Download presentation
Presentation is loading. Please wait.
1
Stanford CS223B Computer Vision, Winter 2007 Lecture 8 Structure From Motion Professors Sebastian Thrun and Jana Košecká CAs: Vaibhav Vaish and David Stavens Slide credit: Gary Bradski, Stanford SAIL
2
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Summary SFM n Problem –Determine feature locations (=structure) –Determine camera extrinsic (=motion) n Two Principal Solutions –Bundle adjustment (nonlinear least squares, local minima) –SVD (through orthographic approximation, affine geometry) n Correspondence –(RANSAC) –Expectation Maximization
3
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion camera features Recover: structure (feature locations), motion (camera extrinsics)
4
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 SFM = Holy Grail of 3D Reconstruction n Take movie of object n Reconstruct 3D model n Would be commercially highly viable live.com
5
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion (1) [Tomasi & Kanade 92]
6
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion (2) [Tomasi & Kanade 92]
7
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion (3) [Tomasi & Kanade 92]
8
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion (4a): Images Marc Pollefeys
9
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion (4b) Marc Pollefeys
10
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion (5) http://www.cs.unc.edu/Research/urbanscape
11
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion n Problem 1: –Given n points p ij =(x ij, y ij ) in m images –Reconstruct structure: 3-D locations P j =(x j, y j, z j ) –Reconstruct camera positions (extrinsics) M i =(A j, b j ) n Problem 2: –Establish correspondence: c(p ij )
12
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion camera features Recover: structure (feature locations), motion (camera extrinsics)
13
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Recovery Problems 1 image2+ images Location knowncalibrationstereo Location unknown SFM, stitching
14
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 SFM: General Formulation fZ X O -x
15
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 SFM: Bundle Adjustment fZ X O -x
16
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Bundle Adjustment n SFM = Nonlinear Least Squares problem n Minimize through –Gradient Descent –Conjugate Gradient –Gauss-Newton –Levenberg Marquardt common method n Prone to local minima
17
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Count # Constraints vs #Unknowns n m camera poses n n points n 2mn point constraints n 6m+3n unknowns n Suggests: need 2mn 6m + 3n n But: Can we really recover all parameters???
18
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 How Many Parameters Can’t We Recover? 036781012nmnm Place Your Bet! We can recover all but… m = #camera poses n = # feature points
19
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Count # Constraints vs #Unknowns n m camera poses n n points n 2mn point constraints n 6m+3n unknowns n Suggests: need 2mn 6m + 3n n But: Can we really recover all parameters??? –Can’t recover origin, orientation (6 params) –Can’t recover scale (1 param) n Thus, we need 2mn 6m + 3n - 7
20
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Are we done? n No, bundle adjustment has many local minima.
21
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 The “Trick Of The Day” Replace Perspective by Orthographic Geometry Replace Euclidean Geometry by Affine Geometry Solve SFM linearly via PCA (“closed” form, globally optimal) Post-Process to make solution Euclidean Post-Process to make solution perspective By Tomasi and Kanade, 1992
22
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Orthographic Camera Model Orthographic = Limit of Pinhole Model: Extrinsic Parameters Rotation Orthographic Projection
23
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Orthographic Projection Limit of Pinhole Model: Orthographic Projection
24
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 The Orthographic SFM Problem subject to
25
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 The Affine SFM Problem subject to drop the constraints
26
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Count # Constraints vs #Unknowns n m camera poses n n points n 2mn point constraints n 8m+3n unknowns n Suggests: need 2mn 8m + 3n n But: Can we really recover all parameters???
27
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 How Many Parameters Can’t We Recover? 036781012nmnm Place Your Bet! We can recover all but…
28
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 The Answer is (at least): 12 CAA ii '
29
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Points for Solving Affine SFM Problem n m camera poses n n points n Need to have: 2mn 8m + 3n-12
30
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Affine SFM Fix coordinate system by making p i0 = P 0 =origin Proof: Rank Theorem: Q has rank 3
31
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 The Rank Theorem n elements 2m elements
32
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Singular Value Decomposition
33
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Affine Solution to Orthographic SFM Gives also the optimal affine reconstruction under noise
34
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Back To Orthographic Projection Find C for which constraints are met Search in 9-dim space (instead of 8m + 3n-12)
35
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Back To Projective Geometry Orthographic (in the limit) Projective
36
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Back To Projective Geometry fZ X O -x Optimize Using orthographic solution as starting point
37
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 The “Trick Of The Day” Replace Perspective by Orthographic Geometry Replace Euclidean Geometry by Affine Geometry Solve SFM linearly via PCA (“closed” form, globally optimal) Post-Process to make solution Euclidean Post-Process to make solution perspective By Tomasi and Kanade, 1992
38
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Structure From Motion n Problem 1: –Given n points p ij =(x ij, y ij ) in m images –Reconstruct structure: 3-D locations P j =(x j, y j, z j ) –Reconstruct camera positions (extrinsics) M i =(A j, b j ) n Problem 2: –Establish correspondence: c(p ij )
39
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 The Correspondence Problem View 1View 3View 2
40
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Correspondence: Solution 1 n Track features (e.g., optical flow) n …but fails when images taken from widely different poses
41
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Correspondence: Solution 2 n Start with random solution A, b, P n Compute soft correspondence: p(c|A,b,P) n Plug soft correspondence into SFM n Reiterate See Dellaert/Seitz/Thorpe/Thrun, Machine Learning Journal, 2003
42
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Example
43
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Results: Cube
44
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Animation
45
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Tomasi’s Benchmark Problem
46
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Reconstruction with EM
47
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 3-D Structure
48
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Correspondence: Alternative Approach n Ransac [Fisher/Bolles] = Random sampling and consensus n Will be discussed Wednesday
49
Sebastian Thrun and Jana Košecká CS223B Computer Vision, Winter 2007 Summary SFM n Problem –Determine feature locations (=structure) –Determine camera extrinsic (=motion) n Two Principal Solutions –Bundle adjustment (nonlinear least squares, local minima) –SVD (through orthographic approximation, affine geometry) n Correspondence –(RANSAC) –Expectation Maximization
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.