Robust ultrasonic flaw detection using order statistic CFAR threshold estimators

Ultrasonic nondestructive imaging of critical defects is often hampered by the presence of interfering and random scatterers (i.e., clutter or grain echoes) associated with the defect´s environment. The detection of broadband ultrasonic echoes for nondestructive evaluation (NDE) can be improved by using optimal bandpass filtering to resolve flaw echoes surrounded by grain scatterers. The optimal bandpass filtering is achieved by examining spectral information of the flaw and the grain echoes where frequency differences have been experimentally shown to be predictable. The flaw echoes can then be discriminated by using adaptive thresholding techniques based on surrounding range cells. The authors present order statistic (OS) based processors (ranked and trimmed mean) to robustly estimate the threshold while censoring outliers. The design of these OS processors is accomplished analytically based on constant false alarm rate (CFAR) detection. The detector performance is compared with the cell averaging method. It is shown that the OS based CFAR processors can detect flaw echoes robustly