Phase-Based Video Motion Processing
Neal Wadhwa
Michael Rubinstein
Fr do Durand
e
William T. Freeman
MIT Computer Science and Articial Intelligence Lab
t
x
(a) Source
(b) Linear [Wu et al. 2012]
(c) Phase-based (this paper)
Figure 1: Motion magnication of
Image Smoothing via L0 Gradient Minimization
Li Xu
Cewu Lu
Yi Xu
Jiaya Jia
Department of Computer Science and Engineering
The Chinese University of Hong Kong
Figure 1: L0 smoothing accomplished by global small-magnitude gradient removal. Our method suppre
Eulerian Video Magnication for Revealing Subtle Changes in the World
Hao-Yu Wu1
Michael Rubinstein1
1
Eugene Shih2
MIT CSAIL
2
John Guttag1
Fr do Durand1
e
William Freeman1
Quanta Research Cambridge, Inc.
y
time
(a) Input
y
(b) Magnified
time
(c) Spatiote
topics
units for light
units for reection
equations for reection
algorithims for computing lighting
assumptions
light is a eld of photons ying through free space
photons y in straight lines and at a constant speed
no scattering in space
eld is in
Monte Carlo integration
pixel color is a denite integral
hard to do in closed form
approximate integral as weighted sum over samples
I[i][j]
I(x, y)Fi,j (x, y)dxdy
=
I(xk , yk )Fi,j (xk , yk )wi,j,k
k
evenly spaced samples = supersampling
deterministi
photon mapping
variation on these themes, and others
makes very good tradeos
not quite monte carlo
6 terms become 4 terms
split up brdf into specular and diuse part
split up incoming equillibrium light into Le , LS + , L.D.
write an equation of outg
PRT
trades os precomputation+space for time
uses linearity of light transport
uses rotational invariance of s.h. for rigid transforms
linearity
recall
Lt = Le + BLt
Lt BLt = Le
(I B)Lt = Le
Lt = (I B)1 Le
dene the example solutions Lt = (I B)1 Le for
Energy Minimization via Graph Cuts: Settling What is Possible
Daniel Freedman and Petros Drineas
Computer Science Department, Rensselaer Polytechnic Institute, Troy, NY 12180
Abstract
contributions of this paper are the following:
1. What is possible in t
A Closed-Form Solution to Natural Image Matting
The motivation of this paper is matting, which is the process of extracting foreground objects
from a given image. In achieving the extraction task, matting generates a matte. The task is
challenging because