A Heuristic Algorithm for the Reconstruction and Extraction of Defect Shape Features in Magnetic Flux Leakage Testing
Alimey Fred John,Libing Bai,Yuhua Cheng,Haichao Yu
DOI: https://doi.org/10.1109/TIM.2020.2998561
IF: 5.6
2020-01-01
IEEE Transactions on Instrumentation and Measurement
Abstract:The first step to quantifying complex defects is knowing its geometry, shape, and orientation. This is evidently challenging in electromagnetic nondestructive testing (ENDT), especially for subsurface complex defect detection, in cases where no information about defect or material is known or given. In this article, we propose a heuristic approach for the visualization, verification, and validation of complex defects in magnetic flux leakage (MFL) testing. This method is based on MFL experiment using magneto-optical images (MOIs) that are obtained from four different magnetization patterns. Using the proposed magnetization patterns, images of complex defect are captured with the aid of a charge-coupled device (CCD) camera, based on the interaction of magnetic field distribution and detected defects, from different angles and directions. An enhanced ant colony algorithm (EACA) is then used to reconstruct complex defect shapes from captured images using a mean image approach. The reconstructed image (mean image) reveals the defect shape with high precision for which the EACA is able to extract important defect features such as specific edges that might be hidden or blurred. This approach based on results has proved to provide promising solution to visually verifying complex defects in MFL, leading to a more simplified and faster way to characterize these defects as compared with conventional methods, using their shapes and orientation.