Corrosion discrimination method from surface images

Igarashi, Takahiro  ; Otani, Kyohei   ; Aoyama, Takahito   ; Kato, Chiaki   

It is one of the important issues to understand the corrosion condition inside the material nondestructively in the integrity assessment of large structures such as bridges and plants. In this presentation, we attempted to determine the corrosion state from the surface image before rust removal by using the contour extraction method based on the discontinuity of luminance in the neighboring pixels of the image and the feature point extraction method using the gradient of the corner and luminance of the neighboring pixels. As a result, it was found that the Adaptive Gaussian and Canny methods could be used for contour extraction, and the Agast Feature Detector method could be used for feature point extraction to predict the vicinity of corrosion points.