Pothole 3D Reconstruction With a Novel Imaging System and Structure From Motion Techniques

Machine vision based evaluation systems are receiving increased attention, day by day, for automated quality inspection of roads. Industrial pavement scanners consist laser and very expensive, hence inaccessible everyone. The proposed work presents a simple novel approach 3D reconstruction potholes an road surface evaluation. technique utilizes Structure from Motion algorithm, along with triangulation, to generate point clouds potholes. Alongside, low-cost system, consisting single camera pointer, is also proposed. Keypoint matching techniques employed, the 5-point on successive image frames cloud. However, this cloud not metric yet, without scale information. ambiguity solved making use using principle triangulation. spot detected in same sequence that used point-cloud building, cutting down capturing processing overhead. system has been benchmarked artificial indentations known dimensions, proving robustness measurement scheme hardware. Static dynamic tests have performed. mean depth errors made imager statically at speeds 10 km/hr, 15 20 km/hr 5.3%, 7.9%, 14.4%, 26.6%, whereas perimeter 5.2%, 6.83 %, 11.8%, 27.8%. proposed, shows promising results generating