Enhancement of magnetic flux leakage images of defects in carbon steel using Eigen vector based approach

Abstract

Magnetic flux leakage images of defects are affected by incoherent noise due to non-linear permeability, surface roughness, stresses and lift-off variations. In this paper, an approach that utilizes the energy packing property of the Eigen values is used to process the tangential component images of machined notches in carbon steel plates. These images are obtained by scanning Giant-magneto resistive (GMR) sensor. In this approach, Eigen values and vectors of image covariance matrix are determined and images are reconstructed after considering the most significant Eigen pair that has the information of defects. The quality of reconstruction is assessed using signal to noise ratio and a new factor called noise reduction factor. Significant enhancement in image quality is observed enabling reliable detection of sub-surface notch located 8.6 mm below surface and a 1.1 mm deep surface-breaking notch.