0.25
0.5
0.75
1.25
1.5
1.75
2
Compressive Sensing for Computer Vision: Hype vs Hope
Published on Dec 01, 200916280 Views
Related categories
Chapter list
Compressive Sensing: Opportunities and pitfalls for Computer Vision00:00
Outline05:36
Sparsity: Motivation08:20
Compressive sensing08:54
Compressive sampling10:20
How does it work?11:56
Restricted Isometry Property (RIP)12:24
Insight from the 80’s14:35
CS signal recovery15:41
CS recovery algorithms17:50
BPDN18:46
Recent developments in computer vision/graphics19:48
Gradient domain processing: vision and graphics24:50
Few applications25:22
Gradient fields and integrability25:54
Non-integrable gradient fields26:46
Discrete domain26:58
Interpretation27:54
CS and reconstruction from gradient fields28:25
l1 - minimization (1)29:24
l1 - minimization (2)29:42
l1 - minimization (3)30:02
Face recognition via sparse representations30:06
Formulation (1)32:37
Formulation (2)33:38
Handling registration and illumination34:37
Iris recognition - 136:44
Iris recognition - 237:28
Basic formulation (1)37:55
Basic formulation (2)38:09
Sparsity for image selection38:25
Selection and recognition algorithm38:41
Results38:49
Sectored random projections For cancelable iris biometrics39:25
Random projections for cancelability40:07
Random projections40:08
Results – Recognition performance42:53
Model-based compressive sensing for reflectance fields42:57
Capturing surface properties43:05
Reflectance field of a scene43:25
Compressive acquisition43:42
Reflectance field acquisition44:45
Enforcing spatial coherency45:17
Compressive sensing set up (Peers et al 2009)45:41
Relighting results46:13
Real Signals have more structure than just Sparsity46:29
Images …46:38
General signal models46:41
Effectiveness of the hybrid subspace space model47:33
Solving under the HSS model47:50
Advantages of the HSS Model47:59
Compressive acquisition48:09
Compressive measurements48:23
Scene relighting Results48:31
Other works not discussed48:51
Remarks49:47