SCALED AND SCALING-FREE McCLELLAN TRANSFORMATIONS FOR THE DESIGN OF 2-D DIGITAL FILTERS
In this paper, the scaling problem of McClellan transformation used for the design 2-D digital filters with elliptical and circular cutoff contours is critically analyzed. Several new and extremely simple formulas for calculating the maximum and minimum values of the transformation function are presented. Using these extreme values, simple formulas for scaling factors and scaled McClellan transformations are derived for the various cases of each type of contour. Using the necessary and sufficient conditions for a scaling-free transformation, the scaling-free McClellan transformations are also derived. It is shown that the number of independent coefficients in them to be optimized varies from 0 to 2 depending on the case, compared to 3 in the original transformation. They are very useful in deriving analytical expressions for the transformation coefficients, and the 2-D filters designed by employing them have a smaller number of multiplications per output sample and hence, attractive for real-time applications. A new scaling formula is presented and some of its properties and effects on the derived formulas are discussed. The major advantage of this new formula is that from the unscaled transformation coefficients for a given type of 2-D filter, we can generate the scaled transformation coefficients for a different type of 2-D filter. Some examples are presented to demonstrate the usefulness of the derived formulas.