3d computer vision

81
3D computer vision techniques v. 4b2 1 3D computer vision techniques KH Wong

Upload: vuongnhu

Post on 02-Jan-2017

242 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: 3D computer vision

3D computer vision techniques v.4b2 1

3D computer vision techniques

KH Wong

Page 2: 3D computer vision

3D computer vision techniques v.4b2 2

Seminar Title: 3D computer vision techniques. Abstract

In this talk, the ideas of obtaining 3D information of objects (or called 3D reconstruction) using different techniques are discussed. Currently, the most popular one is the image based method that uses 2D cameras for 3D reconstruction; in particular reconstruction based on one-image, two-image and multiple-image are discussed. Moreover, batch and sequential treatments of input data are studied. I will also talk about novel techniques, such as using multiple cameras and laser based methods to obtain 3D information. And I will discuss how 3D computer vision is used in film and game production. Finally naked-eye 3D display technologies will be mentioned.

Page 3: 3D computer vision

3D computer vision techniques v.4b2 3

Overview (part1) Introduction From 2D to 3D

Camera systems/calibration Feature extraction/correspondence Reconstruction algorithms

2 views, 3 views , N views Real-time algorithms/Kalman filter

Previous projects Virtual viewer/ Projector camera systems Keystone correction

Novel setups Multiple cameras/ Camera array

Obtain 3D directly Structured light Laser approach Kinect approach Photometric stereo

Page 4: 3D computer vision

3D computer vision techniques v.4b2 4

Overview (part 2) Applications

Photos from tourists (photo tourism) http://phototour.cs.washington.edu/

3D displays Possible future research

Classification based on 3D information Content search 3D based on 3D keys Merging with sound information

Page 5: 3D computer vision

3D computer vision techniques v.4b2 5

Motivation We live in a 3D world We see 2D images but perceive the world in

3D Intelligent robot should have this 3D

reconstruction capability

Page 6: 3D computer vision

3D computer vision techniques v.4b2 6

How to obtain 3D information? Cameras-2D Range sensors-3D

Page 7: 3D computer vision

3D computer vision techniques v.4b2 7

Challenges Obtain 3D information for tasks in a 3D world.

2D-to-3D reconstruction from a camera 3D directly— laser range sensor, kinect sensor

Novel sensors Camera array/ multiple camera One pixel camera light field camera

Page 8: 3D computer vision

3D computer vision techniques v.4b2 8

2D-to-3D reconstruction (feature based method) Camera (perspective projection) Features-extraction and correspondences Methods

One-image method Two-image (Stereo) method Three-image method N-image method

Bundle adjustment Kalman filter

Page 9: 3D computer vision

3D computer vision techniques v.4b2 9

Camera: 3D to 2D projection

Perspective model u=F*X/Z (nonlinear relation) v=F*Y/Z

FZ

Y

v

World center

F

Thin lensor a pin hole

Virtual Screenor CCD sensor

RealScreenOr CCD sensor

Pinhole Camera

http://upload.wikimedia.org/wikipedia/en/8/81/Pinhole-camera.png

Page 10: 3D computer vision

3D computer vision techniques v.4b2 10

Perspective Projective

Model M at t=1

c (Image center, ox,oy)

F=focal length

image

Oc=(0,0,0) (Camera center)

Xc-axis

Zc-axis

Yc-axis

v-axis

u-axis

X,Y,Z

(u,v)

(0,0) of image plane

Camera Coordinates.

WorldCoordinates

YwZw

Xw Rc,Tc

Principal axis

Page 11: 3D computer vision

3D computer vision techniques v.4b2 11

In paintings Western Fresco by Raphael, 1510 -

1511, Stanza della Signatura, Vatican Palace, Rome.

Chinese 《富春山居圖》是元朝畫家黃公望的作品,創作於

1347年至 1350年 Dwelling in the

Fuchun Mountains (富春山居圖 ) by Huang Gongwang (1269–1354)

http://www.es.flinders.edu.au/~mattom/science+society/lectures/illustrations/lecture17/schoolathens.htmlhttp://jsl641124.blog.163.com/blog/static/17702514320115219508530/

Page 12: 3D computer vision

3D computer vision techniques v.4b2 12

Feature correspondences--Camera moved, find correspondences for neighboring images--We can use feature to identify the motions of projected 3D features in 2D.

Image at t=t0

(or left image)Image at t=t0+dt (or right image)

Area a

Page 13: 3D computer vision

3D computer vision techniques v.4b2 13

Demo Youtube Movie http://www.youtube.com/watch?v

=azl-DGK6e1U

Page 14: 3D computer vision

3D computer vision techniques v.4b2 14

One-image 2D-to-3D reconstruction

Page 15: 3D computer vision

3D computer vision techniques v.4b2 15

One image 2D-to-3D reconstruction method Difficult and with ambiguity

http://ai.stanford.edu/~asaxena/reconstruction3d/

Page 16: 3D computer vision

3D computer vision techniques v.4b2 16

One image 2D-to-3D Using prior knowledge (e.g. face)

http://www.wisdom.weizmann.ac.il/~ronen/papers/Hassner Basri - Example Based 3D Reconstruction from Single 2D Images.pdf

Page 17: 3D computer vision

3D computer vision techniques v.4b2 17

Two-image 2D-to-3D reconstruction

Page 18: 3D computer vision

3D computer vision techniques v.4b2 18

Two-image 2D-to-3D reconstruction method: stereo vision

Objectives: Basic idea of stereo vision Stereo reconstruction by epipolar geometry

Stereo camera pair calibration (find Fundamental matrix F)

Construct the 3D (graphic) model from 2 images

Graphic model

Inside a computer

Page 19: 3D computer vision

3D computer vision techniques v.4b2 19

X’l X’r

FocalLengthf

ObjectPx(x,y,z)

z

Left camera center (reference point) Horizontal

Disparity=xL-xR

b (Baseline)

Left CameraPrincipleaxis

Right CameraPrincipleaxis

LeftImage plane

RightImage plane

,

( )

elimate (

'l

'r

' 'l r

x x=z fx - b x =

z ff bx z =x - x )

By similar triangle,w.r.t left camera lens center

By similar triangle,w.r.t right camera lens center

if camera motion is pure translation : Triangular calculation

One major problem is to locate x’l and x’r The correspondence problem

Page 20: 3D computer vision

3D computer vision techniques v.4b2 20

If camera motion is NOT pure translation : Use Epipolar Geometry

X

O2O1

right Frame Plane-2 2

left Frame Plane-1 1

e1e2

Left epipolar lineRight epipolar line

R,T

(x1,y1)

(x2,y2)

Left side is the reference

Focal length=f1

Focal length=f2

Base line=||T||

Plane-3 3

Perpendicular to TX2 or TX1

Right_image_pointT*E*left_image_point=0

Page 21: 3D computer vision

3D computer vision techniques v.4b2 21

Method: 8-point algorithmhttp://www.cs.manchester.ac.uk/ugt/COMP37111/papers/Hartley.pdf

Find 8 point corresponded ( Map 8 Right_image_points to left_image_point

Solve the epeiolar formula Right_image_pointT*E*left_image_point=0 Find E. From E we can find camera R (rotation) ,T

(translation) From R,T we can find model (3D positions of the

left feature points (using left as reference)

Page 22: 3D computer vision

3D computer vision techniques v.4b2 22

An example of stereo reconstruction An example Short-Baseline

Stereo Systems for Mobile Devices

http://www.lelaps.de/videos.html#SQx5vU8BA-Mhttp://www.lelaps.de/projects/stmobile.html

Page 23: 3D computer vision

3D computer vision techniques v.4b2 23

Stereo-based Free-space Estimation Another example

http://www.lelaps.de/videos.html#VrKBNtAN03ohttp://www.lelaps.de/projects/freespace.html

Page 24: 3D computer vision

3D computer vision techniques v.4b2 24

Three-image 2D-to-3D reconstruction

Page 25: 3D computer vision

3D computer vision techniques v.4b2 25

Three-image 2D-to-3D reconstruction method More robust using 3 views It contains 3 epipolar

relations Stereo1: view1,2 , Stereo2: view2,3, Stereo3 :view 3,1.

Combine 3 epipolar geometry information together.

Similar to the algorithm in epipolar geometry (apply 3 times)

http://www.cs.unc.edu/~marc/tutorial/node45.html

M=3-D model pointM, m’, m” are image pointsC,C’,C” are camera centers

Page 26: 3D computer vision

3D computer vision techniques v.4b2 26

Example of 3-image reconstruction Example

LIBVISO: Feature Matching for Visual Odometry http://www.youtube.com/watch?v=DPLh6MoxPAk

Page 27: 3D computer vision

3D computer vision techniques v.4b2 27

N-image

2D-to-3D reconstruction(batched method: order of

images can be random )

Page 28: 3D computer vision

3D computer vision techniques v.4b2 28

N-image 2D-to-3D reconstruction method Bundle adjustment approach

Guess iteratively the solution for 3D to explain the measurements of feature points in all images

Math: Q(u,v)=g(X), g is nonlinear (projection) because u=fX/Z v=fY/Z, f=focal length Given Q (image measurement) , we want to find X=(X,Y,Z)i

from image points (u,v)i of all N model points (i=1,,,N), g is the projection formulas

A typical non linear optimization problem, Gauss-Newton for non linear optimization method is used.

Page 29: 3D computer vision

3D computer vision techniques v.4b2 29

Batched method: order of images can be random From measurement [u,v]I find X

Camera motion

Imaget=1

Imaget=2 Image

t=3Imaget=m

v1

v2v3

vm

X

[u,v]2

[u,v]1

[u,v]3

[u,v]m O1

O2

O3

Om

R2,T2R3,T3

Rm,Tm

R1,T1

Page 30: 3D computer vision

3D computer vision techniques v.4b2 30

Example Bundle adjustment reconstruction

http://www.cse.cuhk.edu.hk/%7Ekhwong/demo/canyon2b2.mpg

Page 31: 3D computer vision

3D computer vision techniques v.4b2 31

N-image

2D-to-3D reconstruction(Sequential method: order of images are used like in a move )

Page 32: 3D computer vision

3D computer vision techniques v.4b2 32

Sequential method: order of images are used like in a move From measurement [u,v]I find X

Camera motion

Imaget=1

Imaget=2 Image

t=3Imaget=m

v1

v2v3

vm

X

[u,v]2

[u,v]1

[u,v]3

[u,v]m O1

O2

O3

Om

R2,T2R3,T3

Rm,Tm

R1,T1

Page 33: 3D computer vision

3D computer vision techniques v.4b2 33

Kalman Filter

33pictures by Ko Hoi Fung

Page 34: 3D computer vision

3D computer vision techniques v.4b2 34

Kalman filter example

34

t = 0:t = 0:Position = x0Position = x0Velocity = v0Velocity = v0

t’ = 1:t’ = 1:Position = x1’Position = x1’Velocity = v1’Velocity = v1’x1’ = v0 * t + x0x1’ = v0 * t + x0

t = 1:t = 1:Position = x1Position = x1

PredictionPrediction

UpdateUpdate

Page 35: 3D computer vision

3D computer vision techniques v.4b2 35

Example Hernan Badino and Takeo

Kanade:"A Head-Wearable Short-Baseline Stereo System for the Simultaneous Estimation of Structure and Motion".IAPR Conference on Machine Vision Applications (MVA), Nara, Japan, June

2011 http://www.youtube.com/watch?v=SQx5vU8BA-M

Page 36: 3D computer vision

3D computer vision techniques v.4b2 36

Novel sensors : Camera array/ Multiple camera systems Camera array/ multiple camera: High

Performance Imaging - Using Large Camera Array

http://www.youtube.com/watch?v=0W_1Ce2lTBohttp://graphics.stanford.edu/papers/CameraArray/

Page 37: 3D computer vision

3D computer vision techniques v.4b2 37

The Self-Reconfigurable Camera Array

http://chenlab.ece.cornell.edu/projects/MobileCamArray/

Demo moviehttp://chenlab.ece.cornell.edu/projects/MobileCamArray/videos/train.movhttp://chenlab.ece.cornell.edu/projects/MobileCamArray/videos/self_reconfiguration.mov

Each camera

Page 38: 3D computer vision

3D computer vision techniques v.4b2 38

Applications

Page 39: 3D computer vision

3D computer vision techniques v.4b2 39

Photo tourism

http://phototour.cs.washington.edu/

Page 40: 3D computer vision

3D computer vision techniques v.4b2 40

Projector-camera system

Application of computer vision

Page 41: 3D computer vision

3D computer vision techniques v.4b2 41

A Projector-Camera system

Page 42: 3D computer vision

3D computer vision techniques v.4b2 42

Projector-Camera calibration

http://www.youtube.com/watch?v=YHhQSglmuqY&feature=channel_page

Page 43: 3D computer vision

3D computer vision techniques v.4b2 43

Our setup

Page 44: 3D computer vision

3D computer vision techniques v.4b2 44

Calibration procedure

Page 45: 3D computer vision

3D computer vision techniques v.4b2 45

Quadrangle tracking

Page 46: 3D computer vision

3D computer vision techniques v.4b2 46

Experiments

Page 47: 3D computer vision

3D computer vision techniques v.4b2 47

Projection result

Page 48: 3D computer vision

3D computer vision techniques v.4b2 48

Results

Page 49: 3D computer vision

3D computer vision techniques v.4b2 49

Hand held direct manipulation 3D Display

http://www.youtube.com/watch?v=vVW9QXuKfoQ&feature=relmfu

Page 50: 3D computer vision

3D computer vision techniques v.4b2 50

Keystone correction

Configuration

Page 51: 3D computer vision

3D computer vision techniques v.4b2 51

Aim of this work Desired Results

Keystoned projection Corrected projection

Page 52: 3D computer vision

3D computer vision techniques v.4b2 52

Overview Three major modules

Projector-camera pair calibration Projection region detection and tracking Automatic keystone correction

3D projectionregion recovery

Projector-camera pair calibration

G, K

Corrected projection

Keystoned projection

Pre-warped image3D rectangle

Keystone correctionProjection region tracking

Camera frame

Flow chart

Page 53: 3D computer vision

3D computer vision techniques v.4b2 53

Pre-warp projection image

Pre-warped projection image Display result

http://www.youtube.com/watch?v=y5XYdeh8Bno&list=UUfy2EumiHMeoUorMFR0woZA&index=1&feature=plcp

Page 54: 3D computer vision

3D computer vision techniques v.4b2 54

Keystone correction Some real correction results

Page 55: 3D computer vision

3D computer vision techniques v.4b2 55

Obtain 3D directly Laser range sensor

Time of flight Kinect

Page 56: 3D computer vision

3D computer vision techniques v.4b2 56

Photometric stereo

http://www.taurusstudio.net/research/photex/ps/equation.htm

Lamertian light formula

•Given 3 or more known light source we can find the normal N•From the set of N we can approximate the surface

http://www.wisdom.weizmann.ac.il/~vision/photostereo/

Page 57: 3D computer vision

3D computer vision techniques v.4b2 57

Photometric stereo using multiple cameras and multiple light sources

Demo

Dynamic Shape Capture using Multi-View Photometric Stereo SIGGRAPH 2009 http://www.youtube.com/watch?v=9hgs5zN38lk

Page 58: 3D computer vision

3D computer vision techniques v.4b2 58

Multiple cameras fro human body reconstruction

Homepage://media.au.tsinghua.edu.cn

Page 59: 3D computer vision

3D computer vision techniques v.4b2 59

Experimental Results

59 3D Modeling Using MVML Dome 23/5/3

Page 60: 3D computer vision

3D computer vision techniques v.4b2 60

Multiple camera doom

http://www.mpi-inf.mpg.de/~yliu/

Page 61: 3D computer vision

3D computer vision techniques v.4b2 61

Structured light method Calculate the shape by how the strip is

distorted.

http://www.laserfocusworld.com/articles/2011/01/lasers-bring-gesture-recognition-to-the-home.html

Page 62: 3D computer vision

3D computer vision techniques v.4b2 62

Real time Virtual 3D Scanner - Structured Light Technology Demo

http://www.youtube.com/watch?v=a6pgzNUjh_s

Page 63: 3D computer vision

3D computer vision techniques v.4b2 63

Time of flight laser method Send the IR-laser light to

different directions and sense how each beam is delayed.

Use the delay to calculate the distance of the object point

http://www.laserfocusworld.com/articles/2011/01/lasers-bring-gesture-recognition-to-the-home.html

Page 64: 3D computer vision

3D computer vision techniques v.4b2 64

LIDAR light detection and ranging scanner

http://hodcivil.edublogs.org/2011/11/06/lidar-%E2%80%93-light-detection-and-ranging/http://commons.wikimedia.org/wiki/File:Lidar_P1270901.jpg

Leica terrestrial lidar (light detection and ranging) scanner

http://www.youtube.com/watch?v=MuwQTc8KK44

Page 65: 3D computer vision

3D computer vision techniques v.4b2 65

3D Laser Scanning - Underground Mine Mapping Demo

http://www.youtube.com/watch?v=BZbvz8fePeQ

Page 66: 3D computer vision

3D computer vision techniques v.4b2 66

Motion capture for film production (MOCAP)

http://upload.wikimedia.org/wikipedia/commons/7/73/MotionCapture.jpghttp://www.naturalpoint.com/optitrack/products/s250e/indepth.html

IR light emitterand camera

http://www.youtube.com/watch?v=IxJrhnynlN8

Page 67: 3D computer vision

3D computer vision techniques v.4b2 67

3D body scanner

http://www.cyberware.com/products/scanners/ps.htmlhttp://www.cyberware.com/products/scanners/wbx.html

http://www.youtube.com/watch?v=86hN0x9RycM

Page 68: 3D computer vision

3D computer vision techniques v.4b2 68

3-D Face capture

http://www.captivemotion.com/products/

http://www.youtube.com/watch?v=-TTR0JrocsI&feature=related

Page 69: 3D computer vision

3D computer vision techniques v.4b2 69

Dimensional Imaging 4D Video Face Capture with Textures

Dimensional Imaging 4D Video Face Capture with Textures

http://www.youtube.com/watch?v=XtTN7tWaXTM&feature=related

Page 70: 3D computer vision

3D computer vision techniques v.4b2 70

Kinect Another structure light

method Use dost rather than

strips

http://www.laserfocusworld.com/articles/2011/01/lasers-bring-gesture-recognition-to-the-home.html

Page 71: 3D computer vision

3D computer vision techniques v.4b2 71

Kinect Hardware

Page 72: 3D computer vision

3D computer vision techniques v.4b2 72

See the IR-dots emitted by KINECT

http://www.youtube.com/watch?v=-gbzXjdHfJA

http://www.youtube.com/watch?v=dTKlNGSH9Po&feature=related

Page 73: 3D computer vision

3D computer vision techniques v.4b2 73

Novel sensors : light field cameraSpin off from Stanford camera array

light field camera : LYTRO camera

Be able to refocus after the picture is taken

https://www.lytro.com/camera

http://www.youtube.com/watch?v=7QV152jc3Ac

Page 74: 3D computer vision

3D computer vision techniques v.4b2 74

light field camera How does it work

http://www.quora.com/Lytro/How-does-the-new-Lytro-camera-work

Page 75: 3D computer vision

3D computer vision techniques v.4b2 75

3D (Volumetric) displayRendering for an Interactive 360º Light Field Display

SIGGRAPH 2007 Papers Proceedings

http://gl.ict.usc.edu/Research/3DDisplay/

http://www.youtube.com/watch?v=h6aUIS44ezE

Page 76: 3D computer vision

3D computer vision techniques v.4b2 76

Occlusion-Capable Volumetric 3D Display by Cossairt,etal.

Actuality Systems, Inc

http://www.3dcgi.com/cooltech/displays/displays.htm

http://www.youtube.com/watch?v=8KaQmn2VTzs

Page 77: 3D computer vision

3D computer vision techniques v.4b2 77

3D display Using a lattice with thin slits, viewer's eyes see different pixels on the screen to create 3d perception

http://www.televisions.com/tv-articles/TV-in-3D/Displaying-3D-Without-Glasses.php

Page 78: 3D computer vision

3D computer vision techniques v.4b2 78

The future Content search in 3D video data bases Shot boundary detection Video data mining Video classification

Page 79: 3D computer vision

3D computer vision techniques v.4b2 79

Appendix

Page 80: 3D computer vision

3D computer vision techniques v.4b2 80

Essential matrix E (a 3x3 matrix) P.110[2] X1 is 3-D X in left camera (reference) system X2 is 3-D X in right camera system

RTEE

RT(i)

TT

TT TTTTT

TTTTiRTT

TTR

T

T

TT

BABxA

where0** hence

0)( , from

0)(X as same ,0)(X so

)( lar toperpendicu is plane, same theon are and0 andmatrix symmetricskew

and ][ Since )(

by sides bothmultiply ,

12

12

2222

222

12

12

BABxA

x

x

TTTT

tttt

tt

ttt

T

][

00

0][

12

13

23

3

2

1

Exercise1:Draw vectors TX2 or TX1 in the diagram

Page 81: 3D computer vision

3D computer vision techniques v.4b2 81

Essential Matrix E

Right_image_pointT*E*left_image_point=0

2 1

2 2 2 1 1 1

2 1

2 1

2 12 2 2 1 1 1

2 1

2 2 1 1 12

X * * X 0

* * 0

_ _ _ _ _ tan _ ,

* * 0

* * 0

since X, ,

T

T

T

T

T

E

X Y Z E X Y Z

f fno harm to prefix some cons t termsZ Z

f fX Y Z E X Y ZZ Z

x y f E x y f

X YX Y Z and x f y fZ Z