Abstract .
The paper proposes a new and powerful technique to determine the optimal values of coefficients of McClellan transformation used to map very effectively 1-D filter prototypes in 2-D FIR filters. The new algorithm permits the mapping of 1-D prototype in an arbitrary chosen 2-D contour. The algorithm minimizes all along the given contour, in a least squares sense, the total error between the ideal contour of the imposed boundary of the 2-D filter bandpass and an isopotential curve described by the McClellan transform. Additional initial designing conditions can be expressed as constraints for this minimization problem, which is solved as a constrained least squares minimization problem. Numerical examples, which include circularly symmetric and diamond-shaped filters, elliptically symmetric and 2-D fan filters demonstrate by lower error values attained by comparison with previous works the effectiveness and the usefulness of this method.