875: recent a dvances in geometric c omputer v ision & recognition

44
875: Recent Advances in Geometric Computer Vision & Recognition Jan-Michael Frahm Spring 2014

Upload: zarita

Post on 24-Feb-2016

44 views

Category:

Documents


1 download

DESCRIPTION

875: Recent A dvances in Geometric C omputer V ision & Recognition. Jan-Michael Frahm Spring 2014. Introductions. Grade Requirements. Presentation of 2 papers in class 30 min talk, 10 min questions Papers for selection must come from: top journals: IJCV, PAMI, CVIU, IVCJ - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

875: Recent Advances in Geometric Computer Vision & Recognition

Jan-Michael FrahmSpring 2014

Page 2: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

2

Introductions

Page 3: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

3

Grade Requirements• Presentation of 2 papers in class

30 min talk, 10 min questions

• Papers for selection must come from: top journals: IJCV, PAMI, CVIU, IVCJ top conferences: CVPR (2010,2011), ICCV (2011),

ECCV (2010), approval for all other venues is needed

• Final project evaluation, extension of a recent method from

the above

Page 4: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

4

Grading• 20% first presentation• 20% second presentation• 30% final project • 30% attendance & class participation

Page 5: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

5

Schedule• Jan. 7th, Introduction

• Jan 7th, Uncertainty in Stereo (guest Philippos Mordohai) (substitute for Jan 13th class)

• Jan 15th , Large-scale image localization basic concepts, First paper selection (Large –scale localization)

• Jan 20th, MLK holiday no class

• Jan 22nd-29th, Large-scale localization basic concepts

• Feb. 3rd, 1. round of presentations starts

• Mar. 10th, 12th Spring break (no class)

• Mar. 17th, Modeling dynamic objects/scenes basic concepts, Second paper selection, final project definition

• Mar. 19st, Modeling dynamic objects

• Mar. 24th, 2. round of presentations starts

• Apr. 21st, 23rd , final project presentation

Page 6: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

6

How to give a great presentation

• Structure of the talk: Motivation (motivate and explain the

problem) Overview Related work (short concise discussion) Approach Experiments Conclusion and future work

Page 7: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

7

How to give a great presentation

• Use large enough fonts 5-6 one line bullet items on a slide

max• Keep it simple• No complex formulas in your talk• Bad Powerpoint slides• How to for presentations

Page 8: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

8

How to give a great presentation

• Abstract the material of the talk provide understanding beyond

details• Use pictures to illustrate

find pictures on the internet create a graphic (in ppt, graph tool) animate complex pictures

Page 9: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

9

How to give a good presentation

• Avoid bad color schemes no red on blue looks awful

• Avoid using laser pointer (especially if you are nervous)

• Add pointing elements in your presentation

• Practice to stay within your time! • Don’t rush through the talk!

Page 10: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Brush up on Stereo Reconstruction

10

Page 11: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Stereo• Extraction of 3D information from 2D images

11

Images 3D Point Cloud

Stereo

Page 12: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Binocular stereo• Given a calibrated binocular stereo pair, fuse it

to produce a depth image Humans can do it

Stereograms: Invented by Sir Charles Wheatstone, 1838

Page 13: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

13

Depth Recovery by Stereo

reference image matching imageDepth

d1d2

d3d4

d5d6

d7d8

d9

Search Space

Epipolar line

Page 14: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

14

Depth Recovery from Stereo

reference image matching imageDepth

d1d2

d3d4

d5d6

d7d8

d9

Search Space

Epipolar line

depthPixel similarity: measured by color differences

Matching Cost

Ground Truth Pixel Matching

Depth Map

Page 15: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Matching criteria• Raw pixel values (correlation)• Band-pass filtered images [Jones & Malik 92]• “Corner” like features [Zhang, …]• Edges [many people…]• Gradients [Seitz 89; Scharstein 94]• Rank statistics [Zabih & Woodfill 94]• Intervals [Birchfield and Tomasi 96]• Overview of matching metrics and their performance:

H. Hirschmüller and D. Scharstein, “Evaluation of Stereo Matching Costs on Images with Radiometric Differences”, PAMI 2008

slide: R. Szeliski

Page 16: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Adaptive Weighting• Boundary Preserving• More Costly

Page 17: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Simplest Case: Parallel images

• Image planes of cameras are parallel to each other and to the baseline

• Camera centers are at same height

• Focal lengths are the same

slide: S. Lazebnik

Page 18: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Simplest Case: Parallel images

• Image planes of cameras are parallel to each other and to the baseline

• Camera centers are at same height

• Focal lengths are the same

• Then, epipolar lines fall along the horizontal scan lines of the images

slide: S. Lazebnik

Page 19: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Essential matrix for parallel images

RtExExT ][,0

0000

000][

TTRtE

R = I t = (T, 0, 0)

Epipolar constraint:

00

0][

xy

xz

yz

aaaa

aaa

t

x

x’

Page 20: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Essential matrix for parallel images

RtExExT ][,0

0000

000][

TTRtE

Epipolar constraint:

R = I t = (T, 0, 0)

t

x

x’

Page 21: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

21

Aggregation Structure

depth

Matching Cost

Pixelwise Costs

Search Space

Jan-Michael Frahm
Jan-Michael Frahm
pixel wise
Jan-Michael Frahm
Here you should also indicate that the minimum is used to compute the depth of the pixel.
Page 22: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

22

Aggregation Structure

Cost Volume

Cost aggregation: cutting the cost volume.

Search Space

Se arch Space

Page 23: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

23

Aggregation Structure

Cost Volume

Fronto-Parallel Plane

Treat neighbors equally

Cost of the center pixel

Costs of neighboring

pixels

Sum of Absolute Differences (SAD)

Depth Map

Page 24: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

24

Aggregation Structure

Adaptive WeightYoon and Kweon, PAMI 2006

Depth Map

Cost Volume

•Color differences •Spatial distances

Weighted cost of the center

pixel

Weighted costs of neighboring

pixels

Page 25: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

25

Aggregation Structure

Adaptive Weight

Depth Map

Oriented Plane

Cost Volume

Lu et al., CVPR 2013

Page 26: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Your basic stereo algorithm

For each epipolar lineFor each pixel in the left image

• compare with every pixel on same epipolar line in right image• pick pixel with minimum match cost

Improvement: match windows• This should look familar...

slide: R. Szeliski

Page 27: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

)( 4NO

)( 3NO

Depth Map Computation• Local methods

Depth with the minimum cost Complexity:

• Global methods Pairwise interactions Complexity:

Scharstein and Szeliski, “A taxonomy and evaluation of dense two-frame stereo correspondence algorithms", IJCV 2002

Image Resolution : the total number of pixels

28

N pixels

aN

pixe

ls

bN pixels

)( 2NO

2~ N

Page 28: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Depth from disparity

f

x x’

BaselineB

z

O O’

X

f

zfBxxdisparity

Disparity is inversely proportional to depth!

Page 29: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Depth Sampling Depth sampling for integer pixel disparity

Quadratic precision loss with depth!

Page 30: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Depth Sampling Depth sampling for wider baseline

Page 31: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Depth Sampling Depth sampling is in O(resolution6)

Page 32: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Failures of correspondence search

Textureless surfaces Occlusions, repetition

Non-Lambertian surfaces, specularitiesslide: S. Lazebnik

Page 33: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

How can we improve window-based matching?

• The similarity constraint is local (each reference window is matched independently)

• Need to enforce non-local correspondence constraints

slide: S. Lazebnik

Page 34: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Non-local constraints• Uniqueness

For any point in one image, there should be at most one matching point in the other image

slide: S. Lazebnik

Page 35: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Non-local constraints• Uniqueness

For any point in one image, there should be at most one matching point in the other image

• Ordering Corresponding points should be in the same order

in both views

slide: S. Lazebnik

Page 36: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Non-local constraints• Uniqueness

For any point in one image, there should be at most one matching point in the other image

• Ordering Corresponding points should be in the same order in

both views

Ordering constraint doesn’t holdslide: S. Lazebnik

Page 37: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Non-local constraints• Uniqueness

For any point in one image, there should be at most one matching point in the other image

• Ordering Corresponding points should be in the same order in

both views• Smoothness

We expect disparity values to change slowly (for the most part)

slide: S. Lazebnik

Page 38: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

I1 I2 I10

Multiple-baseline stereo results

M. Okutomi and T. Kanade, “A Multiple-Baseline Stereo System,” IEEE Trans. on Pattern Analysis and Machine Intelligence, 15(4):353-363 (1993).

Page 39: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Plane Sweep Stereo• Choose a reference view• Sweep family of planes at different depths with

respect to the reference camera

Each plane defines a homography warping each input image into the reference view

reference camera

input image

R. Collins. A space-sweep approach to true multi-image matching. CVPR 1996.

input image

Page 40: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Real-time 3D reconstruction from video

“Real-Time Plane-sweeping Stereo with Multiple Sweeping Directions", CVPR 2007

3D scene SAD as similarity (darker is higher

similarity)

warped images

46

Page 41: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Real-time 3D reconstruction from video

47

“Real-Time Plane-sweeping Stereo with Multiple Sweeping Directions", CVPR 2007

3D scene

warped images

SAD as similarity (darker is higher

similarity)

Page 42: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

Real-time 3D reconstruction from video

49

“Real-Time Plane-sweeping Stereo with Multiple Sweeping Directions", CVPR 2007

3D scene

warped images

SAD as similarity (darker is higher

similarity)

Multi-way sweep

Page 43: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

3D reconstruction from video

view 1 view N

50

Page 44: 875: Recent  A dvances  in  Geometric  C omputer  V ision  &  Recognition

3D reconstruction from video

51