Poisson Image Editing

Patrick Pérez∗ Michel Gangnet† Andrew Blake‡

Microsoft Research UK

Abstract able effect. Conversely, the second-order variations extracted by

the Laplacian operator are the most significant perceptually.
Using generic interpolation machinery based on solving Poisson Secondly, a scalar function on a bounded domain is uniquely de-
equations, a variety of novel tools are introduced for seamless edit- fined by its values on the boundary and its Laplacian in the interior.
ing of image regions. The first set of tools permits the seamless The Poisson equation therefore has a unique solution and this leads
importation of both opaque and transparent source image regions to a sound algorithm.
into a destination region. The second set is based on similar math- So, given methods for crafting the Laplacian of an unknown
ematical ideas and allows the user to modify the appearance of the function over some domain, and its boundary conditions, the Pois-
image seamlessly, within a selected region. These changes can be son equation can be solved numerically to achieve seamless filling
arranged to affect the texture, the illumination, and the color of ob- of that domain. This can be replicated independently in each of
jects lying in the region, or to make tileable a rectangular selection. the channels of a color image. Solving the Poisson equation also
CR Categories: I.3.3 [Computer Graphics]: Picture/Image has an alternative interpretation as a minimization problem: it com-
Generation—Display algorithms; I.3.6 [Computer Graphics]: putes the function whose gradient is the closest, in the L2 -norm, to
Methodology and Techniques—Interaction techniques; I.4.3 [Im- some prescribed vector field — the guidance vector field — under
age Processing and Computer Vision]: Enhancement—Filtering; given boundary conditions. In that way, the reconstructed function
interpolates the boundary conditions inwards, while following the
Keywords: Interactive image editing, image gradient, guided in- spatial variations of the guidance field as closely as possible. Sec-
terpolation, Poisson equation, seamless cloning, selection editing tion 2 details this guided interpolation.
We will examine a number of possible choices for the guidance
vector field. We show in particular that this interpolation machinery
1 Introduction leverages classic cloning tools, both in terms of ease of use and ca-
pabilities. The resulting cloning allows the user to remove and add
Image editing tasks concern either global changes (color/intensity objects seamlessly. By mixing suitably the gradient of the source
corrections, filters, deformations) or local changes confined to a se- image with that of the destination image, it also becomes possible
lection. Here we are interested in achieving local changes, ones to add transparent objects convincingly. Furthermore, objects with
that are restricted to a region manually selected, in a seamless and complex outlines including holes can be added automatically with-
effortless manner. The extent of the changes ranges from slight dis- out the need for painstaking cutting out. These different cloning
tortions to complete replacement by novel content. Classic tools to facilities are presented in Section 3.
achieve that include image filters confined to a selection, for slight As shown in Section 4, the same machinery can also be used
changes, and interactive cut-and-paste with cloning tools for com- to modify the appearance of an image within a restricted domain,
plete replacements. With these classic tools, changes in the selected while avoiding visible discontinuities on the domain boundary. In
regions result in visible seams, which can be only partly hidden, particular, the color, the texture, or the illumination of an object
subsequently, by feathering along the border of the selected region. can easily be modified without any need for precise delineation of
We propose here a generic machinery from which different tools object boundaries. Also, a rectangular image region can be made
for seamless editing and cloning of a selection region can be de- seamlessly tileable.
rived. The mathematical tool at the heart of the approach is the
Poisson partial differential equation with Dirichlet boundary con-
ditions which specifies the Laplacian of an unknown function over Related work The Poisson equation has been used extensively in
the domain of interest, along with the unknown function values over computer vision. It arises naturally as a necessary condition in the
the boundary of the domain. The motivation is twofold. solution of certain variational problems. In the specific context of
First, it is well known to psychologists [Land and McCann 1971] image editing applications three previous pieces of work are related
that slow gradients of intensity, which are suppressed by the Lapla- to the use of the Poisson equation proposed here.
cian operator, can be superimposed on an image with barely notice- In [Fattal et al. 2002], the gradient field of a High Dynamic
Range (HDR) image is rescaled non-linearly, producing a vector
∗ e-mail: [email protected] field that is no longer a gradient field. A new image is then ob-
† e-mail:[email protected]
tained by solving a Poisson equation with the divergence of this
‡ e-mail:[email protected]
vector field as right-hand-side and under Neumann boundary con-
ditions specifying that the value of the gradient of the new image
in the direction normal to the boundary is zero. In contrast, the
method we are proposing here can be applied to arbitrary patches
selected from an image, not just to the entire image. In order to do
this, Neumann boundary conditions on a rectangular outline must
be replaced by Dirichlet conditions on an arbitrary outline. A fur-
Permission to make digital/hard copy of part of all of this work for personal or ther generalization is to extend the range of nonlinear operations
classroom use is granted without fee provided that the copies are not made or
applied to gradients, to include maximum operations and suppres-
distributed for profit or commercial advantage, the copyright notice, the title of the
publication, and its date appear, and notice is given that copying is by permission sion of small gradients, both of which have useful editing functions.
of ACM, Inc. To copy otherwise, to republish, to post on servers, or to redistribute In [Elder and Goldberg 2001], a system is introduced to edit an
to lists, requires prior specific permission and/or a fee. image via a sparse set of its edge elements (edgels). To suppress an
© 2003 ACM 0730-0301/03/0700-0313 $5.00

313
object, associated edgels are removed; to add an object, associated S f∗
edgels as well as color values on both sides of each of these edgels f
are incorporated. The new image is then obtained by interpolat-
ing smoothly the colors associated to the new set of edgels. This Ω
amounts to solving a Laplace equation (a Poisson equation with a
null right hand side) with Dirichlet boundary conditions given by ∂Ω
colors around edgels. Editing edgels and associated colors is not v g
always simple. In addition, image details are lost when converting
to and from the contour domain, which might be undesirable. The Figure 1: Guided interpolation notations. Unknown function f
sparse edgel-based representation is indeed incomplete, as opposed interpolates in domain Ω the destination function f ∗ , under guid-
to the related representation based on wavelet extrema [Mallat and ance of vector field v, which might be or not the gradient field of a
Zhong 1992], which are complete but less adapted to manual edit- source function g.
ing.

In [Lewis 2001], spots are removed from fur images by separat- 2 Poisson solution to guided interpolation
ing out the brightness component from details in a selected region
and replacing the brightness by harmonic interpolation (solving a Guided Interpolation In this section, we detail image interpola-
Laplace equation) of the brightness at the selection boundary. tion using a guidance vector field. As it is enough to solve the inter-
polation problem for each color component separately, we consider
only scalar image functions. Figure 1 illustrates the notations: let
In terms of image editing functionalities, two existing techniques
S, a closed subset of R2 , be the image definition domain, and let
achieve seamless cloning as the basic instance of our system does.
c Photoshop c 7’s Healing Brush [Adobe c Ω be a closed subset of S with boundary ∂ Ω. Let f ∗ be a known
The first one is Adobe
scalar function defined over S minus the interior of Ω and let f be
2002]. To the best of our knowledge, the technique used by this
an unknown scalar function defined over the interior of Ω. Finally,
tool has not been published. Therefore, we don’t know whether or
let v be a vector field defined over Ω.
not it uses a Poisson solver.
The simplest interpolant f of f ∗ over Ω is the membrane inter-
polant defined as the solution of the minimization problem:
The second technique is the multiresolution image blending pro-
posed in [Burt and Adelson 1983]. The idea is to use a multires-
ZZ
min |∇ f |2 with f |∂ Ω = f ∗ |∂ Ω , (1)
olution representation, namely a Laplacian pyramid, of the images f Ω
of interest. The content of the source image region is mixed, within
each resolution band independently, with its new surrounding in where ∇. = [ ∂∂ x. , ∂∂ y. ] is the gradient operator. The minimizer must
the destination image. The final composite image is then recov- satisfy the associated Euler-Lagrange equation
ered by adding up the different levels of the new composite Lapla-
cian pyramid thus obtained. The technique results in multiresolu- ∆ f = 0 over Ω with f |∂ Ω = f ∗ |∂ Ω , (2)
tion mixing where finest details are averaged very locally around
where ∆. = ∂∂ x2. + ∂∂ y2. is the Laplacian operator. Equation 2 is a
2 2
the boundary of the selection, while lower frequencies are mixed
over much larger distances around these boundaries. This fast tech- Laplace equation with Dirichlet boundary conditions. For image
nique achieves an approximate insertion of the source Laplacian editing applications, this simple method produces an unsatisfactory,
in the destination region (on the first level of the Laplacian pyra- blurred interpolant, and this can be overcome in a variety of ways.
mid) whereas we perform this Laplacian insertion exactly via the One is to use a more complex differential equation as in the “in-
solution of a Poisson equation. More importantly, multiresolution painting” technique of [Bertalmio et al. 2000]. The route proposed
blending incorporates data from distant source and destination pix- here is to modify the problem by introducing further constraints in
els, via the upper levels of the pyramid, within the final composite the form of a guidance field as explained below.
image. This long range mixing, which might be undesirable, does A guidance field is a vector field v used in an extended version
not occur in our technique. In addition, our system offers extended of the minimization problem (1) above:
functionality besides opaque seamless cloning, see Sections 3 and ZZ
4. min |∇ f − v|2 with f |∂ Ω = f ∗ |∂ Ω , (3)
f Ω
Finally, whereas we propose a guided interpolation framework, whose solution is the unique solution of the following Poisson equa-
with the guidance being specified by the user, e.g., in the form of tion with Dirichlet boundary conditions:
a source image in the case of seamless cloning, various interpo-
lation methods have been proposed to fill in image regions auto- ∆ f = divv over Ω, with f |∂ Ω = f ∗ |∂ Ω , (4)
matically using only the knowledge of the boundary conditions.
A first class of such approaches is composed of inpainting tech- where divv = ∂∂ ux + ∂∂ vy is the divergence of v = (u, v). This is the
niques [Ballester et al. 2001; Bertalmio et al. 2000] where PDE- fundamental machinery of Poisson editing of color images: three
based interpolation methods are devised such as to continue the Poisson equations of the form (4) are solved independently in the
isophotes hitting the boundary of the selected region. The PDEs three color channels of the chosen color space. All the results re-
to be solved are more complex than the Poisson equation, and work ported in this paper were obtained in the RGB color space, but sim-
only for bridging fairly narrow gaps in relatively texture-free re- ilar results were obtained in CIE-Lab for instance.
gions. Example-based interpolation methods [Barret and Cheney When the guidance field v is conservative, i.e., it is the gradient
2002; Bornard et al. 2002; Efros and Leung 1999] where the new of some function g, a helpful alternative way of understanding what
image region is synthesized using an arrangement of many small Poisson interpolation does is to define the correction function f˜ on
patches are an interesting alternative to inpainting. These methods Ω such that f = g + f˜. The Poisson equation (4) then becomes the
handle large holes and textured boundaries in a more convincing following Laplace equation with boundary conditions:
way. Moreover, they can also be used to import textures as shown
in [Efros and Freeman 2001; Hertzmann et al. 2001]. ∆ f˜ = 0 over Ω, f˜|∂ Ω = ( f ∗ − g)|∂ Ω . (5)

314
Therefore, inside Ω, the additive correction f˜ is a membrane inter-
polant of the mismatch ( f ∗ − g) between the source and the desti-
nation along the boundary ∂ Ω. This particular instance of guided
interpolation is used for seamless cloning in Section 3.

Discrete Poisson solver The variational problem (3), and the

associated Poisson equation with Dirichlet boundary conditions (4),
can be discretized and solved in a number of ways.
For discrete images the problem can be discretized naturally us-
ing the underlying discrete pixel grid. Without loss of generality,
we will keep the same notations for the continuous objects and their
discrete counterparts: S, Ω now become finite point sets defined on
an infinite discrete grid. Note that S can include all the pixels of an
image or only a subset of them. For each pixel p in S, let N p be
the set of its 4-connected neighbors which are in S, and let hp, qi
denote a pixel pair such that q ∈ N p . The boundary of Ω is now Figure 2: Concealment. By importing seamlessly a piece of the
∂ Ω = {p ∈ S \ Ω : N p ∩ Ω 6= 0}.
/ Let f p be thevalue of f at p. The background, complete objects, parts of objects, and undesirable ar-
task is to compute the set of intensities f |Ω = f p , p ∈ Ω . tifacts can easily be hidden. In both examples, multiple strokes (not
For Dirichlet boundary conditions defined on a boundary of ar- shown) were used.
bitrary shape, it is best to discretize the variational problem (3) di-
rectly, rather than the Poisson equation (4). The finite difference
discretization of (3) yields the following discrete, quadratic opti- As for the numerical implementation, the continuous specifica-
mization problem: tion (9) translates into

min ∑
f |Ω hp,qi∩Ω6=0/
( f p − fq − v pq )2 , with f p = f p∗ , for all p ∈ ∂ Ω, (6) for all hp, qi, v pq = g p − gq , (11)

which is to be plugged into (7).

where v pq is the projection of v( p+q The seamless cloning tool thus obtained ensures the compliance
2 ) on the oriented edge [p, q],
of source and destination boundaries. It can be used to conceal un-
i.e., v pq = v( p+q ) · p
~ q. Its solution satisfies the following simulta-
2 desirable image features or to insert new elements in an image, but
neous linear equations:
with much more flexibility and ease than with conventional cloning,
as illustrated in Figs. 2-4. From the perspective of user input, most
for all p ∈ Ω, |N p | f p − ∑ fq = ∑ fq∗ + ∑ v pq . (7)
tasks will simply require very loose lasso selections, as shown for
q∈N p ∩Ω q∈N p ∩∂ Ω q∈N p
instance in Fig. 3. However, when features of the source have to
be aligned with corresponding features in the destination, as in the
When Ω contains pixels on the border of S, which happens for in-
fence example in Fig. 2 (bottom row) or the face example in Fig.
stance when Ω extends to the edge of the pixel grid, these pixels
4 (top row), the positioning of the source and destination regions
have a truncated neighborhood such that |N p | < 4. Note that for
must be more precise. Finally, in situations where seamless cloning
pixels p interior to Ω, that is, N p ⊂ Ω, there are no boundary terms
involves mostly pieces of texture, as in the face touch-up example
in the right hand side of (7), which reads:
in Fig. 2 (top row) the texture swap example in Fig. 4 (bottom row)
applying repeatedly broad brush strokes is the more effective way.
|N p | f p − ∑ fq = ∑ v pq . (8)
Up to global changes induced by the interpolation process, the
q∈N p q∈N p
full content of the source image is retained . In some circumstances,
Equations (7) form a classical, sparse (banded), symmetric, it is desirable to transfer only part of the source content. The most
positive-definite system. Because of the arbitrary shape of bound- common instance of this problem is the transfer of the intensity
ary ∂ Ω, we must use well-known iterative solvers. Results shown in pattern from the source, not the color. A simple solution is to turn
this paper have been computed using either Gauss-Seidel iteration the source image monochrome beforehand, see Fig. 5.
with successive overrelaxation or V-cycle multigrid. Both methods
are fast enough for interactive editing of medium size color image Mixing gradients With the tool described in the previous sec-
regions, e.g., 0.4 s. per system on a Pentium 4 for a disk-shaped re- tion, no trace of the destination image f ∗ is kept inside Ω. However,
gion of 60,000 pixels. As demonstrated in [Bolz et al. 2003], multi- there are situations where it is desirable to combine properties of f ∗
grid implementation on a GPU will provide a solution for much with those of g, for example to add objects with holes, or partially
larger regions. transparent ones, on top of a textured or cluttered background.
An example is shown in Fig.6, in which a text layer is to be
peeled off the source image and applied to the destination image,
3 Seamless cloning without the need for complex selection operations. One possible
approach is to define the guidance field v as a linear combination
Importing gradients The basic choice for the guidance field v of source and destination gradient fields but this has the effect of
is a gradient field taken directly from a source image. Denoting washing out the textures, see Fig. 6.
by g this source image, the interpolation is performed under the However, the Poisson methodology allows non-conservative
guidance of guidance fields to be used, which gives scope to more compelling
v = ∇g, (9) effect. At each point of Ω, we retain the stronger of the variations
in f ∗ or in g, using the following guidance field:
and (4) now reads
∇ f ∗ (x) if |∇ f ∗ (x)| > |∇g(x)|,

∆ f = ∆g over Ω, with f |∂ Ω = f ∗ |∂ Ω . (10) for all x ∈ Ω, v(x) = (12)
∇g(x) otherwise.

315
 source/destination

sources destinations cloning seamless cloning color transfer monochrome transfer

Figure 5: Monochrome transfer. In some cases, such as tex-

ture transfer, the part of the source color remaining after seamless
cloning might be undesirable. This is fixed by turning the source
image monochrome beforehand.

Figs. 6 and 7.

cloning seamless cloning

sources/destinations

Figure 3: Insertion. The power of the method is fully expressed

when inserting objects with complex outlines into a new back-
ground. Because of the drastic differences between the source and
the destination, standard image cloning cannot be used in this case. (a) color-based cutout and paste (b) seamless cloning

(c) seamless cloning and destination av-

(d) mixed seamless cloning
eraged

Figure 6: Inserting objects with holes. (a) The classic method,

color-based selection and alpha masking might be time consuming
and often leaves an undesirable halo; (b-c) seamless cloning, even
averaged with the original image, is not effective; (d) mixed seam-
less cloning based on a loose selection proves effective.
source/destination cloning seamless cloning

swapped textures
source destination

Figure 4: Feature exchange. Seamless cloning allows the user to

replace easily certain features of one object by alternative features.
In the second example of texture swapping multiple broad strokes
(not shown) were used. Figure 7: Inserting transparent objects. Mixed seamless cloning
facilitates the transfer of partly transparent objects, such as the rain-
bow in this example. The non-linear mixing of gradient fields picks
The discrete counterpart of this guidance field is: out whichever of source or destination structure is the more salient
 ∗ at each location.
f p − fq∗ if | f p∗ − fq∗ | > |g p − gq |,
v pq = (13) This mixed seamless cloning is also useful when adding one ob-
g p − gq otherwise,
ject from a source image very close to another object in the desti-
for all hp, qi. The effect of this guidance field is demonstrated in nation image, see Fig. 8.

316
 source/destination seamless cloning mixed seamless cloning

Figure 8: Inserting one object close to another. With seamless

cloning, an object in the destination image touching the selected
region Ω bleeds into it. Bleeding is inhibited by using mixed gradi- Figure 9: Texture flattening. By retaining only the gradients
ents as the guidance field. at edge locations, before integrating with the Poisson solver, one
washes out the texture of the selected region, giving its contents a
flat aspect.
4 Selection editing
In the two previous sections, the guidance field depended, partly A natural extension is to restrict the correction to a selected re-
or wholly, on the gradient field of a source image g. Alternatively, gion Ω, using appropriate Dirichlet conditions on ∂ Ω. Using a sim-
in-place image transformations can be defined by using a guidance plified version of the Fattal et al. transformation [Fattal et al. 2002],
field depending entirely on the original image. Based on this idea, the guidance field is defined in the log-domain by:
this section details texture flattening, spatially selective illumination
changes, background or foreground color modifications, and seam- v = α β |∇ f ∗ |−β ∇ f ∗ , (16)
less tiling. The first two effects rely on non-linear modifications
of the original gradient field ∇ f ∗ in the selected region. The latter with α = 0.2 times the average gradient norm of f ∗ over Ω, and
effects rely on in-place seamless cloning after the original image β = 0.2. As shown in Fig. 10, this tool can be used for instance
has been modified either inside the domain, providing a new source to correct an under-exposed object of interest, or to reduce specular
image, or outside, providing new boundary conditions. reflections.

Texture flattening The image gradient ∇ f ∗ is passed through a

sparse sieve that retains only the most salient features:

for all x ∈ Ω, v(x) = M(x)∇ f ∗ (x), (14)

where M is a binary mask turned on at a few locations of interest.

A good choice for M is an edge detector, in which case the dis-
crete version of (14), to be plugged into (7), is:

f p − fq if an edge lies between p and q,


v pq = (15)
0 otherwise,

for all hp, qi. As shown in Fig. 9, the content of the selection Ω
gets a flattened appearance, with small grain details washed out,
and the main structure preserved. The extent of this effect depends
obviously on the sparsity of the sieve. The more selective the edge
detector, the sparser the edge map, and the more pronounced the
effect.
Note that this instance of Poisson editing has strong connections
with the contour-domain editing system of Elder and Goldberg [El- Figure 10: Local illumination changes. Applying an appropriate
der and Goldberg 2001]. The difference is that we specify approx- non-linear transformation to the gradient field inside the selection
imately the gradient vectors at edge locations through sparse guid- and then integrating back with a Poisson solver, modifies locally
ance (14), whereas their system relies on an exact specification of the apparent illumination of an image. This is useful to highlight
color values on both sides of each edgel. under-exposed foreground objects or to reduce specular reflections.

Local illumination changes As pointed out by the authors, the

method of [Fattal et al. 2002] is not limited to HDR images and can Local color changes Poisson editing is also a powerful tool for
be applied to ordinary images in order to modify smoothly their dy- manipulating colors. Given an original color image and a selec-
namic range. First, the gradient field of the logarithm of the image tion Ω, two differently colored versions of this image can be mixed
is transformed in order to reduce the large gradients and to increase seamlessly: one version provides the destination function f ∗ out-
the small ones. The transformed vector field v is then used to recon- side Ω, the other one provides the source function g to be modified
struct the logarithm of the image, f , by solving the Poisson equa- within Ω according to (10).
tion ∆ f = divv over the whole image domain under the Neumann For example, the task of turning everything in an image
boundary conditions. monochrome except some object of interest would classically be

317
 image inside the selection, such as texture, illumination, or color.
An important common characteristic of all these tools is that there
is no need for precise object delineation, in contrast with the classic
tools that address similar tasks. This is a valuable feature, whether
one is interested in small touch-up operations or in complex photo-
montages.
Although not illustrated in this paper, it is clear that the cloning
facilities described in Section 3 can be combined with the editing
ones introduced in Section 4. It is for instance possible to insert
an object while flattening its texture to make it match the style of a
texture-free destination.
Finally, it is worth noting that the range of editing facilities de-
Figure 11: Local color changes. Left: original image showing rived in this paper from the same generic framework could prob-
selection Ω surrounding loosely an object of interest; center: back- ably be extended further. Appearance changes could for instance
ground decolorization done by setting g to the original color image also deal with the sharpness of objects of interest, thus allowing the
and f ∗ to the luminance of g; right: recoloring the object of interest user to make apparent changes of focus.
by multiplying the RGB channels of the original image by 1.5, 0.5,
and 0.5 respectively to form the source image. Image credits Two landscapes and swimming bear in Fig. 3,
c
flower in Fig.11: from Corel Professional Photos, copyright 2003
Microsoft Research and its licensors, all rights reserved; rainbow in
Fig. 7 courtesy Professor James B. Kaler, University of Illinois.

References
c c
A DOBE . 2002. Photoshop 7.0 User Guide. Adobe Systems Incorpo-
rated.
BALLESTER , C., B ERTALMIO , M., C ASELLES , V., S APIRO , G., AND
V ERDERA , J. 2001. Filling-in by Joint Interpolation of Vector Fields
Figure 12: Seamless tiling. Setting periodic boundary values on the and Gray Levels. IEEE Trans. Image Processing 10, 8, 1200–1211.
border of a rectangular region before integrating with the Poisson BARRET, A., AND C HENEY, A. 2002. Object-Based Image Editing. ACM
solver yields a tileable image. Transactions on Graphics 21, 3, 777–784.
B ERTALMIO , M., S APIRO , G., C ASELLES , V., AND BALLESTER , C.
2000. Image Inpainting. In Proceedings of ACM SIGGRAPH 2000,
performed by precisely selecting an object and then setting its com- ACM Press / ACM SIGGRAPH, New-York, E. Fiume, Ed., Computer
plement to monochrome. In contrast, Poisson editing frees the user Graphics Proceedings, Annual Conference Series, ACM, 417–424.
from the tedium of precise selection: given a source color image g, B OLZ , J., FARMER , I., G RINPSUN , E., AND S CHR ÖDER , P. 2003. Sparse
(a) the destination image f ∗ is set to be the luminance channel from Matrix Solvers on the GPU: Conjugate Gradients and Multigrid. ACM
g, (b) the user selects a region Ω containing the object, and this Transactions on Graphics. to appear.
may be somewhat bigger than the actual object, and (c) the Pois- B ORNARD , R., L ECAN , E., L ABORELLI , L., AND C HENOT, J.-H. 2002.
son equation (10) is solved in each color channel. An example is Missing Data Correction in Still Images and Image Sequences. In Proc.
presented in Fig. 11. Note that, although the result seems to offer ACM International Conference on Multimedia.
also a precise segmentation of the object for free, this is not actually B URT, P., AND A DELSON , E. 1983. A Multiresolution Spline with Appli-
the case as there is some residual contamination of the destination cation to Image Mosaics. ACM Transactions on Graphics 2, 4, 217–236.
image outside the object. E FROS , A., AND F REEMAN , W. 2001. Image Quilting for Texture Synthe-
Conversely Poisson image editing can be used to modify the sis and Transfer. In Proceedings of ACM SIGGRAPH 2001, ACM Press
color of a loosely selected object. Before solving the Poisson equa- / ACM SIGGRAPH, New-York, E. Fiume, Ed., Computer Graphics Pro-
tion (10), the original image is copied to the destination f ∗ and a ceedings, Annual Conference Series, ACM, 341–346.
version with modified colors is copied to the source g, see Fig. 11. E FROS , A., AND L EUNG , T. 1999. Texture Synthesis by Non-Parametric
Sampling. In Proc. Int. Conf. Computer Vision, 1033–1038.
Seamless tiling When the domain Ω is rectangular, its content E LDER , J., AND G OLDBERG , R. 2001. Image Editing in the Contour
can be made tileable by enforcing periodic boundary conditions Domain. IEEE Trans. Pattern Anal. Machine Intell. 23, 3, 291–296.
with the Poisson solver. The source image g is the original im- FATTAL , R., L ISCHINSKI , D., AND W ERMAN , M. 2002. Gradient Domain
age, and the boundary conditions are derived from the boundary High Dynamic Range Compression. ACM Transactions on Graphics 21,
values of g, such that opposite sides of the rectangular domain cor- 3, 249–256.
respond to identical Dirichlet conditions. In Fig. 12, we have cho- H ERTZMANN , A., JACOBS , C., O LIVER , N., C URLESS , B., AND
∗
sen fnorth ∗
= fsouth = 0.5(gnorth + gsouth ), and similarly for the east S ALESIN , D. 2001. Image Analogies. In Proceedings of SIGGRAPH
and west borders. 2001, ACM Press / ACM SIGGRAPH, New-York, E. Fiume, Ed., Com-
puter Graphics Proceedings, Annual Conference Series, ACM, 327–340.
L AND , E., AND M C C ANN , J. 1971. Ligthness and Retinex Theory. J. Opt.
5 Conclusion Soc. Amer. 61, 1–11.
Using the generic framework of guided interpolation, we have in- L EWIS , J., 2001. Lifting Detail from Darkness. SIGGRAPH 2001 Tech
troduced a variety of tools to edit in a seamless and effortless Sketch.
manner the contents of an image selection. The extent of possi- M ALLAT, S., AND Z HONG , S. 1992. Characterization of Signals from
ble changes ranges from replacement by, or mixing with, another Multi-Scale Edges. IEEE Trans. Pattern Anal. Machine Intell. 14, 710–
source image region, to alterations of some aspects of the original 732.

318