“…We use 8 sequences comprising over 500 images in the experiment. Both algorithms are applied to the same set of face sequences using the FastNCC algorithm (Tiddeman and Chen, 2007), Gauss-Newton algorithm (Hager and Belhumeur, 1998), (Matthews and Baker, 2004), Powell's method (Press. et al, 2007) as the Figure 9: Each row consists of a set of selected frames from a tracking sequence with the synthetic texture patches drawn on which indicates the location of the features.…”