Temporal Based Approach for Face Information: A Modified Solution to AAM
To gain high efficiency, accurate alignment and for handling face deformations we use Temporal Matching Active Appearance Model(TM-AAM). The major component in judging a person’s intention, personality and effective state are facial expressions and human face is a rich source of information for the viewer. The specialty of TM-AAM is that it can generalize to any valid example as it contains a statistical model of the shape and grey level appearance of the object of interest. Robust and fast deformable image matching can be done by TM-AAM by combining texture and shape models. Image matching can be done by measuring the residuals and use the model for current parameter changes prediction to yield a better fit. This technique is widely used in high surveillance cameras, fugitive identification, security checks and medical interpretations.