Mathematical morphology is concerned with the analysis of shapes
of objects in images. This means e.g. that by
morphological filtering it is possible to simplify structures
in an image depending on their scale, or just to keep structures
up to a selected scale, or to enhance flow-like structures.
The field of mathematical morphology arised in the mid of the
sixties, and today it is one of the most successful
areas in image processing.
There exist two basic processes in mathematical morphology:
dilation and erosion. In a dilation process, at each pixel the
given grey value is changed making use of the maximum of grey values
within a certain pixel mask. This means, light structures are
enhanced, and dark structures vanish. In an erosion process this role
of light and dark is switched. All other filters rely on a
suitable combination of dilation/erosion.
On the modelling side, there are two ways
to realise the basic processes dilation/erosion. The classic way is to
define a mask, or a set of masks, over which the maximum/minimum
is computed. In our group we mainly follow an alternative proceeding
using partial differential equations (PDEs). The arising
PDEs are hyperbolic, and they are variations of so-called
Eikonal equations arising e.g. in physics in the field of
geometric optics.
The research efforts in our group proceed in three
main directions.
-
Establishing a concept of morphology of tensor-valued data,
as arising e.g. in DT-MRI (diffusion tensor magnetic resonance imaging).
-
Construction of high-quality numerical schemes for PDE-based
morphology.
-
Theoretical investigation of the relation between
the scale space concepts of morphological and linear
scale spaces.
Contributions in Morphology for Tensor-valued Data
The processing of matrix-valued data sets (so-called tensor fields)
is becoming an increasingly important task in modern digital imaging
since the tensor concept provides an adequate description of anisotropic
behaviour: Prominent examples are the diffusive behaviour of water molecules
in tissue, visualised by diffusion tensor magnetic resonance imaging (DT-MRI),
or the so-called structure tensor used in fields ranging from motion
analysis to texture segmentation. The tensors that arise in these ways have
real-valued entries and are symmetric. For general
information on filtering such data, see also
this webpage.
-
Tensor-Valued Morphology, Algebraic-Geometric Approaches
Dilation/erosion rely on maximum and minimum operations.
Their generalisation to the tensor-valued setting is not straightforward
due to the lack of suitable ordering relations for symmetric matrices.
In [4] we have proposed such extensions exploiting
analytic-algebraic and geometric properties of symmetric matrices while
preserving positive definiteness of the initial data. Unfortunately the
proposed approaches have their weaknesses when it comes to computational issues.
This led to the development of more appropriate
extensions of dilation and erosion operations for matrix-valued data.
-
Tensor-Valued Morphology, Ordering-Based Approach
In the matrix valued setting there is a partial ordering, the so-called Loewner ordering,
that is suitable for the definition of appropriate maximal and minimal elements of a set of
matrices. The task to determine these extremal elements is already
non-trivial in the case of 2 by 2 matrices [4]
and becomes more intricate
in higher dimensions [5],[6];
it requires tools from convex analysis and computational geometry.
Nevertheless, the dilation and erosion operations inferred from the ordering
approach exhibit the desired rotational invariance,
preservation of positive definiteness, and
continuous dependence on the initial matrix data.
This makes it possible to define morphological
derivatives and shock filters for matrix-fields.
-
The PDE-based Approach
Dilation/erosion can be described by nonlinear hyperbolic first order PDEs as
introduced by R. van den Boomgaard in 1992.
Based on novel mathematical notions for symmetric matrices, in [7]
we were able to find the truly matrix-valued counterpart to these nonlinear
morphological PDEs. For the appropriate numerical treatment of those novel
matrix-valued PDEs we devised matrix-valued generalisations of
known real-valued schemes, see especially [9].
Numerical Contributions
-
Shock-Capturing Schemes for PDEs in Greyscale Morphology
In 1992 R. van den Boomgaard proposed nonlinear
hyperbolic partial differential equations that mimic the dilation and erosion of images
with structuring elements of increasing size. It is quite challenging to
devise numerical schemes that are capable of following a corresponding
evolution of edges adequately, that means, with as little numerical blurring as possible.
In [8] we develop a novel flux corrected transport
(FCT) scheme that can accurately
cope with the peculiarities of the morphological processes of dilation and erosion.
Later on, this method has been extended to deal with general structuring
elements [10]. In the mentioned works, it is shown
that the method performs better than other PDE-based schemes in the field.
Also, the FCT scheme is competitive to set-based, algebraic solutions.
-
Investigating Relations between Fully Discrete and Continuous-Scale Morphology
In 2005, Lerallut et al. have proposed a fully discrete means
to incorporate tonal differences into adaptive morphological filters.
In [13] we validate that such morphological amoebas
are a discretisation of so-called self-snakes, a PDE-based process
somewhat similar to mean curvature motion.
-
PDE-based Methods for Tensor-Valued Morphology
Relying on the developed PDEs for tensor-valued morphology,
we devised the first high-resolution-type
schemes for morphology of tensor fields [9].
Also, the FCT-scheme introduced in [8]
for processing grey value images has been successfully extended
to this setting showing
superior resolution of edge-like structures compared with other methods.
-
PDE-based Anisotropic Morphology
Relying on the developed PDEs for grey-valued morphology as well as for
tensor-valued morphology, we developed the concept
of anisotropic morphology for enhancing flow-like, coherent
image structures. Beginning with the investigation in [12],
we refined the modeling as well as the numerics in
subsequent steps [11], [10].
Contributions in Scale Space Theory
-
Explaining the Logarithmic Relation between Linear and Morphological Systems
In 1994 Dorst/van den Boomgaard and Maragos introduced the
slope transform as the morphological equivalent of the Fourier
transform. As a result an almost logarithmic connection between linear
and morphological systems became apparent.
In [1] we give an explanation for
this relation by revealing that morphology, in essence, is linear
systems theory based on the (max,+)-algebra resp. the (min,+)-algebra.
The fundamental operations of dilation and erosion turn out to be
convolutions with respect to these algebras.
The so-called Cramer transform as the conjugate of the
logarithmic Laplace transform establishes a homeomorphism between
morphological and Gaussian scale space.
These articles provide a step towards the unification of
linear and morphological scale-space theory.
-
Families of Morphological Scale Spaces
In [2] morphological dilation and
erosion scale spaces with parabolic structuring functions
are embedded into two-parameter families of scale spaces which
include the Gaussian scale space as a limit case.
The scale space families are obtained by deforming the algebraic
operations underlying the morphological operations.
Aspects of these scale space families such as
continuity, invariance and separability are discussed.
The article enriches the picture of structural analogies
between the classes of scale spaces mentioned above.
-
B. Burgeth, J. Weickert:
An Explanation for the Logarithmic Connection between
Linear and Morphological Systems.
In L.D. Griffin, M Lillholm (Eds.): Scale Space Methods in Computer Vision.
Lecture Notes in Computer Science, Vol. 2695, Springer, Berlin, 325-339, 2003.
-
M. Welk:
Families of Generalised Morphological Scale Spaces.
In L.D. Griffin, M Lillholm (Eds.): Scale Space Methods in Computer Vision.
Lecture Notes in Computer Science, Vol. 2695, Springer, Berlin, 770-784, 2003.
-
B. Burgeth, M. Welk, C. Feddern, J. Weickert:
Morphological operations on matrix-valued images.
In T. Pajdla, J. Matas (Eds.): Computer Vision - ECCV 2004.
Lecture Notes in Computer Science, Vol. 3024, 155-167, Springer, Berlin, 2004.
-
B. Burgeth, M. Welk, Ch. Feddern, J. Weickert:
Mathematical Morphology on Tensor Data Using the Loewner Ordering.
In J. Weickert, H. Hagen (Eds.):
Visualization and Processing of Tensor Fields.
Mathematics and Visualization, 357-367, Springer, Berlin, 2006.
Revised version of
Technical Report No. 160, Department of Mathematics,
Saarland University, Saarbrücken, Germany, December 2005.
-
B. Burgeth, N. Papenberg, A. Bruhn, M. Welk, C. Feddern, J. Weickert:
Morphology for higher-dimensional tensor data via Loewner ordering.
In C. Ronse, L. Najman, E. Decencière (Eds.):
Mathematical Morphology: 40 Years On. Computational Imaging and Vision, Vol. 30,
Springer, Dordrecht, 407–416, 2005.
-
B. Burgeth, N. Papenberg, A. Bruhn, M. Welk, J. Weickert:
Mathematical Morphology for Tensor Data Induced by the Loewner Ordering in
Higher Dimensions.
Signal Processing, Vol. 87, No. 2, 291-308, February 2007.
Revised version of
Technical Report No. 161, Department of Mathematics,
Saarland University, Saarbrücken, Germany, December 2005.
-
B. Burgeth, A. Bruhn, S. Didas, J. Weickert, M. Welk:
Morphology for Tensor Data: Ordering versus PDE-Based Approach.
Image and Vision Computing, Vol. 25, No. 4, 496-511, 2007.
Revised version of
Technical Report No. 162, Department of Mathematics, Saarland University,
Saarbrücken, Germany, December 2005.
-
M. Breuß, J. Weickert:
A shock-capturing algorithm for the differential equations
of dilation and erosion.
Journal of Mathematical Imaging and Vision, Vol. 25, 187-201, (2006).
Revised version of
Technical Report No. 153, Department of Mathematics,
Saarland University, Saarbrücken, Germany, September 2005.
-
B. Burgeth, M. Breuß, S. Didas, J. Weickert:
PDE-based morphology for matrix fields: Numerical solution schemes.
Tensors in Image Processing and Computer Vision,
Advances in Pattern recognition, pp. 125-150.
Springer, London, 2009.
Revised version of
Technical Report No. 220, Department of Mathematics,
Saarland University, Saarbrücken, Germany, September 2008.
-
M. Breuß, J. Weickert:
Highly accurate PDE-based morphology for general structuring
elements.
In X.-C. Tai et al. (Eds.):
Scale Space and Variational Methods in Computer Vision.
Lecture Notes in Computer Science, Vol. 5567, 758 - 769,
©
Springer-Verlag Berlin Heidelberg 2009.
- L. Pizarro, B. Burgeth, M. Breuß, J. Weickert
A directional Rouy-Tourin scheme for adaptive matrix-valued morphology.
In M.H.F. Wilkinson and J.B.T.M. Roerdink (Eds.): Proc. Ninth
International Symposium on Mathematical Morphology (ISMM 2009),
Lecture Notes in Computer Science,
vol. 5720, pp. 250 -260, Springer, Berlin, 2009.
©
Springer-Verlag Berlin Heidelberg 2009.
-
M. Breuß, B. Burgeth, J. Weickert:
Anisotropic continuous-scale morphology.
In Proceedings of the 3rd Iberian Conference on Pattern Recognition and
Image Analysis, IbPRIA, June 6–8, 2007, Girona, Spain,
Lecture Notes in Computer Science, Springer, Berlin, 2007.
-
M. Welk, M. Breuß, O. Vogel:
Morphological amoebas are self-snakes.
To appear in Journal of Mathematical Imaging and Vision.
Revised version of
Technical Report No. 259, Department of Mathematics,
Saarland University, Saarbrücken, Germany, February 2010.
-
M. Breuß, J. Weickert:
Highly Accurate Schemes for PDE-Based Morphology
with General Convex Structuring Elements.
To appear in International Journal of Computer Vision.
Revised version of
Technical Report No. 236, Department of Mathematics,
Saarland University, Saarbrücken, Germany, May 2009.
-
B. Burgeth, L. Pizarro, M. Breuß, J. Weickert:
Adaptive continuous-scale morphology for matrix fields.
To appear in International Journal of Computer Vision.
Revised version of
Technical Report No. 237, Department of Mathematics,
Saarland University, Saarbrücken, Germany, May 2009.
|