SlideShare a Scribd company logo
1 of 33
Structure	
  and	
  Mo-on	
  
	
  
	
  3D	
  Reconstruc-on	
  of	
  Cameras	
  and	
  Structure	
  
Vision and Perception 2009/2010
Prof.ssa Maria Fiora Pirri
- Filippo Bianchi - Murru Giovanni
- Mariani Marco - Ligios Antonello
	
  
Assignment	
  
	
  	
  	
  	
  The goal is the estimation of the structure of a
3D object (e.g. a 3D point- cloud) and the
motion on the cameras acquiring it.
Reconstruction is required for two data sets:
- the public collection of images available
here http://cs.gmu.edu/
%7Ekosecka/oldhouse2.zip
- a data set acquired by you (e.g. using a
simple camera)
A sparse reconstruction is sufficient; a metric
reconstruction is required.
Problems to be solved
•  Derive the structure of a scene from a sequence
of images
•  Estimation of Motion of the camera acquiring the
Structure
•  Reconstruction for two Data Sets
– a	
  public	
  collec+on	
  of	
  images	
  
– a	
  personal	
  dataset	
  	
  
Troubleshooting (1/2)
How to solve the problem for two images:
Objective *
Given two uncalibrated images compute a metric reconstruction (PM, P’M, {XMi}) of the
cameras and scene structure, i.e. a reconstruction that is within a similarity
transformation of the true cameras and scene structure.
Algorithm
(i) Compute a projective reconstruction (P, P', {Xi}):
(a) Compute the fundamental matrix from point correspondences xi <-> x’i
between the images.
(b) Camera retrieval: compute the camera matrices P, P' from the fundamental
matrix.
(c) Triangulation: for each point correspondence xi <-> x’i, compute the point Xi
in space that projects to these two image points.
(ii) Rectify the projective reconstruction to metric:
• either Direct method: Compute the homography H such that XEi = HXi from five or
more ground control points XEi with known Euclidean positions. Then the metric
reconstruction is
PM=PH-1, P‘M=P‘H-1, XMi=HXi
Troubleshoo-ng	
  (2/2)	
  
• or Stratified method:
(a) Affine reconstruction: Compute the plane at infinity, π∞, then upgrade the
projective reconstruction to an affine reconstruction with the homography
(b) Metric reconstruction: Compute the image of the absolute conic, ω,
and then upgrade the affine reconstruction to a metric reconstruction with
the homography
where A is obtained by Cholesky factorization from the equation AAT =
(MTωM)-1, and M is the first 3x3 submatrix of the camera in the affine
reconstruction for which ω is computed.
*Multiple View Geometry, Hurtley & Zisserman
The	
  Public	
  Data	
  Set	
   The	
  Personal	
  Data	
  Sets	
  
Example	
  of	
  Euclidian	
  reconstruc9on	
  using	
  
the	
  set	
  of	
  Rubik	
  
• 	
  This	
  is	
  an	
  example	
  of	
  the	
  
Euclidian	
  reconstruc-on	
  
performed	
  using	
  the	
  Rubik	
  
set.	
  
• 	
  The	
  points	
  are	
  taken	
  
manually	
  in	
  order	
  to	
  verify	
  
the	
  correctness	
  of	
  the	
  
output.	
  
• 	
  The	
  output	
  Euclidian	
  
represents	
  the	
  points	
  aGer	
  
the	
  the	
  projec-ve,	
  affine	
  
and	
  metric	
  reconstruc-ons	
  
are	
  performed.	
  
What	
  is	
  different	
  in	
  our	
  case?	
  (1/2)	
  
• we don’t have two images, but a set.
• we have to use stratified method
Ø  transition from the projective
reconstruction to metric reconstruction
through the affine one
What	
  is	
  different	
  in	
  our	
  case?	
  (2/2)	
  
•  The approach to reconstruction is to begin with a
projective reconstruction and then to refine it
progressively to an affine and finally a metric
reconstruction, if possible.
•  Of course, affine and metric reconstruction are not
possible without further information about the scene.
Projec-ve	
  Reconstruc-on	
  
Projective reconstruction can be divided into three
phases:
• Compute the fundamental Matrix from point
correspondences
• Compute the camera matrices P and P1 from the
fundamental matrix
• Triangulation
	
  
	
  
Projec-ve	
  Reconstruc-on	
  (example)	
  
Image	
  acquisi-on	
  
•  This method reads the images from a given folder in the
order in which they are. The images are read directly
with the function imread and saved in a cell array.
•  Order :
In old House the images are ordered, but it is not always.
Then to resolve it we use a particular function:
getCorrOrd(instead of getCorr)
getCorrOrd	
  
•  sorts the images according to correspondence found by
the match function, which does a match on keypoints
extracted from Sift
•  Result: I have pictures ordered by match
•  images similar to each other are adjacent
•  side effects :
•  computational time increases
•  small details may be lost at the expense of greater
precision on the most significant elements of the image
	
  
Correspondence	
  Match	
  
—  The SIFT algorithm is used to find the keypoints, which are
definted by:
—  Invariant descriptor of each keypoint
—  Localization (vector containing the row, column, orientation and scale of the
keypoint)
—  SIFT extract features from all images and memorized in two
vectors.
—  For each new image the features are compared with
precedent (in the order estabilished) image’s features,
previously memorized in the vectors.
—  The comparison is realized using dot products between unit
vector rather than Euclidean distances.
—  The correspondence between the images is positive only
when the ratio of vector angles (inverse cosine of the dot prod
of descriptors features) from the nearest to second nearest
neighbor is less than 0.6
	
  
SIFT	
  vs	
  HARRIS	
  
Need to find points to be correlated in different images
•  SIFT : greater robustness to rotation, translation, scaling
and changes in brightness.With the implementation
Lowe in C (Matlab interface) improves the computational
complexity
•  HARRIS : computationally easy to find, little strength,
high number of false matches
SIFT	
  Examples	
  
Original
Applying SIFT
Correspondence Filtering
—  The correspondences are filtered using RANSAC algorithm.
—  RANdom SAmple Consensus
	
  The key-points correspondences
determined with the matching algorithm,
can be divided in outliers and inliers. The
outliers are discarded because of their
numeric distance from the rest of the
data.
RANSAC	
  uses	
  a	
  threshold	
  parameter.	
  
This	
  is	
  chosen	
  empirically,	
  a	
  small	
  value	
  implies	
  
greater	
  precision	
  because	
  more	
  outliers	
  will	
  be	
  
discarded	
  
Fundamental Matrix F
•  Is the algebraic representation of epipolar geometry since Fx
describes the line on which the corresponding point x’ on the other
image must lie.
•  It correlates each point correspondence between two images.
•  The algorithm implemented in the function of Peter Kovesi is used to
estimate the RANSAC filtered correspondences and the F matrix
related to that.
0' =Fxx T
In	
  the	
  image,	
  is	
  shown	
  as	
  given	
  a	
  pair	
  of	
  images,	
  for	
  each	
  point	
  x	
  	
  in	
  one	
  image,	
  
there	
  	
  exist	
  a	
  corresponding	
  epipolar	
  line	
  l'	
  	
  in	
  the	
  other	
  image.	
  Any	
  point	
  x'	
  in	
  the	
  
second	
  image	
  matching	
  the	
  point	
  x	
  must	
  lie	
  on	
  the	
  epipolar	
  line	
  l'	
  .	
  
The GetF Function
• 	
  AGer	
  compu-ng	
  	
  the	
  correspondences,	
  	
  this	
  func-on	
  is	
  
used	
  for	
  es-ma-ng	
  the	
  fundamental	
  matrix	
  and	
  the	
  
epipoles	
  between	
  each	
  pair	
  of	
  corresponding	
  	
  images.	
  
• 	
  Construct	
  a	
  cell	
  of	
  size	
  N-­‐1,	
  where	
  N	
  is	
  the	
  number	
  of	
  
images.	
  
• 	
  The	
  matrix	
  F	
  is	
  directly	
  used	
  only	
  for	
  es-ma-ng	
  the	
  
second	
  camera:	
  P2	
  =	
  	
  [skewsimm(e1)*F|e1]	
  
• Anyway,	
  because	
  RANSAC	
  was	
  used	
  for	
  es-ma-ng	
  
matrix	
  F,	
  trough	
  this	
  procedure	
  it’s	
  possible	
  to	
  obtain	
  a	
  
set	
  of	
  inliers	
  and	
  outliers	
  for	
  each	
  pair	
  of	
  correponding	
  
images,	
  that	
  are	
  saved	
  in	
  corrispondencesR.	
  
3D points: estimation and updating
• 	
  The	
  first	
  es-ma-on	
  of	
  3D	
  points	
  is	
  derived	
  from	
  the	
  first	
  2	
  images	
  
	
  
• For every remaining images :
- Add new 3d points
- Update older one
• we introduce	
  a	
  new	
  line	
  in	
  the	
  array	
  of	
  3D	
  points	
  which	
  hold	
  a	
  new	
  
parameter	
  that	
  counts	
  the	
  number	
  of	
  images	
  that	
  update	
  every	
  single	
  
point.	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  Accuracy e precision will incraese
• we introduce a new threshold parameter used to remove noisy
associations (with elimina_associazioni function)to reduce the
error
Affine Reconstruction (1/4)
•  The essence of affine reconstruction is to
locate the plane at infinity by some means
•  We need to identify the matrix of
Homography “H” that allows us to
transform the 3D points (coming from
projective computation) and the camera
matrices P and P1.
Affine Reconstruction (2/4)
	
  	
  	
  We	
  iden-fy	
  the	
  plane	
  at	
  infinity	
  iden-fying	
  3	
  
vanishing	
  points	
  on	
  the	
  space.	
  
	
  	
  	
  To	
  compute	
  it	
  we	
  use	
  the	
  parallel	
  lines’	
  method	
  	
  :	
  
	
  	
  	
  	
  1)	
  Take	
  2	
  lines	
  that	
  in	
  reality	
  are	
  parallel	
  from	
  the	
  
first	
  image.	
  This	
  pair	
  of	
  lines	
  iden-fies	
  a	
  vanishing	
  
point	
  in	
  2D,	
  named	
  v	
  
2)	
  Take	
  a	
  line	
  l’	
  from	
  the	
  second	
  image	
  that	
  
corresponds	
  to	
  one	
  of	
  the	
  two	
  parallel	
  lines	
  
	
  	
  	
  	
  	
  
Affine Reconstruction (3/4)
•  The corresponding vanishing point in the
second image v’, can be computed as the
intersection of l’ and the epipolar line Fv
derived from the vanishing point v
computed in the first image.
•  Finding the null space of a matrix A like
is equivalent to find the 3D
vanishing point that
corresponds to v⎟⎟
⎠
⎞
⎜⎜
⎝
⎛
=
''
][
Pl
Pv
A T
x
Affine Reconstruction (4/4)	
  
•  Hence,	
  repea-ng	
  this	
  procedure	
  3	
  -mes,	
  we	
  
can	
  find	
  3	
  vanishing	
  point	
  in	
  the	
  space	
  that	
  
uniquely	
  iden-fies	
  the	
  plane	
  at	
  infinity.	
  
•  Now	
  using	
  the	
  plane	
  at	
  infinity	
  we	
  can	
  
calculate	
  the	
  affine	
  homography:	
  
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛Ι
= Τ
∞π
0|
H
Pairs	
  of	
  parallel	
  lines	
  	
  
Pairs	
  of	
  parallel	
  lines	
  
Under a pure translational camera motion, 3D points appear to slide along parallel
rails. The images of these parallel lines intersect in a vanishing point
corresponding to the translation direction. The epipole e is the vanishing point.
Figure	
  1.2	
  
Metric Reconstruction (1/5)
• The key to metric reconstruction is the identification of the absolute conic .
• Compute the IAC (image of asbsolute conic), ω, and then upgrade the affine
reconstruction to a metric, with the homography
• The projection of the absolute conic into an image depends only on the
calibration matrix of the camera and not on position and orientation.
• Usually images taken with the same camera, implies that both cameras P and
P1 have the same calibration matrix. Hence ω= ω1, i.e. IAC is the same in both
images.
⎟
⎟
⎠
⎞
⎜
⎜
⎝
⎛ Α
=
−
1
1
H
Metric	
  Reconstruc-on	
  (2/5)	
  
•  The constraints arising from scene
orthogonality are used
•  5 constraints are needed
•  3 can be computed from
orthogonal pairs of parallel lines
•  2 can be retrieved supposing
zero skew and unit aspect ratio
•  w2=0
•  w1=w3.
•  The algorithm used to compute the ω
is that explained in the paper :
Crea-ng	
  Architectural	
  Models	
  from	
  Images,	
  
by	
  David	
  Liebowitz,	
  Antonio	
  Criminisi	
  and	
  
Andrew	
  Zisserman	
  cited	
  on	
  the	
  right.
Metric	
  Reconstruc-on	
  (3/5)	
  
The 2 constraints related to zero skew and unit aspect ratio are
equivalent to those expressed by square pixels:
ω₁₂ = ω₂₁ = 0
ω₁₁ = ω₂₂
These 3 pairs of parallel lines are mutually
orthogonal to each other and we can say
that 3 constraints arise from this scene:	
  
	
  
v₁T ω v₂ = 0
	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  v₁T ω v₃ = 0
v₂T ω v₃ = 0	
  
Prima coppia rette parallele
Seconda coppia rette parallele
Terza coppia rette parallele
• 	
  The	
  two	
  condi-ons	
  arising	
  from	
  
square	
  pixels	
  can	
  be	
  wrieen	
  as	
  rows	
  in	
  
a	
  matrix	
  Ā,	
  
a₄=	
  [0	
  1	
  0	
  0	
  0	
  0]	
  
a₅=	
  [1	
  0	
  -­‐1	
  0	
  0	
  0]	
  
	
  
• 	
  a₃	
  a₂	
  a	
  ₁	
  are	
  the	
  rows	
  corresponding	
  
to	
  the	
  constraints	
  arising	
  from	
  the	
  
vanishing	
  point	
  generated	
  by	
  
orthogonal	
  pairs	
  of	
  parallel	
  lines,	
  
expressed	
  in	
  the	
  way	
  illustrated	
  in	
  the	
  
paper,	
  and	
  in	
  the	
  generic	
  algorithm	
  on	
  
the	
  right.	
  
	
  
Ā	
  =	
  [a	
  ₁;	
  a₂;	
  	
  a₃;	
  a₄;	
  a₅]	
  	
  
	
  
• 	
  To	
  find	
  the	
  null	
  space	
  of	
  Ā	
  	
  
• 	
  apply	
  svd(Ā)	
  	
  and	
  get	
  the	
  last	
  
column	
  of	
  V	
  
• 	
  Equivalent	
  to	
  solve	
  	
  
• 	
  Ā	
  w	
  =	
  0	
  	
  
Metric	
  Reconstruc-on	
  (4/5)	
  
Metric	
  Reconstruc-on	
  (5/5)	
  
•  The knowledge of w implies the knowledge of ω
•  Hence we can compute A (that is not the A previously
computed in SVD), such that :
§  AA T = (MTωM)- 1 = Z
§  where M = P(1:3, 1:3)
•  To compute A we use the cholesky decomposition:
§  A= chol (Z,’lower’)
•  In fact this returns Cholesky matrix A that satisfies AA T = Z
•  Using the A value we can build the metric homography such
that is possible to upgrade the affine to the metric
reconstruction
⎟
⎟
⎠
⎞
⎜
⎜
⎝
⎛ Α
=
−
1
1
H

More Related Content

What's hot

CV_Chap 6 Motion Representation
CV_Chap 6 Motion RepresentationCV_Chap 6 Motion Representation
CV_Chap 6 Motion RepresentationKhushali Kathiriya
 
Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...
Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...
Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...Vincent Sitzmann
 
Scale Invariant Feature Transform
Scale Invariant Feature TransformScale Invariant Feature Transform
Scale Invariant Feature Transformkislayabhi
 
CG OpenGL surface detection+illumination+rendering models-course 9
CG OpenGL surface detection+illumination+rendering models-course 9CG OpenGL surface detection+illumination+rendering models-course 9
CG OpenGL surface detection+illumination+rendering models-course 9fungfung Chen
 
Object tracking presentation
Object tracking  presentationObject tracking  presentation
Object tracking presentationMrsShwetaBanait1
 
Passive stereo vision with deep learning
Passive stereo vision with deep learningPassive stereo vision with deep learning
Passive stereo vision with deep learningYu Huang
 
A Beginner's Guide to Monocular Depth Estimation
A Beginner's Guide to Monocular Depth EstimationA Beginner's Guide to Monocular Depth Estimation
A Beginner's Guide to Monocular Depth EstimationRyo Takahashi
 
Lecture9 camera calibration
Lecture9 camera calibrationLecture9 camera calibration
Lecture9 camera calibrationzukun
 
Computer Graphics: Visible surface detection methods
Computer Graphics: Visible surface detection methodsComputer Graphics: Visible surface detection methods
Computer Graphics: Visible surface detection methodsJoseph Charles
 
3-d interpretation from single 2-d image for autonomous driving II
3-d interpretation from single 2-d image for autonomous driving II3-d interpretation from single 2-d image for autonomous driving II
3-d interpretation from single 2-d image for autonomous driving IIYu Huang
 

What's hot (20)

CV_Chap 6 Motion Representation
CV_Chap 6 Motion RepresentationCV_Chap 6 Motion Representation
CV_Chap 6 Motion Representation
 
Ray tracing
Ray tracingRay tracing
Ray tracing
 
Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...
Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...
Tutorial on Generalization in Neural Fields, CVPR 2022 Tutorial on Neural Fie...
 
Human Action Recognition
Human Action RecognitionHuman Action Recognition
Human Action Recognition
 
Edge detection
Edge detectionEdge detection
Edge detection
 
Scale Invariant Feature Transform
Scale Invariant Feature TransformScale Invariant Feature Transform
Scale Invariant Feature Transform
 
CG OpenGL surface detection+illumination+rendering models-course 9
CG OpenGL surface detection+illumination+rendering models-course 9CG OpenGL surface detection+illumination+rendering models-course 9
CG OpenGL surface detection+illumination+rendering models-course 9
 
Image pyramid
Image pyramidImage pyramid
Image pyramid
 
Object tracking final
Object tracking finalObject tracking final
Object tracking final
 
Object tracking presentation
Object tracking  presentationObject tracking  presentation
Object tracking presentation
 
Passive stereo vision with deep learning
Passive stereo vision with deep learningPassive stereo vision with deep learning
Passive stereo vision with deep learning
 
A Beginner's Guide to Monocular Depth Estimation
A Beginner's Guide to Monocular Depth EstimationA Beginner's Guide to Monocular Depth Estimation
A Beginner's Guide to Monocular Depth Estimation
 
Introduction to OpenCV
Introduction to OpenCVIntroduction to OpenCV
Introduction to OpenCV
 
Lecture9 camera calibration
Lecture9 camera calibrationLecture9 camera calibration
Lecture9 camera calibration
 
SIFT
SIFTSIFT
SIFT
 
Stereo vision
Stereo visionStereo vision
Stereo vision
 
Moving object detection
Moving object detectionMoving object detection
Moving object detection
 
Computer Graphics: Visible surface detection methods
Computer Graphics: Visible surface detection methodsComputer Graphics: Visible surface detection methods
Computer Graphics: Visible surface detection methods
 
fractals
fractalsfractals
fractals
 
3-d interpretation from single 2-d image for autonomous driving II
3-d interpretation from single 2-d image for autonomous driving II3-d interpretation from single 2-d image for autonomous driving II
3-d interpretation from single 2-d image for autonomous driving II
 

Viewers also liked

"High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro...
"High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro..."High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro...
"High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro...Edge AI and Vision Alliance
 
AI Strategies for Solving Poker Texas Hold'em
AI Strategies for Solving Poker Texas Hold'emAI Strategies for Solving Poker Texas Hold'em
AI Strategies for Solving Poker Texas Hold'emGiovanni Murru
 
Expert System - Artificial intelligence
Expert System - Artificial intelligenceExpert System - Artificial intelligence
Expert System - Artificial intelligenceDr. Abdul Ahad Abro
 
3DTV - Past, Present and Future
3DTV - Past, Present and Future3DTV - Past, Present and Future
3DTV - Past, Present and FutureTouradj Ebrahimi
 
Fingerprint sensor applications and technologies – Consumer market focus - 20...
Fingerprint sensor applications and technologies – Consumer market focus - 20...Fingerprint sensor applications and technologies – Consumer market focus - 20...
Fingerprint sensor applications and technologies – Consumer market focus - 20...Yole Developpement
 
Sensors for Cellphones and Tablets - 2016 Report by Yole Developpement
Sensors for Cellphones and Tablets - 2016 Report by Yole DeveloppementSensors for Cellphones and Tablets - 2016 Report by Yole Developpement
Sensors for Cellphones and Tablets - 2016 Report by Yole DeveloppementYole Developpement
 
Imaging Technologies for Automotive 2016 Report by Yole Developpement
Imaging Technologies for Automotive 2016 Report by Yole Developpement	Imaging Technologies for Automotive 2016 Report by Yole Developpement
Imaging Technologies for Automotive 2016 Report by Yole Developpement Yole Developpement
 
Deep Learning - The Past, Present and Future of Artificial Intelligence
Deep Learning - The Past, Present and Future of Artificial IntelligenceDeep Learning - The Past, Present and Future of Artificial Intelligence
Deep Learning - The Past, Present and Future of Artificial IntelligenceLukas Masuch
 

Viewers also liked (9)

"High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro...
"High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro..."High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro...
"High-resolution 3D Reconstruction on a Mobile Processor," a Presentation fro...
 
AI Strategies for Solving Poker Texas Hold'em
AI Strategies for Solving Poker Texas Hold'emAI Strategies for Solving Poker Texas Hold'em
AI Strategies for Solving Poker Texas Hold'em
 
Expert System - Artificial intelligence
Expert System - Artificial intelligenceExpert System - Artificial intelligence
Expert System - Artificial intelligence
 
3DTV - Past, Present and Future
3DTV - Past, Present and Future3DTV - Past, Present and Future
3DTV - Past, Present and Future
 
Advanced driver assistance systems
Advanced driver assistance systemsAdvanced driver assistance systems
Advanced driver assistance systems
 
Fingerprint sensor applications and technologies – Consumer market focus - 20...
Fingerprint sensor applications and technologies – Consumer market focus - 20...Fingerprint sensor applications and technologies – Consumer market focus - 20...
Fingerprint sensor applications and technologies – Consumer market focus - 20...
 
Sensors for Cellphones and Tablets - 2016 Report by Yole Developpement
Sensors for Cellphones and Tablets - 2016 Report by Yole DeveloppementSensors for Cellphones and Tablets - 2016 Report by Yole Developpement
Sensors for Cellphones and Tablets - 2016 Report by Yole Developpement
 
Imaging Technologies for Automotive 2016 Report by Yole Developpement
Imaging Technologies for Automotive 2016 Report by Yole Developpement	Imaging Technologies for Automotive 2016 Report by Yole Developpement
Imaging Technologies for Automotive 2016 Report by Yole Developpement
 
Deep Learning - The Past, Present and Future of Artificial Intelligence
Deep Learning - The Past, Present and Future of Artificial IntelligenceDeep Learning - The Past, Present and Future of Artificial Intelligence
Deep Learning - The Past, Present and Future of Artificial Intelligence
 

Similar to Structure and Motion - 3D Reconstruction of Cameras and Structure

Anish_Hemmady_assignmnt1_Report
Anish_Hemmady_assignmnt1_ReportAnish_Hemmady_assignmnt1_Report
Anish_Hemmady_assignmnt1_Reportanish h
 
Image processing second unit Notes
Image processing second unit NotesImage processing second unit Notes
Image processing second unit NotesAAKANKSHA JAIN
 
Report bep thomas_blanken
Report bep thomas_blankenReport bep thomas_blanken
Report bep thomas_blankenxepost
 
Basics of CT- Lecture 9.ppt
Basics of CT- Lecture 9.pptBasics of CT- Lecture 9.ppt
Basics of CT- Lecture 9.pptMagde Gad
 
05 contours seg_matching
05 contours seg_matching05 contours seg_matching
05 contours seg_matchingankit_ppt
 
Kulum alin-11 jan2014
Kulum alin-11 jan2014Kulum alin-11 jan2014
Kulum alin-11 jan2014rolly purnomo
 
Building 3D Morphable Models from 2D Images
Building 3D Morphable Models from 2D ImagesBuilding 3D Morphable Models from 2D Images
Building 3D Morphable Models from 2D ImagesShanglin Yang
 
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...cscpconf
 
Multiple region of interest tracking of non rigid objects using demon's algor...
Multiple region of interest tracking of non rigid objects using demon's algor...Multiple region of interest tracking of non rigid objects using demon's algor...
Multiple region of interest tracking of non rigid objects using demon's algor...csandit
 
Image pre processing
Image pre processingImage pre processing
Image pre processingAshish Kumar
 
matdid950092.pdf
matdid950092.pdfmatdid950092.pdf
matdid950092.pdflencho3d
 
chapter 4 computervision.pdf IT IS ABOUT COMUTER VISION
chapter 4 computervision.pdf IT IS ABOUT COMUTER VISIONchapter 4 computervision.pdf IT IS ABOUT COMUTER VISION
chapter 4 computervision.pdf IT IS ABOUT COMUTER VISIONshesnasuneer
 
Image recogonization
Image recogonizationImage recogonization
Image recogonizationSANTOSH RATH
 
Writeup advanced lane_lines_project
Writeup advanced lane_lines_projectWriteup advanced lane_lines_project
Writeup advanced lane_lines_projectManish Jauhari
 

Similar to Structure and Motion - 3D Reconstruction of Cameras and Structure (20)

Anish_Hemmady_assignmnt1_Report
Anish_Hemmady_assignmnt1_ReportAnish_Hemmady_assignmnt1_Report
Anish_Hemmady_assignmnt1_Report
 
Image processing second unit Notes
Image processing second unit NotesImage processing second unit Notes
Image processing second unit Notes
 
Report bep thomas_blanken
Report bep thomas_blankenReport bep thomas_blanken
Report bep thomas_blanken
 
Final
FinalFinal
Final
 
Basics of CT- Lecture 9.ppt
Basics of CT- Lecture 9.pptBasics of CT- Lecture 9.ppt
Basics of CT- Lecture 9.ppt
 
05 contours seg_matching
05 contours seg_matching05 contours seg_matching
05 contours seg_matching
 
Kulum alin-11 jan2014
Kulum alin-11 jan2014Kulum alin-11 jan2014
Kulum alin-11 jan2014
 
Building 3D Morphable Models from 2D Images
Building 3D Morphable Models from 2D ImagesBuilding 3D Morphable Models from 2D Images
Building 3D Morphable Models from 2D Images
 
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...
MULTIPLE REGION OF INTEREST TRACKING OF NON-RIGID OBJECTS USING DEMON'S ALGOR...
 
Multiple region of interest tracking of non rigid objects using demon's algor...
Multiple region of interest tracking of non rigid objects using demon's algor...Multiple region of interest tracking of non rigid objects using demon's algor...
Multiple region of interest tracking of non rigid objects using demon's algor...
 
Image pre processing
Image pre processingImage pre processing
Image pre processing
 
Medial axis transformation based skeletonzation of image patterns using image...
Medial axis transformation based skeletonzation of image patterns using image...Medial axis transformation based skeletonzation of image patterns using image...
Medial axis transformation based skeletonzation of image patterns using image...
 
998-isvc16
998-isvc16998-isvc16
998-isvc16
 
poster
posterposter
poster
 
matdid950092.pdf
matdid950092.pdfmatdid950092.pdf
matdid950092.pdf
 
chapter 4 computervision.pdf IT IS ABOUT COMUTER VISION
chapter 4 computervision.pdf IT IS ABOUT COMUTER VISIONchapter 4 computervision.pdf IT IS ABOUT COMUTER VISION
chapter 4 computervision.pdf IT IS ABOUT COMUTER VISION
 
Oc2423022305
Oc2423022305Oc2423022305
Oc2423022305
 
Image recogonization
Image recogonizationImage recogonization
Image recogonization
 
Writeup advanced lane_lines_project
Writeup advanced lane_lines_projectWriteup advanced lane_lines_project
Writeup advanced lane_lines_project
 
p927-chang
p927-changp927-chang
p927-chang
 

More from Giovanni Murru

About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...
About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...
About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...Giovanni Murru
 
Controllers for 3R Robot
Controllers for 3R RobotControllers for 3R Robot
Controllers for 3R RobotGiovanni Murru
 
Passive Optical Networks: The MAC protocols
Passive Optical Networks: The MAC protocolsPassive Optical Networks: The MAC protocols
Passive Optical Networks: The MAC protocolsGiovanni Murru
 
About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...
About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...
About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...Giovanni Murru
 
Mackey Glass Time Series Prediction
Mackey Glass Time Series PredictionMackey Glass Time Series Prediction
Mackey Glass Time Series PredictionGiovanni Murru
 
Task Constrained Motion Planning for Snake Robot
Task Constrained Motion Planning for Snake RobotTask Constrained Motion Planning for Snake Robot
Task Constrained Motion Planning for Snake RobotGiovanni Murru
 
About the paper: Development and application of a new steady-hand manipulator...
About the paper: Development and application of a new steady-hand manipulator...About the paper: Development and application of a new steady-hand manipulator...
About the paper: Development and application of a new steady-hand manipulator...Giovanni Murru
 
Practical Augmented Visualization on Handheld Devices for Cultural Heritage
Practical Augmented Visualization on Handheld Devices for Cultural Heritage Practical Augmented Visualization on Handheld Devices for Cultural Heritage
Practical Augmented Visualization on Handheld Devices for Cultural Heritage Giovanni Murru
 

More from Giovanni Murru (9)

GLUTetris
GLUTetrisGLUTetris
GLUTetris
 
About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...
About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...
About the paper USC CINAPS Builds Bridges Observing and Monitoring the Southe...
 
Controllers for 3R Robot
Controllers for 3R RobotControllers for 3R Robot
Controllers for 3R Robot
 
Passive Optical Networks: The MAC protocols
Passive Optical Networks: The MAC protocolsPassive Optical Networks: The MAC protocols
Passive Optical Networks: The MAC protocols
 
About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...
About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...
About the paper: Graph Connectivity Measures for Unsupervised Word Sense Disa...
 
Mackey Glass Time Series Prediction
Mackey Glass Time Series PredictionMackey Glass Time Series Prediction
Mackey Glass Time Series Prediction
 
Task Constrained Motion Planning for Snake Robot
Task Constrained Motion Planning for Snake RobotTask Constrained Motion Planning for Snake Robot
Task Constrained Motion Planning for Snake Robot
 
About the paper: Development and application of a new steady-hand manipulator...
About the paper: Development and application of a new steady-hand manipulator...About the paper: Development and application of a new steady-hand manipulator...
About the paper: Development and application of a new steady-hand manipulator...
 
Practical Augmented Visualization on Handheld Devices for Cultural Heritage
Practical Augmented Visualization on Handheld Devices for Cultural Heritage Practical Augmented Visualization on Handheld Devices for Cultural Heritage
Practical Augmented Visualization on Handheld Devices for Cultural Heritage
 

Recently uploaded

Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)Suman Mia
 
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...Dr.Costas Sachpazis
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...Soham Mondal
 
UNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its PerformanceUNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its Performancesivaprakash250
 
SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )Tsuyoshi Horigome
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSKurinjimalarL3
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130Suhani Kapoor
 
result management system report for college project
result management system report for college projectresult management system report for college project
result management system report for college projectTonystark477637
 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Serviceranjana rawat
 
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...roncy bisnoi
 
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130Suhani Kapoor
 
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordCCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordAsst.prof M.Gokilavani
 
Porous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingPorous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingrakeshbaidya232001
 
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
 
(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
 
Microscopic Analysis of Ceramic Materials.pptx
Microscopic Analysis of Ceramic Materials.pptxMicroscopic Analysis of Ceramic Materials.pptx
Microscopic Analysis of Ceramic Materials.pptxpurnimasatapathy1234
 
Coefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxCoefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxAsutosh Ranjan
 

Recently uploaded (20)

Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)Software Development Life Cycle By  Team Orange (Dept. of Pharmacy)
Software Development Life Cycle By Team Orange (Dept. of Pharmacy)
 
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
Structural Analysis and Design of Foundations: A Comprehensive Handbook for S...
 
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
OSVC_Meta-Data based Simulation Automation to overcome Verification Challenge...
 
UNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its PerformanceUNIT - IV - Air Compressors and its Performance
UNIT - IV - Air Compressors and its Performance
 
SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )
 
Water Industry Process Automation & Control Monthly - April 2024
Water Industry Process Automation & Control Monthly - April 2024Water Industry Process Automation & Control Monthly - April 2024
Water Industry Process Automation & Control Monthly - April 2024
 
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICSAPPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
APPLICATIONS-AC/DC DRIVES-OPERATING CHARACTERISTICS
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
 
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINEDJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
DJARUM4D - SLOT GACOR ONLINE | SLOT DEMO ONLINE
 
result management system report for college project
result management system report for college projectresult management system report for college project
result management system report for college project
 
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
(RIA) Call Girls Bhosari ( 7001035870 ) HI-Fi Pune Escorts Service
 
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
 
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
VIP Call Girls Service Kondapur Hyderabad Call +91-8250192130
 
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete RecordCCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
CCS335 _ Neural Networks and Deep Learning Laboratory_Lab Complete Record
 
Porous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writingPorous Ceramics seminar and technical writing
Porous Ceramics seminar and technical writing
 
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANJALI) Dange Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
 
(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(PRIYA) Rajgurunagar Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
 
Microscopic Analysis of Ceramic Materials.pptx
Microscopic Analysis of Ceramic Materials.pptxMicroscopic Analysis of Ceramic Materials.pptx
Microscopic Analysis of Ceramic Materials.pptx
 
Coefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptxCoefficient of Thermal Expansion and their Importance.pptx
Coefficient of Thermal Expansion and their Importance.pptx
 

Structure and Motion - 3D Reconstruction of Cameras and Structure

  • 1. Structure  and  Mo-on      3D  Reconstruc-on  of  Cameras  and  Structure   Vision and Perception 2009/2010 Prof.ssa Maria Fiora Pirri - Filippo Bianchi - Murru Giovanni - Mariani Marco - Ligios Antonello  
  • 2. Assignment          The goal is the estimation of the structure of a 3D object (e.g. a 3D point- cloud) and the motion on the cameras acquiring it. Reconstruction is required for two data sets: - the public collection of images available here http://cs.gmu.edu/ %7Ekosecka/oldhouse2.zip - a data set acquired by you (e.g. using a simple camera) A sparse reconstruction is sufficient; a metric reconstruction is required.
  • 3. Problems to be solved •  Derive the structure of a scene from a sequence of images •  Estimation of Motion of the camera acquiring the Structure •  Reconstruction for two Data Sets – a  public  collec+on  of  images   – a  personal  dataset    
  • 4. Troubleshooting (1/2) How to solve the problem for two images: Objective * Given two uncalibrated images compute a metric reconstruction (PM, P’M, {XMi}) of the cameras and scene structure, i.e. a reconstruction that is within a similarity transformation of the true cameras and scene structure. Algorithm (i) Compute a projective reconstruction (P, P', {Xi}): (a) Compute the fundamental matrix from point correspondences xi <-> x’i between the images. (b) Camera retrieval: compute the camera matrices P, P' from the fundamental matrix. (c) Triangulation: for each point correspondence xi <-> x’i, compute the point Xi in space that projects to these two image points. (ii) Rectify the projective reconstruction to metric: • either Direct method: Compute the homography H such that XEi = HXi from five or more ground control points XEi with known Euclidean positions. Then the metric reconstruction is PM=PH-1, P‘M=P‘H-1, XMi=HXi
  • 5. Troubleshoo-ng  (2/2)   • or Stratified method: (a) Affine reconstruction: Compute the plane at infinity, π∞, then upgrade the projective reconstruction to an affine reconstruction with the homography (b) Metric reconstruction: Compute the image of the absolute conic, ω, and then upgrade the affine reconstruction to a metric reconstruction with the homography where A is obtained by Cholesky factorization from the equation AAT = (MTωM)-1, and M is the first 3x3 submatrix of the camera in the affine reconstruction for which ω is computed. *Multiple View Geometry, Hurtley & Zisserman
  • 6. The  Public  Data  Set   The  Personal  Data  Sets  
  • 7. Example  of  Euclidian  reconstruc9on  using   the  set  of  Rubik   •   This  is  an  example  of  the   Euclidian  reconstruc-on   performed  using  the  Rubik   set.   •   The  points  are  taken   manually  in  order  to  verify   the  correctness  of  the   output.   •   The  output  Euclidian   represents  the  points  aGer   the  the  projec-ve,  affine   and  metric  reconstruc-ons   are  performed.  
  • 8. What  is  different  in  our  case?  (1/2)   • we don’t have two images, but a set. • we have to use stratified method Ø  transition from the projective reconstruction to metric reconstruction through the affine one
  • 9. What  is  different  in  our  case?  (2/2)   •  The approach to reconstruction is to begin with a projective reconstruction and then to refine it progressively to an affine and finally a metric reconstruction, if possible. •  Of course, affine and metric reconstruction are not possible without further information about the scene.
  • 10. Projec-ve  Reconstruc-on   Projective reconstruction can be divided into three phases: • Compute the fundamental Matrix from point correspondences • Compute the camera matrices P and P1 from the fundamental matrix • Triangulation    
  • 12. Image  acquisi-on   •  This method reads the images from a given folder in the order in which they are. The images are read directly with the function imread and saved in a cell array. •  Order : In old House the images are ordered, but it is not always. Then to resolve it we use a particular function: getCorrOrd(instead of getCorr)
  • 13. getCorrOrd   •  sorts the images according to correspondence found by the match function, which does a match on keypoints extracted from Sift •  Result: I have pictures ordered by match •  images similar to each other are adjacent •  side effects : •  computational time increases •  small details may be lost at the expense of greater precision on the most significant elements of the image  
  • 14. Correspondence  Match   —  The SIFT algorithm is used to find the keypoints, which are definted by: —  Invariant descriptor of each keypoint —  Localization (vector containing the row, column, orientation and scale of the keypoint) —  SIFT extract features from all images and memorized in two vectors. —  For each new image the features are compared with precedent (in the order estabilished) image’s features, previously memorized in the vectors. —  The comparison is realized using dot products between unit vector rather than Euclidean distances. —  The correspondence between the images is positive only when the ratio of vector angles (inverse cosine of the dot prod of descriptors features) from the nearest to second nearest neighbor is less than 0.6  
  • 15. SIFT  vs  HARRIS   Need to find points to be correlated in different images •  SIFT : greater robustness to rotation, translation, scaling and changes in brightness.With the implementation Lowe in C (Matlab interface) improves the computational complexity •  HARRIS : computationally easy to find, little strength, high number of false matches
  • 17. Correspondence Filtering —  The correspondences are filtered using RANSAC algorithm. —  RANdom SAmple Consensus  The key-points correspondences determined with the matching algorithm, can be divided in outliers and inliers. The outliers are discarded because of their numeric distance from the rest of the data. RANSAC  uses  a  threshold  parameter.   This  is  chosen  empirically,  a  small  value  implies   greater  precision  because  more  outliers  will  be   discarded  
  • 18. Fundamental Matrix F •  Is the algebraic representation of epipolar geometry since Fx describes the line on which the corresponding point x’ on the other image must lie. •  It correlates each point correspondence between two images. •  The algorithm implemented in the function of Peter Kovesi is used to estimate the RANSAC filtered correspondences and the F matrix related to that. 0' =Fxx T
  • 19. In  the  image,  is  shown  as  given  a  pair  of  images,  for  each  point  x    in  one  image,   there    exist  a  corresponding  epipolar  line  l'    in  the  other  image.  Any  point  x'  in  the   second  image  matching  the  point  x  must  lie  on  the  epipolar  line  l'  .  
  • 20. The GetF Function •   AGer  compu-ng    the  correspondences,    this  func-on  is   used  for  es-ma-ng  the  fundamental  matrix  and  the   epipoles  between  each  pair  of  corresponding    images.   •   Construct  a  cell  of  size  N-­‐1,  where  N  is  the  number  of   images.   •   The  matrix  F  is  directly  used  only  for  es-ma-ng  the   second  camera:  P2  =    [skewsimm(e1)*F|e1]   • Anyway,  because  RANSAC  was  used  for  es-ma-ng   matrix  F,  trough  this  procedure  it’s  possible  to  obtain  a   set  of  inliers  and  outliers  for  each  pair  of  correponding   images,  that  are  saved  in  corrispondencesR.  
  • 21. 3D points: estimation and updating •   The  first  es-ma-on  of  3D  points  is  derived  from  the  first  2  images     • For every remaining images : - Add new 3d points - Update older one • we introduce  a  new  line  in  the  array  of  3D  points  which  hold  a  new   parameter  that  counts  the  number  of  images  that  update  every  single   point.                                              Accuracy e precision will incraese • we introduce a new threshold parameter used to remove noisy associations (with elimina_associazioni function)to reduce the error
  • 22. Affine Reconstruction (1/4) •  The essence of affine reconstruction is to locate the plane at infinity by some means •  We need to identify the matrix of Homography “H” that allows us to transform the 3D points (coming from projective computation) and the camera matrices P and P1.
  • 23. Affine Reconstruction (2/4)      We  iden-fy  the  plane  at  infinity  iden-fying  3   vanishing  points  on  the  space.        To  compute  it  we  use  the  parallel  lines’  method    :          1)  Take  2  lines  that  in  reality  are  parallel  from  the   first  image.  This  pair  of  lines  iden-fies  a  vanishing   point  in  2D,  named  v   2)  Take  a  line  l’  from  the  second  image  that   corresponds  to  one  of  the  two  parallel  lines            
  • 24. Affine Reconstruction (3/4) •  The corresponding vanishing point in the second image v’, can be computed as the intersection of l’ and the epipolar line Fv derived from the vanishing point v computed in the first image. •  Finding the null space of a matrix A like is equivalent to find the 3D vanishing point that corresponds to v⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ = '' ][ Pl Pv A T x
  • 25. Affine Reconstruction (4/4)   •  Hence,  repea-ng  this  procedure  3  -mes,  we   can  find  3  vanishing  point  in  the  space  that   uniquely  iden-fies  the  plane  at  infinity.   •  Now  using  the  plane  at  infinity  we  can   calculate  the  affine  homography:   ⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛Ι = Τ ∞π 0| H
  • 26. Pairs  of  parallel  lines     Pairs  of  parallel  lines  
  • 27. Under a pure translational camera motion, 3D points appear to slide along parallel rails. The images of these parallel lines intersect in a vanishing point corresponding to the translation direction. The epipole e is the vanishing point. Figure  1.2  
  • 28. Metric Reconstruction (1/5) • The key to metric reconstruction is the identification of the absolute conic . • Compute the IAC (image of asbsolute conic), ω, and then upgrade the affine reconstruction to a metric, with the homography • The projection of the absolute conic into an image depends only on the calibration matrix of the camera and not on position and orientation. • Usually images taken with the same camera, implies that both cameras P and P1 have the same calibration matrix. Hence ω= ω1, i.e. IAC is the same in both images. ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ Α = − 1 1 H
  • 29. Metric  Reconstruc-on  (2/5)   •  The constraints arising from scene orthogonality are used •  5 constraints are needed •  3 can be computed from orthogonal pairs of parallel lines •  2 can be retrieved supposing zero skew and unit aspect ratio •  w2=0 •  w1=w3. •  The algorithm used to compute the ω is that explained in the paper : Crea-ng  Architectural  Models  from  Images,   by  David  Liebowitz,  Antonio  Criminisi  and   Andrew  Zisserman  cited  on  the  right.
  • 30. Metric  Reconstruc-on  (3/5)   The 2 constraints related to zero skew and unit aspect ratio are equivalent to those expressed by square pixels: ω₁₂ = ω₂₁ = 0 ω₁₁ = ω₂₂
  • 31. These 3 pairs of parallel lines are mutually orthogonal to each other and we can say that 3 constraints arise from this scene:     v₁T ω v₂ = 0                                          v₁T ω v₃ = 0 v₂T ω v₃ = 0   Prima coppia rette parallele Seconda coppia rette parallele Terza coppia rette parallele
  • 32. •   The  two  condi-ons  arising  from   square  pixels  can  be  wrieen  as  rows  in   a  matrix  Ā,   a₄=  [0  1  0  0  0  0]   a₅=  [1  0  -­‐1  0  0  0]     •   a₃  a₂  a  ₁  are  the  rows  corresponding   to  the  constraints  arising  from  the   vanishing  point  generated  by   orthogonal  pairs  of  parallel  lines,   expressed  in  the  way  illustrated  in  the   paper,  and  in  the  generic  algorithm  on   the  right.     Ā  =  [a  ₁;  a₂;    a₃;  a₄;  a₅]       •   To  find  the  null  space  of  Ā     •   apply  svd(Ā)    and  get  the  last   column  of  V   •   Equivalent  to  solve     •   Ā  w  =  0     Metric  Reconstruc-on  (4/5)  
  • 33. Metric  Reconstruc-on  (5/5)   •  The knowledge of w implies the knowledge of ω •  Hence we can compute A (that is not the A previously computed in SVD), such that : §  AA T = (MTωM)- 1 = Z §  where M = P(1:3, 1:3) •  To compute A we use the cholesky decomposition: §  A= chol (Z,’lower’) •  In fact this returns Cholesky matrix A that satisfies AA T = Z •  Using the A value we can build the metric homography such that is possible to upgrade the affine to the metric reconstruction ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ Α = − 1 1 H