SlideShare a Scribd company logo
1 of 5
Download to read offline
Robert Collins                                                   Robert Collins
CSE486, Penn State                                               CSE486, Penn State
                                                                                                 Generalized Stereo
                             Lecture 18:                                    Key idea: Any two images showing an overlapping
                        Generalized Stereo:                                 view of the world can be treated as a stereo pair...

                        Epipolar Geometry
                                                                            ... we just have to figure out how the
                                                                                    two views are related.


                                                                       Some of the most “beautiful” math in vision concerns
                                                                       describing how multiple views are related, geometrically.




Robert Collins                                                   Robert Collins
CSE486, Penn State
                     Recall: Epipolar Constraint                 CSE486, Penn State
                                                                                      Review : Simple Stereo System
                                                                                                        P
       Important Stereo Vision Concept:
                                                                                       z                           Left camera
                                                                   y
       Given a point in the left image, we don’t have to
                                                                                                             z
       search the whole right image for a corresponding point.
                                                                                       y                           Right camera
                                                                              x
       The “epipolar constraint” reduces the search space
       to a one-dimensional line.                                           Tx
                                                                                                   Epipolar line
                                                                                                  x
                                                                                                                                 Same Y Coord!
                                                                                            baseline
                                                                            depth
                                                                                                         Equation relating
                                                                                                         depth and disparity
                                                                                            disparity




Robert Collins                                                   Robert Collins
CSE486, Penn State
                     Review: Epipolar Constraint                 CSE486, Penn State
                                                                                                      General Stereo

                                                                                                                                    C2        y2
                                                                       x1      y1                       C1
                                                                                           z1                                            z2
                                                                                                                                              O2
                                                                             O1
                                                                                                                                         x2
                                                                             In general, the cameras may be related by
                                                                             an arbitrary transformation (R,T)
                                                                                                Epipolar Matrix
              Corresponding features are constrained to
              lie along conjugate epipolar lines (on the                      In general, intrinsic camera parameters
                                                                              may be different, and even unknown
              same row in the case of our simple setup).
                                                                                        Fundamental Matrix




                                                                                                                                                   1
Robert Collins                                            Robert Collins
CSE486, Penn State                                                        Epipolar Geometry
                                                          CSE486, Penn State

                                                                                                                  P


                                                                                                        pl              pr
                                                                                                             el       er     Or
                                                                          Ol
                     EPIPOLAR                               Epipoles:
                                                            Epipoles:

                     GEOMETRY                               • el: left image of Or
                                                            • er: right image of Ol
                                                            Epipolar plane:
                                                            • Three points: Ol,Or, and P define an epipolar plane
                                                            Epipolar lines and epipolar constraint:
                                                            • Intersections of epipolar plane with the image planes
                                                            • Corresponding points are on “conjugate” epipolar lines
                                                                                            conjugate”

                                                          The following slides are from Dr.Camps, PSU




Robert Collins                                            Robert Collins
CSE486, Penn State                                        CSE486, Penn State




                     BORING!!!                                                       EPIPOLAR
                        Let’s try again...
                                                                                     GEOMETRY




Robert Collins
CSE486, Penn State    Epipolar Geometry                   Robert Collins
                                                          CSE486, Penn State
                                                                                       Rays to Points in Scene
                          A Visualization
                                                              Tie threads on
                                                              to the pins
                                Would would Pinhead’s
                                                              and connect
                                eye look like close up?
                                                              focal point
                                                              to scene
                                                              points




                                                                    Now what would this look like to a second observer?




                                                                                                                                  2
Robert Collins                                                                                                                                                                          Robert Collins

                Rays Seen from Second Observer
CSE486, Penn State                                                                                                                                                                      CSE486, Penn State
                                                                                                                                                                                                                                                                                                                            Rays Seen by the First Viewer




                                                                                                                                    epipolar lines                                                                                                                                                                                                       epipolar lines



                        Q       u   i
                                    c   k   T   i
                                                m       e   ™           a   n   d       a
                    T   I   F   F       (   L   Z   W       )       d   e   c   o   m       p   r   e   s   s   o   r
                a   r   e       n   e   e   d   e   d       t   o       s   e   e       t   h   i
                                                                                                s       p   i
                                                                                                            c   t   u   r   e   .




      epipole                                                                                                                                                                                                                                                                                                                                                         epipole


                                                                                                                                                 Image 2                                                                                                                                                                                      Image 1




Robert Collins                                                                                                                                                                          Robert Collins
CSE486, Penn State
                                                                                                                                          Epipolar Geometry                             CSE486, Penn State
                                                                                                                                                                                                                                                                                                                                   Epipolar Geometry
                                                            image1                                                                                                 image 2                                                                          image1                                                                                                  image 2




       Epipole : location of cam2                                                                                                                          Epipole : location of cam1                                                                                                                                              Corresponding points
       as seen by cam1.                                                                                                                                    as seen by cam2.                                                                                                                                                     lie on conjugate epipolar lines




Robert Collins                                                                                                                                                                          Robert Collins
CSE486, Penn State
                                                                                                                                          Epipolar Geometry                             CSE486, Penn State
                                                                                                                                                                                                                                                                                                                            Epipole not Necessarily in Image
                                                            image1                                                                                                 image 2



                                                                                                                                                                                                                                                                                                                             epipolar lines



                                                                                                                                    Conjugate epipolar lines induce                                     a
                                                                                                                                                                                                            T
                                                                                                                                                                                                            r
                                                                                                                                                                                                                Q
                                                                                                                                                                                                                I
                                                                                                                                                                                                                e
                                                                                                                                                                                                                    F
                                                                                                                                                                                                                        u
                                                                                                                                                                                                                        F
                                                                                                                                                                                                                        n
                                                                                                                                                                                                                            i
                                                                                                                                                                                                                            c


                                                                                                                                                                                                                            e
                                                                                                                                                                                                                                k
                                                                                                                                                                                                                                (
                                                                                                                                                                                                                                e
                                                                                                                                                                                                                                    T
                                                                                                                                                                                                                                    L
                                                                                                                                                                                                                                    d
                                                                                                                                                                                                                                        i
                                                                                                                                                                                                                                        m
                                                                                                                                                                                                                                        Z
                                                                                                                                                                                                                                        e
                                                                                                                                                                                                                                            W
                                                                                                                                                                                                                                            d
                                                                                                                                                                                                                                                e   ™
                                                                                                                                                                                                                                                    )
                                                                                                                                                                                                                                                    t   o
                                                                                                                                                                                                                                                            d
                                                                                                                                                                                                                                                                a
                                                                                                                                                                                                                                                                e
                                                                                                                                                                                                                                                                s
                                                                                                                                                                                                                                                                    n
                                                                                                                                                                                                                                                                    c
                                                                                                                                                                                                                                                                    e
                                                                                                                                                                                                                                                                        d
                                                                                                                                                                                                                                                                        o
                                                                                                                                                                                                                                                                        e
                                                                                                                                                                                                                                                                            m
                                                                                                                                                                                                                                                                                a


                                                                                                                                                                                                                                                                                t
                                                                                                                                                                                                                                                                                    p
                                                                                                                                                                                                                                                                                    h
                                                                                                                                                                                                                                                                                        r
                                                                                                                                                                                                                                                                                        i
                                                                                                                                                                                                                                                                                        s
                                                                                                                                                                                                                                                                                            e   s
                                                                                                                                                                                                                                                                                                p
                                                                                                                                                                                                                                                                                                    s
                                                                                                                                                                                                                                                                                                    i
                                                                                                                                                                                                                                                                                                    c
                                                                                                                                                                                                                                                                                                        o
                                                                                                                                                                                                                                                                                                        t
                                                                                                                                                                                                                                                                                                            r
                                                                                                                                                                                                                                                                                                            u   r   e   .




                                                                                                                                    a generalized 1D “scan-line” ordering                     epipole
                                                                                                                                    on the images (analogous to traditional
                                                                                                                                    scan line ordering of rows in an image)
                                                                                                                                                                                                                                                                                                                                                        Image 2




                                                                                                                                                                                                                                                                                                                                                                                3
Robert Collins                                                                            Robert Collins
                Epipolar Geometry
CSE486, Penn State
                                                                                              Epipolar Constraint:
                                                                                          CSE486, Penn State

                                                             P                                                                           P

                                              pl                           pr                                            pl                            pr
                                                    el                   er          Or                                         el                   er          Or
                Ol                                                                                       Ol

  Epipoles:
  Epipoles:                                                                                         Epipoles:
                                                                                            Given Epipoles:
  • el: left image of Or                                                                    • el: left image of Or
  • er: right image of Ol                                                                   • er: right image of Ol
  Epipolar plane:                                                                            Given pl:
  • Three points: Ol,Or, and P define an epipolar plane                                      •consider its epipolar line: pl el
  Epipolar lines and epipolar constraint:                                                    •find epipolar plane: Ol,pl,el
  • Intersections of epipolar plane with the image planes                                    •intersect the epipolar plane with the right image plane
  • Corresponding points are on “conjugate” epipolar lines
                                  conjugate”                                                 •search for pr on the right epipolar line

The following slides are from Dr.Camps, PSU




Robert Collins                                                                            Robert Collins
CSE486, Penn State
                                     Essential Matrix                                     CSE486, Penn State
                                                                                                                     Essential Matrix

                                                         P                                                                           P
                                               Pl            Pr                                                            Pl            Pr

                                    pl                              pr                                              pl                          pr
                                              el                  er            Or                                        el                  er            Or
        Ol                                                                                        Ol
                                                         T                                                                           T




       Does this look familiar? Recall world to camera
       transformation by (R,T). Here, we are transforming
       from camera to camera.




Robert Collins                                                                            Robert Collins
CSE486, Penn State
                                     Essential Matrix                                     CSE486, Penn State
                                                                                                                     Essential Matrix
               Epipolar constraint: Pl, T and Pl - T are coplanar:                                      Epipolar constraint: Pl, T and Pl - T are coplanar:
                                         P                                                                                        P
                                 Pl          Pr                                                                           Pl          Pr

                                    pl                              pr                                              pl                          pr
                                              el                  er            Or                                       el                   er            Or
        Ol                                                                                        Ol
                                                         T                                                                           T




                                                                                                                                                                      4
Robert Collins
                      Vector Product as a Matrix                     Robert Collins
CSE486, Penn State

                            Multiplication
                                                                     CSE486, Penn State
                                                                                                 Essential Matrix
                                                                                   Epipolar constraint: Pl, T and Pl - T are coplanar:
                                                                                                             P
                                                                                                     Pl          Pr

                                                                                                pl                       pr
                                                                                                     el                er            Or
                                                                             Ol
                                                                                                             T




                          S has rank 2 ; it depends only on T




Robert Collins                                                       Robert Collins
CSE486, Penn State
                           Essential Matrix                          CSE486, Penn State
                                                                                          Essential Matrix Properties
              Epipolar constraint: Pl, T and Pl - T are coplanar:
                                        P
                                Pl          Pr

                          pl                        pr                 • has rank 2
                               el                 er            Or     • depends only on the EXTRINSIC Parameters (R & T)
        Ol
                                        T



Essential Matrix:                                                                     We will discuss more of the wonderful
                                                                                      properties of this matrix next time...




                                                                                                                                          5

More Related Content

More from zukun

My lyn tutorial 2009
My lyn tutorial 2009My lyn tutorial 2009
My lyn tutorial 2009zukun
 
ETHZ CV2012: Tutorial openCV
ETHZ CV2012: Tutorial openCVETHZ CV2012: Tutorial openCV
ETHZ CV2012: Tutorial openCVzukun
 
ETHZ CV2012: Information
ETHZ CV2012: InformationETHZ CV2012: Information
ETHZ CV2012: Informationzukun
 
Siwei lyu: natural image statistics
Siwei lyu: natural image statisticsSiwei lyu: natural image statistics
Siwei lyu: natural image statisticszukun
 
Lecture9 camera calibration
Lecture9 camera calibrationLecture9 camera calibration
Lecture9 camera calibrationzukun
 
Brunelli 2008: template matching techniques in computer vision
Brunelli 2008: template matching techniques in computer visionBrunelli 2008: template matching techniques in computer vision
Brunelli 2008: template matching techniques in computer visionzukun
 
Modern features-part-4-evaluation
Modern features-part-4-evaluationModern features-part-4-evaluation
Modern features-part-4-evaluationzukun
 
Modern features-part-3-software
Modern features-part-3-softwareModern features-part-3-software
Modern features-part-3-softwarezukun
 
Modern features-part-2-descriptors
Modern features-part-2-descriptorsModern features-part-2-descriptors
Modern features-part-2-descriptorszukun
 
Modern features-part-1-detectors
Modern features-part-1-detectorsModern features-part-1-detectors
Modern features-part-1-detectorszukun
 
Modern features-part-0-intro
Modern features-part-0-introModern features-part-0-intro
Modern features-part-0-introzukun
 
Lecture 02 internet video search
Lecture 02 internet video searchLecture 02 internet video search
Lecture 02 internet video searchzukun
 
Lecture 01 internet video search
Lecture 01 internet video searchLecture 01 internet video search
Lecture 01 internet video searchzukun
 
Lecture 03 internet video search
Lecture 03 internet video searchLecture 03 internet video search
Lecture 03 internet video searchzukun
 
Icml2012 tutorial representation_learning
Icml2012 tutorial representation_learningIcml2012 tutorial representation_learning
Icml2012 tutorial representation_learningzukun
 
Advances in discrete energy minimisation for computer vision
Advances in discrete energy minimisation for computer visionAdvances in discrete energy minimisation for computer vision
Advances in discrete energy minimisation for computer visionzukun
 
Gephi tutorial: quick start
Gephi tutorial: quick startGephi tutorial: quick start
Gephi tutorial: quick startzukun
 
EM algorithm and its application in probabilistic latent semantic analysis
EM algorithm and its application in probabilistic latent semantic analysisEM algorithm and its application in probabilistic latent semantic analysis
EM algorithm and its application in probabilistic latent semantic analysiszukun
 
Object recognition with pictorial structures
Object recognition with pictorial structuresObject recognition with pictorial structures
Object recognition with pictorial structureszukun
 
Iccv2011 learning spatiotemporal graphs of human activities
Iccv2011 learning spatiotemporal graphs of human activities Iccv2011 learning spatiotemporal graphs of human activities
Iccv2011 learning spatiotemporal graphs of human activities zukun
 

More from zukun (20)

My lyn tutorial 2009
My lyn tutorial 2009My lyn tutorial 2009
My lyn tutorial 2009
 
ETHZ CV2012: Tutorial openCV
ETHZ CV2012: Tutorial openCVETHZ CV2012: Tutorial openCV
ETHZ CV2012: Tutorial openCV
 
ETHZ CV2012: Information
ETHZ CV2012: InformationETHZ CV2012: Information
ETHZ CV2012: Information
 
Siwei lyu: natural image statistics
Siwei lyu: natural image statisticsSiwei lyu: natural image statistics
Siwei lyu: natural image statistics
 
Lecture9 camera calibration
Lecture9 camera calibrationLecture9 camera calibration
Lecture9 camera calibration
 
Brunelli 2008: template matching techniques in computer vision
Brunelli 2008: template matching techniques in computer visionBrunelli 2008: template matching techniques in computer vision
Brunelli 2008: template matching techniques in computer vision
 
Modern features-part-4-evaluation
Modern features-part-4-evaluationModern features-part-4-evaluation
Modern features-part-4-evaluation
 
Modern features-part-3-software
Modern features-part-3-softwareModern features-part-3-software
Modern features-part-3-software
 
Modern features-part-2-descriptors
Modern features-part-2-descriptorsModern features-part-2-descriptors
Modern features-part-2-descriptors
 
Modern features-part-1-detectors
Modern features-part-1-detectorsModern features-part-1-detectors
Modern features-part-1-detectors
 
Modern features-part-0-intro
Modern features-part-0-introModern features-part-0-intro
Modern features-part-0-intro
 
Lecture 02 internet video search
Lecture 02 internet video searchLecture 02 internet video search
Lecture 02 internet video search
 
Lecture 01 internet video search
Lecture 01 internet video searchLecture 01 internet video search
Lecture 01 internet video search
 
Lecture 03 internet video search
Lecture 03 internet video searchLecture 03 internet video search
Lecture 03 internet video search
 
Icml2012 tutorial representation_learning
Icml2012 tutorial representation_learningIcml2012 tutorial representation_learning
Icml2012 tutorial representation_learning
 
Advances in discrete energy minimisation for computer vision
Advances in discrete energy minimisation for computer visionAdvances in discrete energy minimisation for computer vision
Advances in discrete energy minimisation for computer vision
 
Gephi tutorial: quick start
Gephi tutorial: quick startGephi tutorial: quick start
Gephi tutorial: quick start
 
EM algorithm and its application in probabilistic latent semantic analysis
EM algorithm and its application in probabilistic latent semantic analysisEM algorithm and its application in probabilistic latent semantic analysis
EM algorithm and its application in probabilistic latent semantic analysis
 
Object recognition with pictorial structures
Object recognition with pictorial structuresObject recognition with pictorial structures
Object recognition with pictorial structures
 
Iccv2011 learning spatiotemporal graphs of human activities
Iccv2011 learning spatiotemporal graphs of human activities Iccv2011 learning spatiotemporal graphs of human activities
Iccv2011 learning spatiotemporal graphs of human activities
 

Lecture18

  • 1. Robert Collins Robert Collins CSE486, Penn State CSE486, Penn State Generalized Stereo Lecture 18: Key idea: Any two images showing an overlapping Generalized Stereo: view of the world can be treated as a stereo pair... Epipolar Geometry ... we just have to figure out how the two views are related. Some of the most “beautiful” math in vision concerns describing how multiple views are related, geometrically. Robert Collins Robert Collins CSE486, Penn State Recall: Epipolar Constraint CSE486, Penn State Review : Simple Stereo System P Important Stereo Vision Concept: z Left camera y Given a point in the left image, we don’t have to z search the whole right image for a corresponding point. y Right camera x The “epipolar constraint” reduces the search space to a one-dimensional line. Tx Epipolar line x Same Y Coord! baseline depth Equation relating depth and disparity disparity Robert Collins Robert Collins CSE486, Penn State Review: Epipolar Constraint CSE486, Penn State General Stereo C2 y2 x1 y1 C1 z1 z2 O2 O1 x2 In general, the cameras may be related by an arbitrary transformation (R,T) Epipolar Matrix Corresponding features are constrained to lie along conjugate epipolar lines (on the In general, intrinsic camera parameters may be different, and even unknown same row in the case of our simple setup). Fundamental Matrix 1
  • 2. Robert Collins Robert Collins CSE486, Penn State Epipolar Geometry CSE486, Penn State P pl pr el er Or Ol EPIPOLAR Epipoles: Epipoles: GEOMETRY • el: left image of Or • er: right image of Ol Epipolar plane: • Three points: Ol,Or, and P define an epipolar plane Epipolar lines and epipolar constraint: • Intersections of epipolar plane with the image planes • Corresponding points are on “conjugate” epipolar lines conjugate” The following slides are from Dr.Camps, PSU Robert Collins Robert Collins CSE486, Penn State CSE486, Penn State BORING!!! EPIPOLAR Let’s try again... GEOMETRY Robert Collins CSE486, Penn State Epipolar Geometry Robert Collins CSE486, Penn State Rays to Points in Scene A Visualization Tie threads on to the pins Would would Pinhead’s and connect eye look like close up? focal point to scene points Now what would this look like to a second observer? 2
  • 3. Robert Collins Robert Collins Rays Seen from Second Observer CSE486, Penn State CSE486, Penn State Rays Seen by the First Viewer epipolar lines epipolar lines Q u i c k T i m e ™ a n d a T I F F ( L Z W ) d e c o m p r e s s o r a r e n e e d e d t o s e e t h i s p i c t u r e . epipole epipole Image 2 Image 1 Robert Collins Robert Collins CSE486, Penn State Epipolar Geometry CSE486, Penn State Epipolar Geometry image1 image 2 image1 image 2 Epipole : location of cam2 Epipole : location of cam1 Corresponding points as seen by cam1. as seen by cam2. lie on conjugate epipolar lines Robert Collins Robert Collins CSE486, Penn State Epipolar Geometry CSE486, Penn State Epipole not Necessarily in Image image1 image 2 epipolar lines Conjugate epipolar lines induce a T r Q I e F u F n i c e k ( e T L d i m Z e W d e ™ ) t o d a e s n c e d o e m a t p h r i s e s p s i c o t r u r e . a generalized 1D “scan-line” ordering epipole on the images (analogous to traditional scan line ordering of rows in an image) Image 2 3
  • 4. Robert Collins Robert Collins Epipolar Geometry CSE486, Penn State Epipolar Constraint: CSE486, Penn State P P pl pr pl pr el er Or el er Or Ol Ol Epipoles: Epipoles: Epipoles: Given Epipoles: • el: left image of Or • el: left image of Or • er: right image of Ol • er: right image of Ol Epipolar plane: Given pl: • Three points: Ol,Or, and P define an epipolar plane •consider its epipolar line: pl el Epipolar lines and epipolar constraint: •find epipolar plane: Ol,pl,el • Intersections of epipolar plane with the image planes •intersect the epipolar plane with the right image plane • Corresponding points are on “conjugate” epipolar lines conjugate” •search for pr on the right epipolar line The following slides are from Dr.Camps, PSU Robert Collins Robert Collins CSE486, Penn State Essential Matrix CSE486, Penn State Essential Matrix P P Pl Pr Pl Pr pl pr pl pr el er Or el er Or Ol Ol T T Does this look familiar? Recall world to camera transformation by (R,T). Here, we are transforming from camera to camera. Robert Collins Robert Collins CSE486, Penn State Essential Matrix CSE486, Penn State Essential Matrix Epipolar constraint: Pl, T and Pl - T are coplanar: Epipolar constraint: Pl, T and Pl - T are coplanar: P P Pl Pr Pl Pr pl pr pl pr el er Or el er Or Ol Ol T T 4
  • 5. Robert Collins Vector Product as a Matrix Robert Collins CSE486, Penn State Multiplication CSE486, Penn State Essential Matrix Epipolar constraint: Pl, T and Pl - T are coplanar: P Pl Pr pl pr el er Or Ol T S has rank 2 ; it depends only on T Robert Collins Robert Collins CSE486, Penn State Essential Matrix CSE486, Penn State Essential Matrix Properties Epipolar constraint: Pl, T and Pl - T are coplanar: P Pl Pr pl pr • has rank 2 el er Or • depends only on the EXTRINSIC Parameters (R & T) Ol T Essential Matrix: We will discuss more of the wonderful properties of this matrix next time... 5