1. A method for image segmentation using a shape prior, comprising the steps of:
(a) applying an initial shape to a first portion of an image;
(b) forming a narrowband around a border of the shape;
(c) calculating a minimum graph cut of a second portion of the image within the narrowband;
(d) adjusting the shape by fitting a result of the minimum graph cut using the shape prior, wherein said shape defines a segmentation result of said image; and
(e) outputting said segmentation result of said image.
2. The method of claim 1, further comprising:
(f) repeating steps (b)-(d) until a stop condition is met prior to step (e).
3. The method of claim 2, wherein said stop condition comprises a specified number of iterations.
4. The method of claim 2, wherein said stop condition comprises convergence of a position of the shape on the image.
5. The method of claim 2, wherein said stop condition comprises convergence of an energy function of the graph of pixels within the narrowband.
6. The method of claim 1, wherein step (a) comprises:
selecting a seed point at said first portion of an image; and
forming said initial shape at said seed point.
7. The method of claim 6, wherein said seed point is selected by one of a user and a computer.
8. The method of claim 6, wherein said initial shape comprises one of a parametric shape having a predetermined size and a mean shape of a statistical shape eigenspace.
9. The method of claim 1, wherein step (b) comprises:
dilating the border of the shape in opposite directions to determine an outer narrowband boundary and an inner narrowband boundary separated by a bandwidth.
10. The method of claim 1, wherein step (c) comprises.
generating a shape mask of the image based on the shape;
calculating a statistical measure of a first intensity of pixels of the image inside the shape and a statistical measure of a second intensity of pixels of the image outside of the shape;
generating a graph of pixels of the image within the narrowband;
calculating energy of the graph based on the shape mask and the statistical measures of the first and second intensities; and
calculating a minimum graph cut of the graph to minimize the energy of the graph.
11. The method of claim 10, wherein said statistical measure of a first intensity comprises a first mean intensity of the pixels inside the shape and said statistical measure of a second intensity comprises a second mean intensity of the pixels outside of the shape.
12. The method of claim 10, wherein said shape mask is a binary image in which pixels inside the shape have a value of 0 and pixels outside of the shape have a value of 1.
13. The method of claim 12, wherein the energy of the graph is calculated using an energy function:
E
=
\u2211
p
\u2208
\u2062
P
\u2062
D
p
\u2061
(
f
p
)
+
\u2211
p
,
q
\u2208
N
\u2062
V
p
,
q
\u2061
(
f
p
,
f
q
)
,
where E is the energy, p and q are adjacent pixels, P is all of the pixels in the graph, N is a neighborhood of pixels, fp is a label that labels a pixel as one of foreground or background, Dp(fp) is a region term that measures a cost of assigning the label fp to pixel p, and Vp,q is a boundary term that measures a cost of assigning labels fp,fq to pixels p and q.
14. The method of claim 13, wherein the region term Dp(fp) of the energy function is expressed as:
Dp(foreground)=e\u2212(I(p)\u2212\u03bci)2(2\u03c32)+\u03bbe\u2212(M(p)\u22121)2(2\u03c32)
Dp(background)=e\u2212(I(p)\u2212\u03bco)2(2\u03c32)+\u03bbe\u2212(M(p)\u22120)2(2\u03c32),
and the boundary term Vp,q of the energy function is expressed by:
Vp,q=e\u2212(I(p)\u2212I(q))2(2\u03c32)+\u03bbe\u2212(M(p)\u2212M(q))2(2\u03c32),
where I represents the intensity of a pixel, M represents the shape mask, \u03bci is a first mean intensity of the pixels inside the shape, \u03bco is a second mean intensity of the pixels outside the shape, \u03c3 is standard deviation of a Gaussian matching function, and \u03bb is a weighting parameter.
15. The method of claim 1, wherein the shape prior comprises a parametric shape.
16. The method of claim 15, wherein said parametric shape comprises an ellipse.
17. The method of claim 1, wherein said shape prior comprises a statistical shape eigenspace determined based on a plurality of training shapes.
18. The method of claim 17, wherein said statistical shape eigenspace comprises a mean shape and one or more modes.
19. A system for graph cut image segmentation using a shape prior, comprising:
means for applying an initial shape to a first portion of an image;
means for forming a narrowband around a border of the shape;
means for calculating a minimum graph cut of a second portion of the image within the narrowband; and
means for adjusting the shape by fitting a result of the minimum graph cut using the shape prior.
20. The system of claim 19, further comprising:
means for determining whether a stop condition is met; and
means for controlling said means for forming, means for calculating, said means for adjusting to iteratively operate until said stop condition is met.
21. The system of claim 19, wherein means for applying comprises:
means for selecting a seed point at said first portion of an image; and
means for forming said initial shape at said seed point.
22. The system of claim 19, wherein said means for forming comprises:
means for dilating the border of the shape in opposite directions to determine an outer narrowband boundary and an inner narrowband boundary separated by a bandwidth.
23. The system of claim 19, wherein said means for calculating comprises:
means for generating a shape mask of the image based on the shape;
means for calculating statistical measures of a first intensity of pixels of the image inside the shape and of a second intensity of pixels of the image outside of the shape;
means for generating a graph of pixels of the image within the narrowband;
means for calculating energy of the graph based on the shape mask and the statistical measures of the first and second intensities; and
means for calculating a minimum graph cut of the graph to minimize the energy of the graph.
24. The system of claim 19, wherein the shape prior comprises one of a parametric shape and a statistical shape eigenspace.
25. The system of claim 24, further comprising:
means for generating said statistical shape eigenspace using one or more training shapes.
26. A computer readable medium storing computer program instructions for performing an image segmentation method using a shape prior, said computer program instructions defining the steps comprising:
(a) applying an initial shape to a first portion of an image;
(b) forming a narrowband around a border of the shape;
(c) calculating a minimum graph cut of a second portion of the image within the narrowband;
(d) adjusting the shape by fitting a result of the minimum graph cut using the shape prior, wherein said shape defines a segmentation result of said image; and
(e) outputting said segmentation result of said image.
27. The computer readable medium of claim 26, further comprising computer program instructions defining the step of:
(f) repeating steps (b)-(d) until a stop condition is met prior to step (e).
28. The computer readable medium of claim 26, wherein the computer program instructions defining step (a) comprise computer program instructions defining the steps of:
selecting a seed point at said first portion of an image; and
forming said initial shape at said seed point.
29. The computer readable medium of claim 26, wherein the computer program instructions defining step (b) comprise computer program instructions defining the step of:
dilating the border of the shape in opposite directions to determine an outer narrowband boundary and an inner narrowband boundary separated by a bandwidth.
30. The computer readable medium of claim 26, wherein the computer program instructions defining step (c) comprise computer program instructions defining the steps of:
generating a shape mask of the image based on the shape;
calculating statistical measures of a first intensity of pixels of the image inside the shape and a second intensity of pixels of the image outside of the shape;
generating a graph of pixels of the image within the narrowband;
calculating energy of the graph based on the shape mask and the statistical measures of the first and second intensities; and
calculating a minimum graph cut of the graph to minimize the energy of the graph.
31. The computer readable medium of claim 26, wherein the shape prior comprises one of a parametric shape and a statistical shape eigenspace.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. A lithographic apparatus comprising a reticle assembly for use in a lithographic process in which a first image field and a second image field are projected onto a first target portion and a second target portion on a substrate, the reticle assembly being arranged to hold a first reticle having the first image field and a second reticle having the second image field such that a distance between the first and second image fields substantially corresponds to a distance between the first and the second target portions, wherein the reticle assembly comprises a holder configured to hold the first reticle and the second reticle, wherein the holder is configured to have the first and second reticles positionable to abut against each other and the holder comprises a set of actuators configured to position one of the first reticle and the second reticle with respect to the other of the first and second reticle so that the distance between the first and the second image field is adjustable.
2. The lithographic apparatus according to claim 1, wherein the reticle assembly is arranged such that the reticle assembly can be used in a single scanning movement of the lithographic process to project subsequently the first image field and the second image field on the first and second target portions.
3. The lithographic apparatus according to claim 1, wherein the reticle assembly comprises a covering element of opaque material arranged to cover a gap between adjacent edges of the first reticle and the second reticle to avoid that light may pass through the gap.
4. The lithographic apparatus according to claim 3, wherein the covering element is arranged to cover a part of at least one of the first and second reticles between the first and second image fields.
5. The lithographic apparatus according to claim 1, wherein the distance between the first image field and the second image field is smaller than 5 mm.
6. The lithographic apparatus according to claim 1, comprising a reticle support configured to hold the holder, wherein the holder is arranged to be loaded on the reticle support while holding the first and second reticle.
7. The lithographic apparatus according to claim 1, comprising an alignment sensor configured to measure an alignment between the first reticle and the second reticle.
8. The lithographic apparatus according to claim 1, wherein a positioning device of the substrate stage is configured to correct for a misalignment between the first reticle and the second reticle.
9. The lithographic apparatus according to claim 1, wherein the second target portion is adjacent the first target portion so that the distance between the first and second target portions is less than a length of the first or second target portion, the reticle assembly and the substrate movable relative to each other so that, in use, the first target portion of the substrate and the adjacent second target portion of the substrate are exposed while the reticle assembly moves in a first scanning direction and a third target portion of the substrate and an adjacent fourth target portion of the substrate are exposed while the reticle assembly moves in a second scanning direction opposite to the first scanning direction, the second target portion and the third target portion being adjacent in a step direction.
10. A method to project two or more image fields on a first target portion and second target portion on a substrate in a lithographic process, comprising:
providing a first reticle having the first image field and a second reticle having the second image field;
placing the first reticle and the second reticle on a same holder of a reticle assembly; and
providing a distance between the first and second image field corresponding to a distance between the first and second target portions,
wherein the holder is configured to have the first and second reticles positionable to abut against each other and the holder comprises a set of actuators configured to position one of the first reticle and the second reticle with respect to the other of the first and second reticle so that the distance between the first and the second image field is adjustable.
11. The method according to claim 10, wherein the second target portion is adjacent the first target portion so that the distance between the first and second target portions is less than a length of the first or second target portion, the reticle assembly including the first and second reticles and the substrate movable relative to each other so that, in use, the first target portion of the substrate and the adjacent second target portion of the substrate are exposed while the reticle assembly moves in a first scanning direction and a third target portion of the substrate and an adjacent fourth target portion of the substrate are exposed while the reticle assembly moves in a second scanning direction opposite to the first scanning direction, the second target portion and the third target portion being adjacent in a step direction.
12. The method according to claim 10, wherein a reticle support is configured to hold the holder, wherein the holder is arranged to be loaded on the reticle support while holding the first and second reticle.