A Novel Fuzzy Edge Detection and Classification Scheme to Aid Hydro-dams Surface Examination

The visual examination of downstream faces of hydro dams is an important component in their monitoring. Few computer vision algorithms exist for this task. Some of them aim to detect and evaluate concrete cracks using edge detection. Other problems can also affect hydro-dams, as roughness, scratches, and can lead in time to surface deterioration. These are not currently reliably addressed by image analysis. The proposed fuzzy edge detection and classification scheme can assist the human expert in the visual inspection for crack examination and non-critical surface faults. To differentiate between them we use a fuzzy c-means classifier on the edge strengths, which are found using Sobel edge detection and enhanced through a fuzzy algorithm. The scheme is validated through experiments on different images. The resulting edge maps emphasize superficial faults or cracks, depending on the user’s selection, providing more accurate results than classical edge detection.