Morphological Processing for Crack Detection in Eddy Current Images of Jet Engine Disks

With the growing use of air transportation, aerospace industries are looking towards automated airplane inspection systems that can provide more rapid and economical services and achieve greater safety. Eddy current nondestructive testing technique is one of the most commonly used methods in aircraft inspection. One demanding application of the eddy current technique is the inspection of jet engine disks. In rotary fan disks in jet engines, cracks often occur in high stressed regions of blade slots such as chamfered edges. This presents problems in terms of flaw detection using the eddy current method since the crack signals are overwhelmed by the edge signals. Therefore, a fairly robust algorithm to extract flaw signals and eliminate contribution from other sources is needed. In this paper, a novel method based on morphological operations is proposed for crack detection in eddy current images of jet engine disks. The method is specially designed to eliminate contribution to the eddy current signals from the large scale geometry of the disk such as slot edges while preserving small scale signals such as indications due to cracks. Initial classification results demonstrate the capability of the proposed method in extracting flaw features of different scales. The algorithm is also simple, fast, and robust in presence of noise.