the probe images 3 .…”

“…Given G i = {( 2 A triangulation of a discrete set of points P is a subdivision of the convex hull of the points into simplices such that any two simplices intersect in a common face or not at all and the set of points that are vertices of the subdividing simplices coincides with P. 3 Triangulation could not be a good choice for extracting graphs in the training stage, because then the number of extracted graphs from training images would not be sufficient for reasonably estimating the graph spaces (discussed in Section 3). we are interested in obtaining the probability of G i belonging to each model set (individual) to assign it to the one that maximizes the posterior probability:…”

“…PCA+LDA [2] provides a linear transformation on PCAprojected feature vectors, by maximizing the betweenclass variance and minimizing the within-class variance. The BIC algorithm [3] projects the feature vector onto extra-personal and intra-personal subspaces and computes the probability that each feature vector came from one or the other subspace.…”

Towards robust face recognition from video

“…Bayesian face recognition [21,22,23] has been proposed to improve robustness in the presence of varying illumination and expression. These approaches employ probabilistic models to characterize intrapersonal and interpersonal differences with a principal component analysis (PCA) or "eigenface" representation.…”