AUTOMATED PARAMETER EXTRACTION FOR ULTRASONIC FLAW ANALYSIS

To make a decision on the nature of a defect contained within a weld s pecimen, it is necessary to reliably detect suspect regions in the ult rasonic inspection images, derive geometrical such as shape, size, pos ition and from each indication and, finally, collate these parameters intelligently by associating each indication with a possible defect ty pe. This procedure is discussed for the case when the segmentation of indications and parameter calculation procedures are performed by the authors' NDT Workbench facility. A real-flaw example, inspected by a s eries of probes, is used to demonstrate the final defect categorisatio n decision. The indication parameters derived during this process can be used either to aid the manual interpreter or as part of a knowledge based system (KBS).