Pipe corrosion recognition through image processing using fault detector robot

Abstract

In industrial literature, damages that occurred due to various reasons on the material are called corrosion. Corrosion cause fatigue and failure of systems and other risks such as financial losses for replacing the system, leakage and pollution of contacted products. In accessible surfaces corrosion detection is done easily, but in cases such as tanks, pipes and particularly long tubes, there is no access to the inside of the pipes so more complex systems are needed. In this research, a new approach is proposed to detect corrosions in industrial pipes. The proposed method is based on image processing algorithms hence it is a kind of non-destructive inspection method. The proposed method offers a new innovative processing algorithm to identify corrosion and also provides a proper lighting method. For this purpose, first the correct lighting system is performed , then obtained images from inspection would be preprocessed for detection phase. Preprocessing step includes color format changing, denoising and smoothing operations. Then corrosions are identified by edge detection algorithms and amount of it, is estimated by morphological operations.