Simultaneous Extraction of Functional Face Subspaces

Facial variation divides into a number of functional sub- spaces. An improved method of measuring these was de- signed, within the space defined by an Appearance Model. Initial estimates of the subspaces (lighting, pose, identi ty, expression) were obtained by Principal Components Ana- lysis on appropriate groups of faces. An iterative algorith m was applied to image codings to maximise the probability of coding across these non-orthogonal subspaces before ob- taining the projection on each sub-space and recalculating the spaces. This procedure enhances identity recognition, reduces overall sub-space variance and produces Principal Components with greater span and less contamination.