Computer Vision: Local Features II (Di, 09.12.2014)

Anmeldung erforderlich

RWTH

Für RWTH-Angehörige und aus dem RWTH-Netz verfügbar

Anmelden
  • Einbetten

Beschreibung:

Scale Invariance, Local Descriptors, SIFT

Kapitel:

00:00:00
Lecture 13: Local Features II
00:00:05
Course Outline
00:01:19
Recap: Local Feature Matching Outline
00:03:31
Recap: Requirements for Local Features
00:04:17
Recap: Harris Detector [Harris88]
00:06:40
Recap: Harris Detector Responses [Harris88]
00:09:28
Recap: Hessian Detector [Beaudet78]
00:10:39
Recap: Hessian Detector Responses [Beaudet78]
00:11:26
Topics of This Lecture
00:12:28
From Points to Regions...
00:13:45
Naïve Approach: Exhaustive Search
00:16:04
Automatic Scale Selection
00:22:54
What Is A Useful Signature Function?
00:26:30
Characteristic Scale
00:27:22
Laplacian-of-Gaussian (LoG)
00:40:11
LoG Detector: Workflow
00:44:56
Difference-of-Gaussian (DoG)
00:47:37
Key point localization with DoG
00:48:59
DoG - Efficient Computation
00:57:45
Results: Lowe's DoG
01:00:28
Harris-Laplace [Mikolajczyk '01]
01:03:21
Summary: Scale Invariant Detection
01:04:41
Topics of This Lecture
01:05:01
Rotation Invariant Descriptors
01:06:05
Orientation Normalization: Computation
01:09:14
Topics of This Lecture
01:09:28
The Need for Invariance
01:11:54
Affine Adaptation
01:14:46
Iterative Affine Adaptation
01:15:46
Affine Normalization/Deskewing
01:16:43
Affine Adaptation Example
01:17:22
Summary: Affine-Inv. Feature Extraction
01:17:52
Invariance vs. Covariance
01:20:23
Topics of This Lecture
01:20:38
References and Further Reading