The Applicaton of a Dynamic Threshold to C-Scan Images with Variable Noise

Abstract

The high-strength alloys used in the aerospace industry for rotating parts are inspected ultrasonically for material anomalies which may have resulted from the manufacturing process. These materials, such as titanium alloys, often have a large macro-grain structure which limits the sensitivity of the ultrasonic inspection to material anomalies. An improved inspection for cylindrical titanium alloy billet was implemented in production in 1995. This inspection, also called the multizone inspection, utilizes bicylindrically focused transducers to inspect the billet using multiple zones. It also features digital C-scan image acquisition and a dual acceptance criterion based on both amplitude relative to a 2/64″ flat-bottom hole (FBH) calibration target and C-scan signal-to-noise ratio (SNR).[1] The SNR was calculated manually by the operators using graphical user interface based image analysis software. The SNR-based criterion has proved to be a more effective detector of material anomalies than the amplitude criterion.[2] It is, however, susceptible to operator variability. The operator variability was leading to a larger than expected number of false positive indications during production inspections. A false positive indication is one which did not meet the acceptance criterion of the test, but no corresponding material anomaly was found on metalography. To resolve this problem, a dynamic threshold algorithm for ultrasonic C-scan images is applied to the images to automatically identify indications which exceed either the amplitude or the SNR acceptance criterion.