Download presentation
Presentation is loading. Please wait.
2
Computer Vision Optical Flow Marc Pollefeys COMP 256 Some slides and illustrations from L. Van Gool, T. Darell, B. Horn, Y. Weiss, P. Anandan, M. Black, K. Toyama
3
Computer Vision 2 last week: polar rectification
4
Computer Vision 3 Last week: polar rectification
5
Computer Vision 4 Stereo matching Optimal path (dynamic programming ) Similarity measure (SSD or NCC) Constraints epipolar ordering uniqueness disparity limit disparity gradient limit Trade-off Matching cost (data) Discontinuities (prior) (Cox et al. CVGIP’96; Koch’96; Falkenhagen´97; Van Meerbergen,Vergauwen,Pollefeys,VanGool IJCV‘02)
6
Computer Vision 5 Hierarchical stereo matching Downsampling (Gaussian pyramid) Disparity propagation Allows faster computation Deals with large disparity ranges ( Falkenhagen´97;Van Meerbergen,Vergauwen,Pollefeys,VanGool IJCV‘02)
7
Computer Vision 6 Disparity map image I(x,y) image I´(x´,y´) Disparity map D(x,y) (x´,y´)=(x+D(x,y),y)
8
Computer Vision 7 Example: reconstruct image from neighboring images
9
Computer Vision 8 Multi-view depth fusion Compute depth for every pixel of reference image –Triangulation –Use multiple views –Up- and down sequence –Use Kalman filter (Koch, Pollefeys and Van Gool. ECCV‘98) Allows to compute robust texture
10
Computer Vision 9 Real-time stereo on graphics hardware Computes Sum-of-Square-Differences Hardware mip-map generation used to aggregate results over support region Trade-off between small and large support window Yang and Pollefeys CVPR03 140M disparity hypothesis/sec on Radeon 9700pro e.g. 512x512x20disparities at 30Hz Shape of a kernel for summing up 6 levels
11
Computer Vision 10 Sample Re-Projections near far
12
Computer Vision 11 (1x1) (1x1+2x2) (1x1+2x2 +4x4+8x8) (1x1+2x2 +4x4+8x8 +16x16) Combine multiple aggregation windows using hardware mipmap and multiple texture units in single pass video
13
Computer Vision 12 Cool ideas Space-time stereo (varying illumination, not shape)
14
Computer Vision 13 More on stereo … The Middleburry Stereo Vision Research Page http://cat.middlebury.edu/stereo/ Recommended reading D. Scharstein and R. Szeliski. A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms. IJCV 47(1/2/3):7-42, April-June 2002. PDF file (1.15 MB) - includes current evaluation. Microsoft Research Technical Report MSR-TR-2001-81, November 2001. PDF file (1.27 MB).PDF file
15
Computer Vision 14 Jan 16/18-Introduction Jan 23/25CamerasRadiometry Jan 30/Feb1Sources & ShadowsColor Feb 6/8Linear filters & edgesTexture Feb 13/15Multi-View GeometryStereo Feb 20/22Optical flowProject proposals Feb27/Mar1Affine SfMProjective SfM Mar 6/8Camera CalibrationSilhouettes and Photoconsistency Mar 13/15Springbreak Mar 20/22SegmentationFitting Mar 27/29Prob. SegmentationProject Update Apr 3/5Tracking Apr 10/12Object Recognition Apr 17/19Range data Apr 24/26Final project Tentative class schedule
16
Computer Vision 15 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
17
Computer Vision 16 Optical Flow: Where do pixels move to?
18
Computer Vision 17 Motion is a basic cue Motion can be the only cue for segmentation
19
Computer Vision 18 Motion is a basic cue Motion can be the only cue for segmentation
20
Computer Vision 19 Motion is a basic cue Even impoverished motion data can elicit a strong percept
21
Computer Vision 20 Applications tracking structure from motion motion segmentation stabilization compression Mosaicing …
22
Computer Vision 21 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
23
Computer Vision 22 Definition of optical flow OPTICAL FLOW = apparent motion of brightness patterns Ideally, the optical flow is the projection of the three-dimensional velocity vectors on the image
24
Computer Vision 23 Caution required ! Two examples : 1. Uniform, rotating sphere O.F. = 0 2. No motion, but changing lighting O.F. 0
25
Computer Vision 24 Caution required !
26
Computer Vision 25 Mathematical formulation I (x,y,t) = brightness at (x,y) at time t Optical flow constraint equation : Brightness constancy assumption:
27
Computer Vision 26 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
28
Computer Vision 27 The aperture problem 1 equation in 2 unknowns
29
Computer Vision 28 The aperture problem 0
30
Computer Vision 29
31
Computer Vision 30 The aperture problem
32
Computer Vision 31 Remarks
33
Computer Vision 32 Apparently an aperture problem
34
Computer Vision 33 What is Optic Flow, anyway? Estimate of observed projected motion field Not always well defined! Compare: –Motion Field (or Scene Flow) projection of 3-D motion field –Normal Flow observed tangent motion –Optic Flow apparent motion of the brightness pattern (hopefully equal to motion field) Consider Barber pole illusion
35
Computer Vision 34
36
Computer Vision 35 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
37
Computer Vision 36 Horn & Schunck algorithm Additional smoothness constraint : besides OF constraint equation term minimize e s + e c
38
Computer Vision 37 Horn & Schunck The Euler-Lagrange equations : In our case, so the Euler-Lagrange equations are is the Laplacian operator
39
Computer Vision 38 Horn & Schunck Remarks : 1. Coupled PDEs solved using iterative methods and finite differences 2. More than two frames allow a better estimation of I t 3. Information spreads from corner-type patterns
40
Computer Vision 39
41
Computer Vision 40 Horn & Schunck, remarks 1. Errors at boundaries 2. Example of regularisation (selection principle for the solution of illposed problems)
42
Computer Vision 41 Results of an enhanced system
43
Computer Vision 42 Structure from motion with OF
44
Computer Vision 43 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
45
Computer Vision 44
46
Computer Vision 45
47
Computer Vision 46
48
Computer Vision 47
49
Computer Vision 48
50
Computer Vision 49 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
51
Computer Vision 50
52
Computer Vision 51
53
Computer Vision 52
54
Computer Vision 53 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
55
Computer Vision 54
56
Computer Vision 55
57
Computer Vision 56
58
Computer Vision 57
59
Computer Vision 58
60
Computer Vision 59
61
Computer Vision 60
62
Computer Vision 61
63
Computer Vision 62
64
Computer Vision 63
65
Computer Vision 64
66
Computer Vision 65
67
Computer Vision 66
68
Computer Vision 67
69
Computer Vision 68
70
Computer Vision 69
71
Computer Vision 70
72
Computer Vision 71
73
Computer Vision 72
74
Computer Vision 73
75
Computer Vision 74
76
Computer Vision 75
77
Computer Vision 76
78
Computer Vision 77
79
Computer Vision 78
80
Computer Vision 79
81
Computer Vision 80
82
Computer Vision 81
83
Computer Vision 82
84
Computer Vision 83 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
85
Computer Vision 84
86
Computer Vision 85
87
Computer Vision 86
88
Computer Vision 87 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
89
Computer Vision 88
90
Computer Vision 89
91
Computer Vision 90
92
Computer Vision 91
93
Computer Vision 92
94
Computer Vision 93 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
95
Computer Vision 94
96
Computer Vision 95
97
Computer Vision 96
98
Computer Vision 97
99
Computer Vision 98
100
Computer Vision 99
101
Computer Vision 100 Optical Flow Brightness Constancy The Aperture problem Regularization Lucas-Kanade Coarse-to-fine Parametric motion models Direct depth SSD tracking Robust flow Bayesian flow
102
Computer Vision 101
103
Computer Vision 102 Rhombus Displays http://www.cs.huji.ac.il/~yweiss/Rhombus/
104
Computer Vision 103
105
Computer Vision 104
106
Computer Vision 105
107
Computer Vision 106
108
Computer Vision 107
109
Computer Vision 108
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.