Millimeter wave imaging of notches in metal specimens under dielectric coating using image processing

Structural health monitoring is one of the main concerns in infrastructure engineering since it focuses on evaluating the life span of structures and detection of defects and damages, which may lead to sudden failure of structural components. This is why, in recent years, defect and damage detection and evaluation have received a wide attention among researchers. More specifically, a local structural health monitoring based on non-destructive testing techniques is an efficient approach to identify the defects and damages of structures in early stages to evaluate the remaining life span, improving safety and health of structures and increasing their serviceability. Therefore, the development and application of non-destructive testing and evaluation techniques including microwave and millimeter wave imaging of structures are very important. This paper presents the application of a relatively simple millimeter wave continuous wave reflectometer with an open-ended waveguide antenna and the developed image processing algorithms for the purpose of detection and evaluation of flaws such as cracks and notches on metal surface covered by dielectric coating such as rubber coating. It is shown that the original images based on measured data may not provide desired information, leading to developing the image processing algorithms to enhance the imaging results. Three different algorithm-based solutions are proposed and developed in this paper for automatic damage detection in infrastructures which complement the edge-based damage detection methods in hazardous scenarios. The proposed algorithms are based on Sobel, Canny, and fuzzy c-mean thresholding algorithms developed using Otsu's thresholding method to facilitate the structural damage identification. The notches in this paper represent damages in the metal plate such as cracks. The proposed algorithms have been compared together and with the traditional ones to investigate their performance and advantages. The proposed algorithms can successfully enhance the quality of images and visualize the damage on the metal surface under dielectric coating.