Docstoc

Computer Vision_ Photometric Stereo

Document Sample
Computer Vision_ Photometric Stereo Powered By Docstoc
					Last lecture
• Passive Stereo
• Spacetime Stereo
• Multiple View Stereo
Today
• Structure from Motion:
   Given pixel correspondences,
   how to compute 3D structure and camera motion?




                                       Slides stolen from Prof Yungyu Chuang
Epipolar geometry &
fundamental matrix
The epipolar geometry




        What if only C,C’,x are known?
The epipolar geometry
                        epipolar geometry demo




         C,C’,x,x’ and X are coplanar
The epipolar geometry




        All points on  project on l and l’
 The epipolar geometry




Family of planes  and lines l and l’ intersect at e and e’
 The epipolar geometry
epipolar pole                  epipolar geometry demo
= intersection of baseline with image plane
= projection of projection center in other image




epipolar plane = plane containing baseline
epipolar line = intersection of epipolar plane with image
  The fundamental matrix F



                               R
                    p                            p’
                C                                 C’
                           T=C’-C

   x  KX                                   x' K' R(X - T)
pK xX-1
                                    p'  K' -1 x'  R (X - T)
The equation of the epipolar plane through X is
                                           
 ( X  T) (T  p)  0               ( R p ' ) (T  p )  0
The fundamental matrix F
                    
          ( R p ' ) (T  p )  0
                         T  p  Sp
                              0         Tz    Ty 
                                                   
                         S   Tz        0      Tx 
                               Ty     Tx      0 
                                                   
          (R p' )  (Sp)  0
              
           (p' R )(Sp)  0
                 
            p' Ep  0                 essential matrix
The fundamental matrix F




                      R
          p                 p’
      C                     C’
                   T=C’-C




               
              p' Ep  0
 The fundamental matrix F
                          
                      p' Ep  0
Let M and M’ be the intrinsic matrices, then
                                          1
           pK x     1
                                   p '  K ' x'

       (K' 1 x' )  E(K 1x)  0
                           1
        x' K' EK x  0
                 
               x' Fx  0              fundamental matrix
The fundamental matrix F




                      R
          p                 p’
      C                     C’
                   T=C’-C

               
              p' Ep  0
                
              x' Fx  0
 The fundamental matrix F
• The fundamental matrix is the algebraic representation
  of epipolar geometry

• The fundamental matrix satisfies the condition that for
  any pair of corresponding points x↔x’ in the two images



                    x' Fx  0
                       T
                                            x'   T
                                                      l'  0   
  The fundamental matrix F

 F is the unique 3x3 rank 2 matrix that satisfies x’TFx=0
     for all x↔x’


1. Transpose: if F is fundamental matrix for (P,P’), then FT
   is fundamental matrix for (P’,P)
2. Epipolar lines: l’=Fx & l=FTx’
3. Epipoles: on all epipolar lines, thus e’TFx=0, x
   e’TF=0, similarly Fe=0
4. F has 7 d.o.f. , i.e. 3x3-1(homogeneous)-1(rank2)
5. F maps from a point x to a line l’=Fx (not invertible)
 The fundamental matrix F




• It can be used for
  – Simplifies matching
  – Allows to detect wrong matches
Estimation of F — 8-point algorithm
• The fundamental matrix F is defined by

                                     
                                 x' Fx  0
     for any pair of matches x and x’ in two images.
                                                               f11     f12    f13 
• Let x=(u,v,1)T and x’=(u’,v’,1)T,                       F   f 21
                                                                       f 22   f 23 
                                                                                    
                                                               f 31
                                                                       f 32   f 33 
                                                                                    
    each match gives a linear equation

uu' f11  vu ' f12  u ' f13  uv' f 21  vv' f 22  v' f 23  uf31  vf 32  f 33  0
 8-point algorithm
                                                       f11 
                                                      f 
                                                       12 
                                                       f13 
 u1u1´ v1u1´ u1´ u1v1´ v1v1´ v1´ u1 v1 1  
u u ´ v u ´ u ´ u v ´ v v ´ v ´ u v 1  f 21 
 2 2     2 2      2    2 2    2 2    2     2    2   f   0
                                           22 
                                                     f 23 
unun ´ vn un ´ un ´ un vn ´ vn vn ´ vn ´ un vn 1  f 
                                                       31 
                                                       f 32 
                                                      f 
                                                       33 
• In reality, instead of solving Af  0, we seek f to
                                          
   minimize Af , least eigenvector of A A     .
 8-point algorithm
• To enforce that F is of rank 2, F is replaced by F’ that
  minimizes F  F' subject to det F'  0 .

                                     
• It is achieved by SVD. Let F  UΣV, where
         1 0    0           1 0 0
   Σ   0  2 0  , let Σ'   0  2 0
                                    
        0
             0 3          0
                                  0 0

  then F'  UΣ' V  is the solution.
8-point algorithm
% Build the constraint matrix
  A = [x2(1,:)‘.*x1(1,:)' x2(1,:)'.*x1(2,:)' x2(1,:)' ...
      x2(2,:)'.*x1(1,:)' x2(2,:)'.*x1(2,:)' x2(2,:)' ...
      x1(1,:)'          x1(2,:)'       ones(npts,1) ];

  [U,D,V] = svd(A);

% Extract fundamental matrix from the column of V
% corresponding to the smallest singular value.
  F = reshape(V(:,9),3,3)';

% Enforce rank2 constraint
  [U,D,V] = svd(F);
  F = U*diag([D(1,1) D(2,2) 0])*V';
8-point algorithm
• Pros: it is linear, easy to implement and fast
• Cons: susceptible to noise
 Problem with 8-point algorithm
                                                                 f11 
                                                                f 
                                                                 12 
                                                                 f13 
 u1u1´ v1u1´ u1´ u1v1´ v1v1´ v1´ u1                   v1    1  
u u ´ v u ´ u ´ u v ´ v v ´ v ´ u                     v2       f 21 
                                                             1
 2 2    2 2     2    2 2     2 2     2     2
                                                                 f 22   0
                                                      
                                                               f 23 
unun ´ vn un ´ un ´ un vn ´ vn vn ´ vn ´ un           vn    1  
                                                                  f 31
 ~10000   ~10000   ~100   ~10000   ~10000   ~100 ~100 ~100   1        
                                                                 f 32 
             Orders of magnitude difference                     f 
             between column of data matrix                       33 
     !
              least-squares yields poor results
   Normalized 8-point algorithm
   normalized least squares yields good results
   Transform image to ~[-1,1]x[-1,1]

(0,500)          (700,500)                        (-1,1)            (1,1)
                              2              
                              700    0     1
                                     2       
                                           1
                                    500      
                                           1
                                                          (0,0)
                                             


(0,0)             (700,0)                         (-1,-1)           (1,-1)
Normalized 8-point algorithm
1. Transform input by xi  Tx i , x'  Tx '
                          ˆ           ˆi  i
                   ˆ ˆ     '
                                       ˆ
2. Call 8-point on x i , x i to obtain F
3. F  T' Τ FTˆ

                      
                   x' Fx  0

                        1
            x' T' FT x  0
            ˆ        ˆ
                      ˆ
                      F
 Normalized 8-point algorithm
[x1, T1] = normalise2dpts(x1);
[x2, T2] = normalise2dpts(x2);
 A = [x2(1,:)‘.*x1(1,:)' x2(1,:)'.*x1(2,:)' x2(1,:)' ...
      x2(2,:)'.*x1(1,:)' x2(2,:)'.*x1(2,:)' x2(2,:)' ...
      x1(1,:)'         x1(2,:)'        ones(npts,1) ];

  [U,D,V] = svd(A);

  F = reshape(V(:,9),3,3)';

  [U,D,V] = svd(F);
  F = U*diag([D(1,1) D(2,2) 0])*V';
% Denormalise
  F = T2'*F*T1;
Normalization
function [newpts, T] = normalise2dpts(pts)

   c = mean(pts(1:2,:)')'; % Centroid
   newp(1,:) = pts(1,:)-c(1); % Shift origin to centroid.
   newp(2,:) = pts(2,:)-c(2);

   meandist = mean(sqrt(newp(1,:).^2 + newp(2,:).^2));
   scale = sqrt(2)/meandist;

   T = [scale   0 -scale*c(1)
         0 scale -scale*c(2)
         0     0      1     ];
   newpts = T*pts;
 RANSAC

repeat
   select minimal sample (8 matches)
   compute solution(s) for F
   determine inliers
until (#inliers,#samples)>95% or too many times


compute F based on all inliers
Results (ground truth)
Results (8-point algorithm)
Results (normalized 8-point algorithm)
From F to R, T

                    x' Fx  0
                               1
          x' M' EM x  0
                        
          E  M' FM                            If we know camera parameters

             E  R[T]
Hartley and Zisserman, Multiple View Geometry, 2nd edition, pp 259
Application: View morphing
Application: View morphing
 Main trick
• Prewarp with a
  homography to rectify
  images
• So that the two views are
  parallel
   • Because linear
     interpolation works when
     views are parallel
Problem with morphing
• Without rectification
                  morph            morph



    prewarp                          prewarp




                          output
          input                            input

homographies
Video demo
Triangulation

• Problem: Given some points in correspondence
  across two or more images (taken from calibrated
  cameras), {(uj,vj)}, compute the 3D location X




Richard Szeliski   CSE 576 (Spring 2005): Computer   40
                               Vision
Triangulation
• Method I: intersect viewing rays in 3D, minimize:



      •   X is the unknown 3D point
      •   Cj is the optical center of camera j
      •   Vj is the viewing ray for pixel (uj,vj)
                                                                   X
      •   sj is unknown distance along Vj
• Advantage: geometrically intuitive                          Vj

                                                              Cj




Richard Szeliski            CSE 576 (Spring 2005): Computer            41
                                        Vision
Triangulation
• Method II: solve linear equations in X
      • advantage: very simple




• Method III: non-linear minimization
      • advantage: most accurate (image plane error)




Richard Szeliski       CSE 576 (Spring 2005): Computer   42
                                   Vision

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:4
posted:8/11/2011
language:English
pages:42