Notice that the hyperbolic paraboloid has four parameters and can be made
to fit through all four points in the neighborhood of Fig. 5.2. Similarly, the
bicubic function has 16 parameters and can fit all 16 points in a 4 4 neigh-
borhood. If a higher-order interpolating function has the same number of
coefficients as the neighborhood has points, then the interpolating surface can
likewise be made to fit at every point. However, if there are more points in the
neighborhood than there are coefficients, then the surface cannot fit all the
points, and a curve-fitting or error-minimization procedure must be used.
Higher-order interpolating functions that are widely used include cubic splines,
Legendre centered functions, and the truncated sinc(x) function