Automatic crack identification using a novel 3D profilometry-based magnetic particle testing method

Abstract

In magnetic particle testing (MPT), magnetic particles are utilized to form magnetic particle indication (MPI) on cracks. Identifying MPI using the visual information of tested objects is the most popular approach for crack detection. However, it requires MPI to exhibit visual distinctiveness from its surroundings. Thus, previous approaches have conducted special treatments on magnetic particles to display MPI, leading to a series of limitations in MPT, especially environmental pollution. This paper shows that MPI has measurable three-dimension (3D) profile, and consequently proposes a 3D profilometry-based magnetic particle testing method (3DPMPT). Unlike visual identification, our approach detects cracks through profile changes caused by MPI on tested objects without any special treatment of magnetic particles. Furthermore, an algorithm for MPI identification is presented. It calculates the MPI profile by comparing the profile of the tested objects before and after the addition of magnetic particles to promote the automation of 3DPMPT. The algorithm includes details of the comparison, denoising based on Euclidean clustering, and removing the profile of fake magnetic particle indication (FMPI) based on the computed results from digital model. Experimenting with artificial and natural cracks on several tested objects, we demonstrate that 3DPMPT has the potential to be employed in automatic ferromagnetic material detection.