Results 1 
5 of
5
An InformationTheoretic External ClusterValidity Measure
 Research Report RJ 10219, IBM
, 2001
"... In this paper we propose a measure of similarity/association between two partitions of a set of objects. Our motivation is the desire to use the measure to characterize the quality or accuracy of clustering algorithms by somehow comparing the clusters they produce with "ground truth" consisting of c ..."
Abstract

Cited by 62 (3 self)
 Add to MetaCart
In this paper we propose a measure of similarity/association between two partitions of a set of objects. Our motivation is the desire to use the measure to characterize the quality or accuracy of clustering algorithms by somehow comparing the clusters they produce with "ground truth" consisting of classes assigned to the patterns by manual means or some other means in whose veracity there is confidence. Such measures are referred to as "external". Our measure also allows clusterings with different numbers of clusters to be compared in a quantitative and principled way. Our evaluation scheme quantitatively measures how useful the cluster labels of the patterns are as predictors of their class labels. When all clusterings to be compared have the same number of clusters, the measure is equivalent to the mutual information between the cluster labels and the class labels. In cases where the numbers of clusters are different, however, it computes the reduction in the number of bits that w...
Scalesets image analysis
 International Journal of Computer Vision
, 2006
"... This paper introduces a multiscale theory of piecewise image modelling, called the scalesets theory, and which can be regarded as a regionoriented scalespace theory. The first part of the paper studies the general structure of a geometrically unbiased regionoriented multiscale image descriptio ..."
Abstract

Cited by 33 (3 self)
 Add to MetaCart
This paper introduces a multiscale theory of piecewise image modelling, called the scalesets theory, and which can be regarded as a regionoriented scalespace theory. The first part of the paper studies the general structure of a geometrically unbiased regionoriented multiscale image description and introduces the scalesets representation, a representation which allows to handle such a description exactly. The second part of the paper deals with the way scalesets image analyses can be built according to an energy minimization principle. We consider a rather general formulation of the partitioning problem which involves minimizing a twotermbased energy, of the form λC+D, where D is a goodnessoffit term and C is a regularization term. We describe the way such energies arise from basic principles of approximate modelling and we relate them to operational rate/distorsion problems involved in lossy compression problems. We then show that an important subset of these energies constitutes a class of multiscale energies in that the minimal cut of a hierarchy gets coarser and coarser as parameter λ increases. This allows us to devise a fast dynamicprogramming procedure to find the complete scalesets representation of this family of minimal cuts. Considering then the construction of the hierarchy from which the minimal cuts are extracted, we end up with an exact and parameterfree
A Minimum Description Length Based Image Segmentation Procedure, and Its Comparison with a CrossValidation Based Segmentation Procedure
, 1999
"... Image segmentation is a very important problem in image analysis, as quite often it is a key component of a good practical solution to a reallife imaging problem. It aims to partition a digital image into a set of nonoverlapping homogeneous regions. One approach to segmenting an image is to fit ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
Image segmentation is a very important problem in image analysis, as quite often it is a key component of a good practical solution to a reallife imaging problem. It aims to partition a digital image into a set of nonoverlapping homogeneous regions. One approach to segmenting an image is to fit a piecewise constant function to the image and define the segmentation by the discontinuity points of the fitted function. The first contribution of this article is to present a new and automatic segmentation procedure which follows this piecewise constant function fitting approach. This procedure is based on Rissanen's minimum description length (MDL) principle and consists of two components: (i) an MDLbased criterion in which the "best" segmentation (i.e., the "best" fitted piecewise constant function) is defined as its minimizer and (ii) a fast merging algorithm which attempts to locate this minimizer. As a second contribution of this article, the new MDLbased procedure is compared wi...
A regionbased matching approach for 3droof reconstruction from hr satellite stereo pairs
 In DICTA
, 2003
"... Abstract. This study is a part of a global project on urban scenes interpretation using high resolution satellite images. Actually, the research is focused on buildings and roads are used to delineate zones of interest. This study details automatic reconstruction of 3D facets using a region matching ..."
Abstract

Cited by 3 (1 self)
 Add to MetaCart
Abstract. This study is a part of a global project on urban scenes interpretation using high resolution satellite images. Actually, the research is focused on buildings and roads are used to delineate zones of interest. This study details automatic reconstruction of 3D facets using a region matching approach based on hierarchical segmentation of images. The novelty consists in dedicating algorithms to satellital context to make them more robust to noise and to deal with low stereopair Base to Height ratio. First of all, a hierarchical segmentation process is explained, then matching regions constraints are detailed. Afterwards, optimal cuts, which corresponds to a set of regions that are likely to represent building rooftops, are processed in both hierarchies. Cuts are processed given matching scores and using regions planarity constraint. In the third part, global matching of both cuts is processed in order to obtain final matchings which will allow 3D scene reconstruction. Eventually, some results are presented.
Segmenting Images Corrupted by Correlated Noise
 IEEE Transactions on Pattern Analysis and Machine Intelligence
, 1998
"... Image segmentation is fundamental to many image analysis problems. It aims to partition a digital image into a set of nonoverlapping homogeneous regions. This paper briefly describes a new segmentation procedure which is designed to segment images corrupted by correlated noise. This procedure is b ..."
Abstract
 Add to MetaCart
Image segmentation is fundamental to many image analysis problems. It aims to partition a digital image into a set of nonoverlapping homogeneous regions. This paper briefly describes a new segmentation procedure which is designed to segment images corrupted by correlated noise. This procedure is based on Rissanen 's minimum description length principle. 1 Introduction One popular approach to segmenting an image based on greyvalue is to approximate the image by a 2D piecewise smooth function and define the segmentation by the discontinuity points of the 2D function. In this paper images that can be well modelled by 2D piecewise constant functions (PCFs) corrupted by additive Gaussian noise will receive primary attention. The main contribution here is the proposal of a new segmentation procedure which is designed to segment such images when the noise is correlated. The proposed segmentation procedure is based on Rissanen's minimum description length (MDL) principle [4] and consists ...