Skeleton-based noise removal algorithm for binary concrete crack image segmentation

Image processing methods for automated concrete crack detection are often challenged by binary noise. Noise removal decrease the false positive pixels of results, at cost a reduction in true positives. This paper proposes novel method noise and segmentation noisy images. The applies an area threshold before reducing pixel groups image to skeleton. Each skeleton is connected its nearest neighbour remaining short skeletons removed using length threshold. A morphological reconstruction follows remove all elements original that do not intersect with Finally, close proximity endpoints resulting reinstated. Testing was conducted on dataset images; proposed (Skele-Marker) obtained recall, precision, intersection over union, F1 score results 77%, 91%, 72%, 84%, respectively. Skele-Marker compared other found literature outperform terms union score. used make more reliable, supporting ever-growing demand inspections structures.