The main objective of the present investigation was to evaluate median filters and median/inverse filters as image preprocessors for facilitating machine recognition. An experiment for preprocessor evaluation is discussed, taking into account a preprocessor which consists of either a median filter alone or a combination of median/inverse filters. It is found that median filtering offers the most effective preprocessing for enhancing machine recognition in cases in which images are contaminated by impulse-type noise. Median filtering is also effective in cases in which images are degraded by blurs and impulse-type noise. Median/inverse filtering improves the performance for cases of images degraded by both Gaussian white noise and blurs.