Automatic cylinder detection using Hough Transform. T.Rabbani t.rabbani@citg.tudelft.nl November 23, 2018 Section of Photogrammetry and Remote Sensing
Hough Transform for automatic cylinder detection 85% of objects found in industrial scenes can be approximated by planes, spheres, cones and cylinders 5-free parameters for cylinder. November 23, 2018
Hough Transform for automatic cylinder detection Space Complexity Time complexity Example: S=100 Cs = 9.3 GB S=200 Cs = 300 GB Solution: Two-step approach Orientation Estimation Position and Radius estimation November 23, 2018
Basic Idea Behaviour of normals on Gaussian sphere November 23, 2018
Step 1: Orientation Estimation November 23, 2018
Step 2: Position and Radius Estimation November 23, 2018
Uniform Sampling of orientation space Uniform sampling essential for Hough transform Place points so that each point represents equal area on sphere surface No closed form solution Iterative solution too slow, as Hough transform needs bi-directional mapping November 23, 2018
Spherical sampling linear sampling of and Highly nonlinear November 23, 2018
Cartesian sampling Use implicit equation of sphere Uniformaly sample x and y November 23, 2018
Approximate Uniform Sampling Number of samples in phi = For mapping back and forth: November 23, 2018
Approximate Uniform Sampling November 23, 2018
Example November 23, 2018
Example November 23, 2018
Results November 23, 2018
Results on NAM modeling 20 million points Processing: Segmentation Automatic cylinder and plane detection Planar patches: 946 Cylinders: 1392 November 23, 2018
Problem: Model Selection Domains of planes and cylinders overlap A cylinder can be represented by a number of planar patches A plane by a set of cylindrical segments (a) (b) November 23, 2018
Solution: Hypothesis verification Hough transform will always give a hit Consider each hit as a strong hypothesis Employ model matching for hypothesis verification Accept only if hypothesis verification is successful November 23, 2018
Planarity Test Histogram of Distances Angles with normal Aspect ratio = Max Extent/Min Extent Max-extent Min-extent November 23, 2018
Cylindricality Test Difference between predicted and actual normal Distribution of points on cylinder surface Max min radius theta November 23, 2018
Results on NAM modeling November 23, 2018
Future Plans Results refinements Merging of multiple hits Connection of separated segments due to occlusions etc Automatic constraint detection Detection and fitting of curves November 23, 2018
Questions: November 23, 2018