Mesh Edge Detection and Sharp Edge Reconstruction

1. Mesh Edge Detection and Sharp Edge Reconstruction Speaker:Ma HaoDi Sep. 27, 2007

2. Author Markus Gross: • A professor of computer science, chair of the institute of computational science, and director of the Computer Graphics Laboratory of the Swiss Federal Institute of Technology (ETH) in Zürich. • Gross was a papers co-chair of the IEEE Visualization '99, the Eurographics 2000 and the IEEE Visualization 2002 conferences. He was chair of the papers committeeof ACM SIGGRAPH 2005. • His research interests include point-based graphics, physically-based modeling, multiresolution analysis, and virtual reality.

3. Author Charlie C. L. Wang(王昌凌): • An Assistant Professor at the Department of Mechanical and Automation Engineering, the Chinese University of Hong Kong. • (1998) Mechatronics Engineering from Huazhong University of Science and Technology, Ph.D. (2002) in Mechanical Engineering from the Hong Kong University of Science and Technology. • A member of IEEE and ASME.

4. Reference • Incremental reconstruction of sharp edges on mesh surfaces Charlie C.L. Wang * Computer-Aided Design 38 (2006) 689–702 • Multiresolution Feature Extraction for Unstructured Meshes Andreas Hubeli, Markus Gross IEEE Visusualization 01, 2001

5. Background • sharp edges and corners are degraded on the resultant surface of evolution: subdivision; restructuring;fairing

6. Motivation • Reconstruction or retain feature information including sharp edges or ridge lines.

7. Mesh Edge Detection

8. Methodology • Classification Phase • Selection of Feature Edges • Patch Construction • Skeletonizing

9. Methodology

10. Second Order Difference(SOD)

11. Extended Second Order Difference(ESOD)

12. Best Fit Polynomial(BFP)

13. Best Fit Polynomial(BFP)

14. Angle Between Best Fit Polynomials(ABBFP)

15. Angle Between Best Fit Polynomials(ABBFP)

16. Results(ABBFP:differentsupport)

17. Detection Phase

18. Detection Phase • First, a subset of feature edgesis constructed. • Hysteresis Thresholding • Next, Construction of the Patches • Finally, the line-type features are extracted • line-type features are extracted using a skeletonizingalgorithm

19. Detection Phase:(step 1  2)

20. Detection Phase:(step 2  3)

21. Detection Phase:(step 2  3) • for all edges e in patch • if (isBoundaryEdge(e) == true) • edgeList.insert(e); • while edgeList is not empty do { • e = edgeList.front(); // Retrieve the first edge • edgeList.pop_front(); // Remove it from the list • if(belongsToPatch(e) == false) { • removeFromPatch(e); • edgeList.insert(newBoundaryEdges); • } • } • }

22. Results

23. Reconstruction of Sharp Edge

24. Methodology • Signals indicating sharpness • UUSOD • Surface sharp edges reconstrction • Geometry predictor

25. Uniformly Supported Second-Order Difference(USSOD) • where d(v,f) returns the Euclidean distance from v to the pointset of f (i.e. not the plane holding f). • P(f i,f j)representing the inner product of unit normal vectors on two faces • ? (uniform support size)

26. Uniformly Supported Second-Order Difference(USSOD) • Better but not completely solve

27. Uniformly Supported Second-Order Difference(USSOD)

28. Geometry predictor • Some definition • static vertices • sharp vertex • static triangle ( all static vertices ) • dynamic triangles( one sharp vertices )

29. Geometry predictor • An ideal position for a vertex v minimizes thedifferencebetween its position and thesmoothness signals—tangentplanes: • contains the static triangles near the vertex v • is the unit normal vector of f • is a point on the static triangle f

30. Geometry predictor

31. Progressive surface prediction • red circles represent sharp vertices and white circles denote static vertices.

32. Results

33. Results

34. Results

35. Results

36. Limitations • The feature that blends smoothly into a flat area may be miss-sharpened • Some unwanted sharpening will be given on small radius • Not a adaptive sharpness identification techniques

37. Results

38. THE END