Frequency diversity ultrasonic flaw detection using order statistic filters

The authors present the theory and application of order statistic (OS) filters in ultrasonic flaw detection problems. The theory suggests that the optimal rank can be founded upon the knowledge of the distribution of flaws and grains. Experimental and simulation results for the OS filters are presented for flaws embedded in steel specimens. OS filters and averaging algorithms are compared with regard to their application for flaw detection. It is shown that OS filters can improve the flaw-to-clutter echo ratio better than averaging algorithms for a broader class of clutter and flaw echo statistics.