Enhanced Defect Detection in NDE Using Pixel Level Data Fusion

Pipeline systems in service continue to degrade with passage of time due to usage. Magnetic flux leakage (MFL) and Eddy Current (EC) based Non destructive evaluation (NDE) methods are most common for inline inspection (ILI) of the metallic pipelines. Many times, a single measurement technique fails to reveal all the characteristics of the damage due to resolution constraints. This paper presents a novel pixel based fusion of both MFL and EC inspection that can detect real defects like cracks and slits of sub millimeter dimension. An automated data based screening rule is developed to judge whether fusion is needed. Then a transform domain based fusion method called complementary sparse representation (CSR) along with post processing steps is developed for fusion of the source images. The efficacy of the fusion algorithm is studied on both the MFL and EC data generated by simulation means.