Thesis title: Inspection of infrastructures to detect the corrosion by machine vision
The bridges are used widely in the transportation system due to economic production, speed of construction versatility. The traditional method of bridge inspection includes the visual and physical activities of the various elements, which require skillful operators, also Non-destructive testing (NDT) recognizes the defects for infrastructure with advantage and disadvantage characteristics according to damage classification. Corrosion is a major problem within transportation infrastructure that propagated on the surface based on the weather condition and material property, etc. The performance and age of the bridge reduced by corrosion propagation therefore the effects of parameters in the corrosion creation investigate by the experimental test in the laboratory.
In this study, the clustering algorithm of machine vision is developed in the bridges are one of the critical structures with a huge volume of different corrosion. The behavior of the corrosion simulates from the measurement of yield & tensile strength, hydrogen concentration and micro structural examination. The image processing algorithm utilizes to defect detection of railway steel truss, pedestrian bridge, and mixed steel-concrete structure according to data based approach. The accuracy of defects detection in the steel structure improved by digital image recognition of RGB technique to automatic detection and inspection of the bridge over time. A new methodology of K-means clustering algorithm propose by machine vision concepts to compare the result of segmentation in edge detection. This approach can detect different types of defects in the images which are captured from the structure through MATLAB R2021a. The outcome of this research eliminates the routine inspection schedule through the predict the defects and FEM analysis to simulate the corrosion and relative Non-Destructive Testing (NDT) by ANSYS MAXWELL & COMSOL.