Optimal-trade-off filters for noise robustness, peak sharpness, and light efficiency in nonoverlapping background noise

A design of filters for pattern recognition is presented that have optimal trade-off for the criteria of noise robustness, peak sharp- ness, and Horner efficiency, when input scene noise is spatially disjoint (nonoverlapping) with the target. Three different criteria—signal-to-noise ratio, peak-to-output energy ratio, and light efficiency—are used to de- sign the filter. Computer simulations for the various types of noise and parameters illustrate the performance of such filters for optical pattern recognition.