THE SYSTEM OF CARGO DETECTION IN OPEN WAGONS BASED ON GENERATING A 3D MODEL
The paper presents the system of cargo detection in open wagons. The system consists of obtaining a 3D open wagon model from a sequence of wagon 2D images and its subsequent analysis. The model is a dense point cloud generated by the fusion of depth maps with rejected erroneous depth estimates. The model analysis include the preliminary model preparing, cargo location estimation, and finding clusters in a model cloud corresponding to the cargo objects. A possible cargo location is a minimal axis aligned rectangular cuboid containing all cargo object clusters.
Authors: V. V. Vihlyav, S. A. Kulakov, G. V. Razumovskiy
Direction: Informatics and Computer Technologies
Keywords: 3D model, structure from motion, clustering, DBSCAN, affine transforms, multi-view stereo, photogrammetry, depth map fusion, unorganized point cloud, cargo detection, 3D Hough transform
View full article