FIG. 3 shows an exemplary mapping of pixels (represented by circles) of the input image 12 into a devised discriminant space that is defined along five axes (or dimensions) corresponding to five different types of features (i.e., Feature 1, Feature 2, . . . , Feature 5).