Benjamin Perret

Ecole Supérieure d'Ingénieur en Electronique et Electrotechnique - Paris | ESIEE · Department of Computer Science

Superpixels through Iterative CLEarcutting (SICLE) is a recently proposed framework for superpixel segmentation. SICLE consists of three steps: (i) seed oversampling; (ii) superpixel generation; and (iii) seed removal; such that, after step (i), steps (ii) and (iii) are repeated until a desired number of superpixels is obtained. Such pipeline showe...

In this article, we propose a method for designing loss functions based on component trees that can be optimized by gradient descent algorithms and are therefore usable in conjunction with recent machine learning approaches such as neural networks. The nodes of this tree are the connected components of the upper level sets of an image and the leave...

Binary Partition Hierarchies (BPH) and minimum spanning trees are fundamental data structures involved in hierarchical analysis such as quasi-flat zones or watershed. However, classical BPH construction algorithms require to have the whole data in memory, which prevent the processing of large images that cannot fit entirely in the main memory of th...

Binary Partition Hierarchies (BPH) and minimum spanning trees are fundamental data structures involved in hierarchical analysis such as quasi-flat zones or watershed. However, classical BPH construction algorithms require to have the whole data in memory, which prevent the processing of large images that cannot fit entirely in the main memory of th...

Superpixel segmentation can be used as an intermediary step in many applications, often to improve object delineation and reduce computer workload. However, classical methods do not incorporate information about the desired object. Deep-learning-based approaches consider object information, but their delineation performance depends on data annotati...

The extension of Mathematical Morphology to colour and multivariate images is challenging due to the need to define a total ordering in the colour space. No one general way of ordering multivariate data exists and, therefore, there is no single, definitive way of performing morphological operations on colour images. In this paper, we propose an ext...

In the context of mathematical morphology, component-graphs are complex but powerful structures for multi-band image modeling, processing, and analysis. In this work, we propose a novel multi-band object detection method relying on the component-graphs and statistical hypothesis tests. Our analysis shows that component-graphs are better at capturin...

Binary partition hierarchies and minimum spanning trees are key structures for numerous hierarchical analysis methods, as those involved in computer vision and mathematical morphology. In this article, we consider the problem of their computation in an out-of-core manner, i.e., by minimizing the size of the data structures that are simultaneously n...

This article extends a classical marker-based image segmentation method proposed by Salembier and Garrido in 2000. In the original approach, the segmentation relies on two sets of pixels which play the role of object and background markers. In the proposed extension, the markers are not represented by crisp sets, but by fuzzy ones, i.e., functions...

In this article, we propose a method to design loss functions based on component trees which can be optimized by gradient descent algorithms and which are therefore usable in conjunction with recent machine learning approaches such as neural networks. We show how the altitudes associated to the nodes of such hierarchical image representations can b...

This editorial presents the Special Issue dedicated to the conference ISMM 2019 and summarizes the articles published in this Special Issue.

We study the problem of fitting an ultrametric distance to a dissimilarity graph in the context of hierarchical cluster analysis. Standard hierarchical clustering methods are specified procedurally, rather than in terms of the cost function to be optimized. We aim to overcome this limitation by presenting a general optimization framework for ultram...

Reflectance confocal microscopy (RCM) is a powerful tool to visualize the skin layers at cellular resolution up to a depth of 200µm. A semi‐quantitative score of skin aging from RCM images has been previously published, requiring visual assessment of the images by experienced dermatologists. We have proposed in a previous publication, new computer‐...

Watershed is a well-established clustering and segmentation method. In this article, we aim to achieve a better theoretical understanding of the hierarchical version of the watershed operator. More precisely, we propose a characterization of hierarchical watersheds in the framework of edge-weighted graphs. The proposed characterization leads to an...

The wide literature on graph theory invites numerous problems to be modeled in theframework of graphs. In particular, clustering and segmentation algorithms designed inthis framework can be applied to solve problems in various domains, including imageprocessing, which is the main field of application investigated in this thesis. In thiswork, we foc...

We propose an efficient algorithm that removes unimportant regions from a hierarchical partition tree, while preserving the hierarchical partition structure. Various experiments demonstrate that applying this algorithm on various classification or segmentation problems does indeed improve the results by a large margin. Code is available online at h...

In [19], the authors provide an evaluation of combinations of hierarchical watersheds, showing that some combinations outperform individual hierarchical watersheds. In this article, we aim to achieve a deeper understanding of those combinations. We study which of the four combinations evaluated in [19] always result in flattened (simplified) hierar...

Superpixel computation can be seen as a process of grouping similar pixels trying to preserve image boundaries. In this work, we propose a label propagation method guided by hierarchy of partitions in the context of the marked (supervised) segmentation problem. The main idea of the proposed method is to propagate labels on a tree modelling a hierar...

Higra — Hierarchical Graph Analysis is a C++/Python library for efficient sparse graph analysis with a special focus on hierarchical methods capable of handling large amount of data. The main aspects of hierarchical graph analysis addressed in Higra are the construction of hierarchical representations (agglomerative clustering, mathematical morphol...

The computation of hierarchies of partitions from the watershed transform is a well-established segmentation technique in mathematical morphology. In this article, we introduce the watersheding operator, which maps any hierarchy into a hierarchical watershed. The hierarchical watersheds are the only hierarchies that remain unchanged under the actio...

We study the problem of fitting an ultrametric distance to a dissimilarity graph in the context of hierarchical cluster analysis. Standard hierarchical clustering methods are specified procedurally, rather than in terms of the cost function to be optimized. We aim to overcome this limitation by presenting a general optimization framework for ultram...

The segmentation of the dermal-epidermal junction (DEJ) in in vivo confocal images represents a challenging task due to uncertainty in visual labeling and complex dependencies between skin layers. We propose a method to segment the DEJ surface, which combines random forest classification with spatial regularization based on a three-dimensional cond...

This article introduces a novel region detector based on hierarchies of partitions, so-called Hierarchy-Based Salient Regions (HBSR). This approach enables to combine the clues given by a high quality contour detector with a custom salient region detection procedure. The evaluation of the proposed method HBSR with a standard feature detection asses...

Hierarchical watersheds are obtained by iteratively merging the regions of a watershed segmentation. In the watershed segmentation of an image, each region contains exactly one (local) minimum of the original image. Therefore, the construction of a hierarchical watershed of any image I can be guided by a total order ≺ on the set of minima of I. The...

The Hit-or-Miss Transform (HMT) is a powerful morphological operation that can be utilised in many digital image analysis problems. Its original binary definition and its extension to grey-level images have seen it applied to various template matching and object detection tasks. However, further extending the transform to incorporate colour or mult...

This paper aims to evaluate the morphological hierarchies (observation scale, watersheds area and volume) in the context of marked video segmentation. Here, we automatically created a set of markers for each video taking into account its ground-truth. In order to evaluate the hierarchies, we have applied several types of markers and their combinati...

This article aims to understand the practical features of hierarchies of morphological segmentations, namely the quasi-flat zones hierarchy and watershed hierarchies, and to evaluate their potential in the context of natural image analysis. We propose a novel evaluation framework for hierarchies of partitions designed to capture various aspects of...

The main goal of this paper is to evaluate the potential of some combinations of watershed hierarchies. We also propose a new combination based on merging level sets of hierarchies. Experiments were performed on natural image datasets and were based on evaluating the segmentations extracted from level sets of each hierarchy against the image ground...

Connected operators provide well-established solutions for digital image processing, typically in conjunction with hierarchical schemes. In graph-based frameworks, such operators basically rely on symmetric adjacency relations between pixels. In this article, we introduce a notion of directed connected operators for hierarchical image processing, b...

We propose a quantitative evaluation of morphological hierarchies (quasi-flat zones, constraint connectivity, watersheds, observation scale) in a novel framework based on the marked segmentation problem. We created a set of automatically generated markers for the one object image datasets of Grabcut and Weizmann. In order to evaluate the hierarchie...

emplate matching is a fundamental problem in image analysis and computer vision. It has been addressed very early by Mathematical Morphology, through the well-known Hit-or-Miss Transform. In this chapter, we review most of the existing works on this morphological template matching operator, from the standard case of binary images to the (not so sta...

This article presents a new approach for constructing connected operators for image processing and analysis. It relies on a hierarchical Markovian unsupervised algorithm in order to classify the nodes of the traditional Max-Tree. This approach enables to naturally handle multivariate attributes in a robust non-local way. The technique is demonstrat...

During the last decade, several theories have been proposed in order to extend the notion of set connections in mathematical morphology. These new theories were obtained by generalizing the definition to wider spaces (namely complete lattices) and/or by relaxing some hypothesis. Nevertheless, the links among those different theories are not always...

The goal of this paper is to provide linear or quasi-linear algorithms for producing some of the various trees used in mathemetical morphology, in particular the trees corresponding to hierarchies of watershed cuts and hierarchies of constrained connectivity. A specific binary tree, corresponding to an ordered version of the edges of the minimum sp...

This paper introduces a generalization of self-dual marked flattenings defined in the lattice of mappings. This definition provides a way to associate a self-dual operator to every mapping that decomposes an element into sub-elements (i.e., gives a cover). Contrary to classical flattenings whose definition relies on the complemented structure of th...

In edge-weighted graphs, we provide a unified presentation of a family of popular morphological hierarchies such as component trees, quasi flat zones, binary partition trees, and hierarchical watersheds. For any hierarchy of this family, we show if (and how) it can be obtained from any other element of the family. In this sense, the main contributi...

We present to the astronomical community an algorithm for the detection of low surface brightness (LSB) galaxies in images, called MARSIAA (MARkovian Software for Image Analysis in Astronomy), which is based on multi-scale Markovian modeling. MARSIAA can be applied simultaneously to different bands. It segments an image into a user-defined number o...

Connections in image processing are an important notion that describes how pixels can be grouped together according to their spatial relationships and/or their gray-level values. In recent years, several works were devoted to the development of new theories of connections among which hyperconnection (h-connection) is a very promising notion. This p...

We propose a new class of hyper-connections in order to improve the consistency of hyper-connected filters and to simplify their design. Our idea relies on the principle that the decomposition of an image into h-components must be necessary and sufficient. We propose a set of three equivalent axioms to achieve this goal. We show that an existing h-...

We present a new method for the parametric decomposition of barred spiral galaxies in multispectral observations. The observation is modelled with a realistic image formation model and the galaxy is composed of physically significant parametric structures. The model also includes a parametric filtering to remove non-desirable aspects of the observa...

The NGVS is mapping the Virgo Cluster with a depth making possible to detect very low surface brightness (LSB) structures, such as faint dwarf galaxies. To extract these from just above the sky noise and make statistical studies of their properties, we use the software MARSIAA (MARkovian Software for Image Analysis in Astronomy). This segmentation...

This thesis proposes a morphological and multiband characterization method for galaxies. Galaxies started to evolve and interact since the very beginning of the Universe and they now present a large variety of appearance. Their shapes in the different spectral bands are thus an important clue to understand the history of the Universe. This work dev...

In this paper, we investigate the possibilities offered by the extension of the connected component trees (cc-trees) to multivariate images. We propose a general framework for image processing using the cc-tree based on the lattice theory and we discuss the possible applications depending on the properties of the underlying ordered set. This theore...

Cette thèse propose une méthode de caractérisation morphologique multibande des galaxies. Ces dernières ont commencé à évoluer et à interagir très tôt dans l'histoire de l'Univers: leurs formes dans les différentes parties du spectre électromagnétique représentent donc un traceur important de cette histoire. Ce travail propose une organisation hiér...

The morphological hit-or-miss transform (HMT) is a powerful tool for digital image analysis. Its recent extensions to grey level images have proven its ability to solve various template matching problems. In this paper we explore the capacity of various existing approaches to work in very noisy environments and discuss the generic methods used to i...

Astronomers still lack a multiwavelength analysis scheme for galaxy classification. In this paper we propose a way of analysing multispectral observations aiming at refining existing classifications with spectral information. We propose a global approach which consists of decomposing the galaxy into a parametric model using physically meaningful st...

We have designed a new technique for the detection of Low Surface Brightness galaxies based on local background/source separation using Markovian analysis. This method helps to estimate smooth local variations of the background and therefore allows for determining source candidates as faint as LSB galaxies. For each source an average density profil...

Astronomical object detection is a particularly difficult but very challenging task. Indeed, astronomical images may contain a high noise level due to huge distance within the Universe or to the low photon flow collected on telescope mirrors [1]. Some astronomical objects of interest such as Low Surface Brightness (LSB) galaxies are characterized b...

Nous proposons une méthode d'analyse d'observations multispectrales de galaxies dans le but d'utiliser l'information spectrale pour raffiner les classifications morphologiques existantes. Nous avons opté pour une approche globale qui consiste à décomposer la galaxie en utilisant un modèle paramétrique basé sur une interprétation physique. Nous espé...