Image Morphing: 15-463: Computational Photography Alexei Efros, CMU, Fall 2005
Uploaded by
xolraxImage Morphing: 15-463: Computational Photography Alexei Efros, CMU, Fall 2005
Uploaded by
xolraxImage Morphing
Alexey Tikhonov
15-463: Computational Photography Alexei Efros, CMU, Fall 2005
Recovering Transformations
?
y x f(x,y) T(x,y) y x g(x,y)
What if we know f and g and want to recover the transform T?
e.g. better align images from Project 1 willing to let user provide correspondences
How many do we need?
Translation: # correspondences?
?
y x T(x,y) y x
How many correspondences needed for translation? How many Degrees of Freedom? What is the transformation matrix? 1 0 p' x px M = 0 1 p' y p y 1 0 0
Euclidian: # correspondences?
?
y x T(x,y) y x
How many correspondences needed for translation+rotation? How many DOF?
Affine: # correspondences?
?
y x T(x,y) y x
How many correspondences needed for affine? How many DOF?
Projective: # correspondences?
?
y x T(x,y) y x
How many correspondences needed for projective? How many DOF?
Example: warping triangles
B B
?
T(x,y) A
Source
C A
Destination
Given two triangles: ABC and ABC in 2D (12 numbers) Need to find transform T to transfer all pixels from one to the other. What kind of transformation is T? How can we compute the transformation matrix: x' a b c x y ' = d e f y 1 0 0 1 1
HINT: warping triangles
(0,1)
1 1
(0,0)
(1,0)
T2
change of basis
Inverse change of basis
C A
Source
Destination
Dont forget to move the origin too! Very useful for Project 2
Morphing = Object Averaging
The aim is to find an average between two objects
Not an average of two images of objects but an image of the average object! How can we make a smooth transition in time?
Do a weighted average over time t
How do we know what the average object looks like?
We havent a clue! But we can often fake something reasonable
Usually required user/artist input
Averaging Points
Q
Whats the average of P and Q?
v=Q-P
Linear Interpolation (Affine Combination): New point aP + bQ, defined only when a+b = 1 So aP+bQ = aP+(1-a)Q
P + 0.5v = P + 0.5(Q P) = 0.5P + 0.5 Q
P + 1.5v = P + 1.5(Q P) = -0.5P + 1.5 Q (extrapolation)
P and Q can be anything:
points on a plane (2D) or in space (3D) Colors in RGB or HSV (3D) Whole images (m-by-n D) etc.
Idea #1: Cross-Dissolve
Interpolate whole images: Imagehalfway = (1-t)*Image1 + t*image2 This is called cross-dissolve in film industry But what is the images are not aligned?
Idea #2: Align, then cross-disolve
Align first, then cross-dissolve
Alignment using global warp picture still valid
Dog Averaging
What to do?
Cross-dissolve doesnt work Global alignment doesnt work
Cannot be done with a global transformation (e.g. affine)
Any ideas?
Feature matching!
Nose to nose, tail to tail, etc. This is a local (non-parametric) warp
Idea #3: Local warp, then cross-dissolve
Morphing procedure: for every t, 1. Find the average shape (the mean dog)
local warping Cross-dissolve the warped images
2. Find the average color
Local (non-parametric) Image Warping
Need to specify a more detailed warp function
Global warps were functions of a few (2,4,8) parameters Non-parametric warps u(x,y) and v(x,y) can be defined independently for every single location x,y! Once we know vector field u,v we can easily warp each pixel (use backward warping with interpolation)
Image Warping non-parametric
Move control points to specify a spline warp Spline produces a smooth vector field
Warp specification - dense
How can we specify the warp?
Specify corresponding spline control points
interpolate to a complete warping function
But we want to specify only a few points, not a grid
Warp specification - sparse
How can we specify the warp?
Specify corresponding points
interpolate to a complete warping function How do we do it?
How do we go from feature points to pixels?
Triangular Mesh
1. Input correspondences at key feature points 2. Define a triangular mesh over the points
Same mesh in both images! Now we have triangle-to-triangle correspondences
3. Warp each triangle separately from source to destination
How do we warp a triangle? 3 points = affine warp! Just like texture mapping
Triangulations
A triangulation of set of points in the plane is a partition of the convex hull to triangles whose vertices are the points, and do not contain other points. There are an exponential number of triangulations of a point set.
An O(n3) Triangulation Algorithm
Repeat until impossible:
Select two sites. If the edge connecting them does not intersect previous edges, keep it.
Quality Triangulations
Let (T) = (1, 2 ,.., 3t) be the vector of angles in the triangulation T in increasing order. A triangulation T1 will be better than T2 if (T1) > (T2) lexicographically. The Delaunay triangulation is the best
Maximizes smallest angles
good
bad
Improving a Triangulation
In any convex quadrangle, an edge flip is possible. If this flip improves the triangulation locally, it also improves the global triangulation.
If an edge flip improves the triangulation, the first edge is called illegal.
Illegal Edges
Lemma: An edge pq is illegal iff one of its opposite vertices is inside the circle defined by the other three vertices. Proof: By Thales theorem.
Theorem: A Delaunay triangulation does not contain illegal edges. Corollary: A triangle is Delaunay iff the circle through its vertices is empty of other sites. Corollary: The Delaunay triangulation is not unique if more than three sites are co-circular.
Nave Delaunay Algorithm
Start with an arbitrary triangulation. Flip any illegal edge until no more exist. Could take a long time to terminate.
Delaunay Triangulation by Duality
General position assumption: There are no four co-circular points. Draw the dual to the Voronoi diagram by connecting each two neighboring sites in the Voronoi diagram. Corollary: The DT may be constructed in O(nlogn) time. This is what Matlabs delaunay function uses.
Image Morphing
We know how to warp one image into the other, but how do we create a morphing sequence?
1. Create an intermediate shape (by interpolation) 2. Warp both images towards it 3. Cross-dissolve the colors in the newly warped images
Warp interpolation
How do we create an intermediate warp at time t?
Assume t = [0,1] Simple linear interpolation of each feature pair (1-t)*p1+t*p0 for corresponding features p0 and p1
Other Issues
Beware of folding
You are probably trying to do something 3D-ish
Morphing can be generalized into 3D
If you have 3D data, that is!
Extrapolation can sometimes produce interesting effects
Caricatures
Dynamic Scene
Project #2: Due Tu, Sept 27
Given two photos, produce a 60-frame morph animation
Use triangulation-based morphing (lots of helpful Matlab tools) Need to write triangle-to-triangle warp (cant use Matlab tools)
