|
00:00:00
|
Lecture 20: Motion and Optical Flow |
|
00:01:10
|
Course Outline |
|
00:01:26
|
Recap: Epipolar Geometry - Calibrated Case |
|
00:02:09
|
Recap: Epipolar Geometry - Uncalibrated Case |
|
00:03:25
|
Recap: The Eight-Point Algorithm |
|
00:05:27
|
Recap: Normalized Eight-Point Algorithm |
|
00:07:30
|
Practical Considerations |
|
00:15:26
|
Topics of This Lecture |
|
00:16:37
|
Video |
|
00:17:22
|
Motion and Perceptual Organization |
|
00:20:41
|
Uses of Motion |
|
00:21:26
|
Motion Estimation Techniques |
|
00:22:46
|
Topics of This Lecture |
|
00:22:56
|
Motion Field |
|
00:24:02
|
Motion Field and Parallax |
|
00:35:01
|
Topics of This Lecture |
|
00:35:36
|
Optical Flow |
|
00:37:01
|
Apparent Motion ≠ Motion Field |
|
00:37:44
|
Estimating Optical Flow |
|
00:39:58
|
The Brightness Constancy Constraint |
|
00:47:40
|
The Aperture Problem |
|
00:48:43
|
The Barber Pole Illusion |
|
00:49:43
|
Solving the Aperture Problem |
|
00:53:41
|
Conditions for Solvability |
|
00:54:52
|
Eigenvectors of AᵀA |
|
00:56:28
|
Interpreting the Eigenvalues |
|
00:57:12
|
Edge |
|
00:58:15
|
Low-Texture Region |
|
00:58:43
|
High-Texture Region |
|
00:59:32
|
Per-Pixel Estimation Procedure |
|
01:00:47
|
Iterative Refinement |
|
01:03:08
|
Optical Flow: Iterative Refinement |
|
01:06:34
|
Some Implementation Issues |
|
01:09:55
|
Extension: Global Parametric Motion Models |
|
01:12:02
|
Example: Affine Motion |
|
01:14:27
|
Problem Cases in Lucas-Kanade |
|
01:16:29
|
Dealing with Large Movements |
|
01:17:05
|
Temporal Aliasing |
|
01:18:52
|
Idea: Reduce the Resolution! |
|
01:19:55
|
Coarse-to-fine Optical Flow Estimation |
|
01:21:30
|
Dense Optical Flow |
|
01:23:10
|
Summary |
|
01:25:00
|
References and Further Reading |