1 / 96

3D photography (II)

3D photography (II). Digital Visual Effects, Spring 2005 Yung-Yu Chuang 2005/5/25. with slides by Szymon Rusinkiewicz, Richard Szeliski, Steve Seitz and Brian Curless. Announcements. Final project will be online tomorrow Proposal presentation on next Wednesday

dgwinn
Télécharger la présentation

3D photography (II)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. 3D photography (II) Digital Visual Effects, Spring 2005 Yung-Yu Chuang 2005/5/25 with slides by Szymon Rusinkiewicz, Richard Szeliski, Steve Seitz and Brian Curless

  2. Announcements • Final project will be online tomorrow • Proposal presentation on next Wednesday • I will send out your current grades by next Wednesday • Scribe (SIGGRAPH 2005, CVPR 2005, readings) • Schedule for the next few weeks • 6/1 proposal • 6/8 making face/human • 6/15 random topics • 6/28 final project presentation

  3. Outline • Range acquisition techniques • Full model reconstruction • ICP • Volumetric reconstruction • Systems, projects and applications • Final project

  4. Range acquisition

  5. Range acquisition

  6. Range acquisition taxonomy mechanical (CMM, jointed arm) inertial (gyroscope, accelerometer) contact ultrasonic trackers magnetic trackers industrial CT rangeacquisition transmissive ultrasound MRI radar non-optical sonar reflective optical

  7. Range acquisition taxonomy shape from X: stereo motion shading texture focus defocus passive opticalmethods active variants of passive methods Stereo w. projected texture Active depth from defocus Photometric stereo active time of flight triangulation

  8. Passive approaches stereo space carving

  9. Active approaches Cyberware whole body scanner shadow scanning

  10. Active variants

  11. Full model reconstruction

  12. 3D Model Acquisition Pipeline 3D Scanner

  13. View Planning 3D Model Acquisition Pipeline 3D Scanner

  14. View Planning Alignment 3D Model Acquisition Pipeline 3D Scanner

  15. View Planning Alignment Merging 3D Model Acquisition Pipeline 3D Scanner

  16. Problem • Align two partially-overlapping meshesgiven initial guessfor relative transform

  17. Aligning 3D data • If correct correspondences are known,it is possible to find correct relative rotation/translation

  18. Find the optimal transformation

  19. Find the optimal transformation

  20. Find the optimal transformation

  21. Aligning 3D data • How to find corresponding points? • Previous systems based on user input,feature matching, surface signatures, etc.

  22. Aligning 3D data • Alternative: assume closest points correspond to each other, compute the best transform…

  23. Aligning 3D Data • … and iterate to find alignment • Iterated Closest Points (ICP) [Besl & McKay 92] • Converges if starting position “close enough“

  24. ICP variants • Variants on the following stages of ICPhave been proposed: • Selecting source points (from one or both meshes) • Matching to points in the other mesh • Weighting the correspondences • Rejecting certain (outlier) point pairs • Assigning an error metric to the current transform • Minimizing the error metric

  25. ICP variants • Selecting source points (from one or both meshes) • Matching to points in the other mesh • Weighting the correspondences • Rejecting certain (outlier) point pairs • Assigning an error metric to the current transform • Minimizing the error metric

  26. Selecting source points • Use all points • Uniform subsampling • Random sampling • Normal-space sampling • Ensure that samples have normals distributedas uniformly as possible

  27. Normal-space sampling Uniform Sampling Normal-Space Sampling

  28. Random sampling Normal-space sampling Normal-space sampling • Conclusion: normal-space sampling better for mostly-smooth areas with sparse features

  29. ICP variants • Selecting source points (from one or both meshes) • Matching to points in the other mesh • Weighting the correspondences • Rejecting certain (outlier) point pairs • Assigning an error metric to the current transform • Minimizing the error metric

  30. Matching • Matching strategy has greatest effect on convergence and speed • Closest point • Normal shooting • Closest compatible point • Projection

  31. Find closest point in other mesh Closest-point matching Closest-point matching generally stable, but slow and requires preprocessing

  32. Project along normal, intersect other mesh Normal shooting Slightly better than closest point for smooth meshes, worse for noisy or complex meshes

  33. Closest compatible point • Can improve effectiveness of both of the previous variants by only matching to compatible points • Compatibility based on normals, colors, etc. • At limit, degenerates to feature matching

  34. Projection to find correspondences • Finding closest point is most expensive stage ofthe ICP algorithm • Idea: use a simpler algorithm to find correspondences • For range images, can simply project point [Blais 95]

  35. Projection-based matching • Slightly worse performance per iteration • Each iteration is one to two orders of magnitude faster than closest-point

  36. ICP variants • Selecting source points (from one or both meshes) • Matching to points in the other mesh • Weighting the correspondences • Rejecting certain (outlier) point pairs • Assigning an error metric to the current transform • Minimizing the error metric

  37. ICP variants • Selecting source points (from one or both meshes) • Matching to points in the other mesh • Weighting the correspondences • Rejecting certain (outlier) point pairs • Assigning an error metric to the current transform • Minimizing the error metric

  38. Point-to-plane error metric • Using point-to-plane distance instead of point-to-point lets flat regions slide along each other [Chen & Medioni 91]

  39. High-speed ICP algorithm • ICP algorithm with projection-based correspondences, point-to-plane matchingcan align meshes in a few tens of ms.(cf. over 1 sec. with closest-point)

  40. Range processing pipeline • Steps 1. manual initial alignment 2. ICP to one existing scan 3. automatic ICP of all overlapping pairs 4. global relaxation to spread out error 5. merging using volumetric method

  41. Range processing pipeline • Steps 1. manual initial alignment 2. ICP to one existing scan 3. automatic ICP of all overlapping pairs 4. global relaxation to spread out error 5. merging using volumetric method

  42. Volumetric reconstruction

  43. Signed distance function

  44. Overview

  45. Weighting

  46. Volumetric integration (Curless and Levoy, Siggraph´96) range surfaces signed distance to surface volume weight (~accuracy) distance depth sensor surface1 • use voxel space • new surface as zero-crossing • (find using marching cubes) • least-squares estimate • (zero derivative=minimum) surface2 combined estimate

  47. Isosurfacing: marching cubes First 2D, marching squares

  48. Marching cubes

  49. Marching cubes

  50. Results

More Related