A frequency domain approach to synthesizing near optimum edge detection filters

A frequency domain approach to synthesizing near optimum edge detection filters is presented. This approach is based on the "method of ideal filter design"[1] which involves first deriving an ideal filter with a weighted least-squares-error realizable filter. For the spatial edge detection application, the high resolution ideal detection filter is employed along with some form of spectral smoothing or averaging to reduce an unwanted cross spectral density term. An example is presented to demonstrate the approach and results of filter application. The synthesis procedure presented here is most applicable to adaptive systems employing an FFT.