The objective of the AYIN team is to provide image processing tools to aid in the solution of problems in remote sensing and in dermato/cosmetology. From the methodological point of view, the AYIN team is focused on the development of hierarchical and stochastic models for image processing. The two principal applicative axis of the team are:
Analysis of very high resolution images (optical and radar), which includes image quality (denoising, deconvolution and super-resolution), object extraction (extraction of structures, population counting), change detection and temporal tracking in image sequences.
Hyperspectral imagery, which includes modelling of physics of hyperspectral sensors, dimensionality reduction adapted for specific applications, development of spectral-spatial approaches for automatic classification.
Yuliya Tarabalka was recruited as Inria CR2 to the AYIN team in September 2012.
Yuliya Tarabalka received Best Reviewer Award of Transactions on Geoscience and Remote Sensing in July 2012.
A new book was published: Zoltan Kato and Josiane Zerubia. Markov Random Fields in Image Segmentation. Collection Foundation and Trends in Signal Processing. Now editor, World Scientific, 168 pages, September 2012.
A patent on skin care was deposited in collaboration with Galderma and the Morpheme research team in November 2012.
Following a Bayesian methodology as far as possible, probabilistic models are used within the AYIN team for three purposes: to describe the class of images to be expected from any given scene, to describe prior knowledge about the scene and to incorporate specific constraints. The models used in AYIN fall into the following two classes.
Markov random fields were introduced to image processing in the Eighties, and were quickly applied to the full range of inverse problems in computer vision. They owe their popularity to their flexible and intuitive nature, which makes them an ideal modelling tool, and to the existence of standard and easy-to-implement algorithms for their solution. In the AYIN team, attention is focused on their use in image modelling, in particular of textures; on the development of improved prior models for segmentation; and on the lightening of the heavy computational load traditionally associated with these techniques, in particular via the study of varieties of hierarchical random fields.
One of the grand challenges of computer vision and image processing is the expression and use of prior geometric information. For satellite and aerial imagery, this problem has become increasingly important as the increasing resolution of the data results in the necessity to model geometric structures hitherto invisible. One of the most promising approaches to the inclusion of this type of information is stochastic geometry, which is an important line of research in the AYIN team. Instead of defining probabilities for different types of image, probabilities are defined for configurations of an indeterminate number of interacting, parameterized objects located in the image. Such probability distribution are called `marked point processes'. Such processes have been recently applied to skin care problems.
One of the most important problems studied in the AYIN team is how to estimate the parameters that appear in the models. For probabilistic models, the problem is quite easily framed, but is not necessarily easy to solve, particularly in the case when it is necessary to extract simultaneously both the information of interest and the parameters from the data.
Another line of research in the AYIN team concerns development of graph-based, in particular, hierarchical models for very high resolution image analysis and classification. A specific hierarchical model recently developed in AYIN represents an image by a forest structure, where leaf nodes represent image regions at the finest level of partition, while other nodes correspond to image regions at the coarser levels of partitions. The AYIN team is interested in developing multi-feature models of image regions as an ensemble of spectral, texture, geometrical and classification features, and establishing new criteria for comparing image regions. Recent research concerns extension of hierarchical models to a temporal dimension, for analyzing multitemporal data series.
The first application domain of the AYIN team concerns analysis and classification of remote sensing images. The very high spatial, spectral and temporal resolution of the last generation of imaging sensors (for instance, GeoEye, Ikonos, Pleiades, COSMO-SkyMed, TerraSAR-X, ...) provides rich information about environment and is very useful in a range of applications, such as investigating urban environments, precision agriculture, natural disasters and mineralogy. The development of these applications presents new challenges of high-dimensional and high-volume data analysis. The methods proposed by the AYIN team are applied for analysis of SAR, multi- and hyperspectral remote sensing images. In particular, the team develops approaches for image segmentation and classification, change detection, extraction of structures and object tracking.
The second application domain of the AYIN team is skin care imaging which mainly consists in image analysis and classification for dermatology and cosmetology. Here we also deal with very high spatial, spectral and temporal resolution of the most recent imaging sensors. In dermatology we are particularly interested in hyperpigmentation detection and disorders severity evaluation (for instance, for melasma, acne, ...). In cosmetology our main goals are analysis, modeling and characterization of the condition of human skin, as well as evaluation of means to influence that condition. Some of the changes in skin over time have to do with chronological aging (such as pinheads for teenagers or wrinkles for mature people), others with extrinsic aging, caused for instance by sun exposure and smoking.
The software MAD V2.0 was transfered to Galderma R&D in November 2012.
The software Scombo v1.1 was transfered to Cutis laboratory (Galderma R& D, Sophia Antipolis) in May 2012, and to the French-Singaporean laboratory IPAL (Image and Pervasive Access Lab) in November 2012.
The software MAD (Melasma Automatic Detector) V2.0 was deposited with the APP in November 2012. A patent has also been deposited jointly by Galderma R& D and Inria during the same month. It deals with the melasma severity scoring from multi-spectral imaging.
The software Scombo (Supervised Classifier of MultiBand Optical images) v1.1 was deposited with the APP in April 2012. It deals with the supervised classification of multiband optical images by using Markov random fields. It was developed with Aurélie Voisin, Vladimir Krylov and Josiane Zerubia.
This work was done in collaboration with DITEN, University of Genoa, with Dr. Gabriele Moser and Pr. Sebastiano B. Serpico with partial financial support of the French Defense Agency, DGA (http://
high resolution, synthetic aperture radar data, multi-sensor data, urban areas, supervised classification, hierarchical Markov random fields, statistical modeling, wavelets, textural features
The classification of remote sensing images including urban areas is relevant in the context of the management of natural disasters (earthquakes, floodings...), and allows to determine land-use and establish land cover maps, or to localise damaged areas. Given the huge amount and variety of data available nowadays, the main difficulty is to find a classifier that takes into account multi-band, multi-resolution, and possibly multi-sensor data. A minor part of our work was also dedicated to the change detection , still in the frame of the management of natural disasters.
We developed a supervised Bayesian classifier that combines a joint class-conditional statistical modeling and a hierarchical Markov random field. The first classification step deals with the statistical modeling for each target class (e.g. vegetation, urban, etc.) by using a finite mixture model, estimated by resorting to a modified stochastic expectation maximization (SEM) algorithm. Such a model is well-adapted to deal with heterogeneous classes, and each mixture component reflects the contribution of the different materials contained in a given class. When considering optical images, the statistics are modeled by using finite mixtures of Gaussian distributions. In the case of SAR amplitude imagery, we favor a finite mixture of generalized Gamma distributions. Then, at each considered resolution, the different input bands are statistically combined by using multivariate copulas. The second classification step relies on the integration of this statistical modeling in a hierarchical Markov random field integrated in a quad-tree structure. Such contextual classifier helps improving the robustness of the method with respect to noise, or to SAR speckle. A variety of algorithms were proposed to estimate the labels on hierarchical graphs. The consideration of a specific graph, here a quad-tree, allows to benefit from its good properties (e.g. causality) and to apply non iterative algorithms. Among the different algorithms employed in the literature, we chose to take into account an exact estimator of the marginal posterior mode (MPM). The cost function associated to this estimator offers the possibility to penalize the errors according to their number and the scale at which they occur: an error at the coarsest scale is stronger penalized than an error at the finest scale. Moreover, we introduce a prior estimation update that experimentally leads to improved results.
The first experiments were run on single-polarized, mono-resolution synthetic aperture radar (SAR) amplitude images. The challenge of the problem considered here is that the given input is at a single resolution and should be integrated in a multi-scale tree. Thus, we extract an extra information in the form of a multi-scale wavelet decomposition from the initial image. Then, at each level, a textural feature map (e.g. Haralick's variance) is obtained from each image in the decomposition stack, and integrated as an additional information that aims at discriminating the urban areas. Finally, at each level, the wavelet image is combined with the textural image by using copulas, as described previously. These results were presented in , .
Such a classifier is sufficiently flexible to take into account different types of data , . Thus, we also tested coregistered data of a given area acquired at different resolutions (e.g., multiresolution SAR images), directly integrated at the different levels of the hierarchical tree. The classification of multisensor (optical/SAR) data is illustrated in Fig. . In this specific example, we consider a GeoEye acquisition (resolution: 65 centimeters) and a coregistered COSMO-SkyMed SAR acquisition (resolution:
We have also run experiments on other types of acquisitions, such as histological images , to prove the robustness of the proposed algorithm with respect to different image sources.
This program has been partially funded by the Direction of International Relations of Inria (DRI).
Bootstrapping, Expectation-Maximization Algorithm, Iterated Conditional Expectation, Markov Random Fields, Simulated Annealing.
We implemented both Bootstrap EM and Iterated Conditional Expectation algorithms for parameter estimation of first order Markov Random Field models followed by Simulated Annealing, for optimal segmentation of gray-scale images. The objective was to perform a quantitative comparison of the two methods. Apart from successful implementation of these algorithms, an extension of these to multispectral images was performed, and the obtained results were found to be of superior quality compared against the original gray-scale ones (see Fig. ).
First, Bootstrapped EM or Iterated Conditional Expectation were performed based on the user's requirement. The estimated parameters were then used to obtain the optimal segmentation of the image via simulated annealing. The algorithm was extended using multivariate Gaussian models to perform the same for multispectral images.
This work has been conducted in collaboration with ACRI-ST (http://
stochastic modeling, marked point process, object extraction, SEM, simulated annealing
Marked point processes have been successfully applied in image processing analysis, when dealing with high resolution images in the purpose of feature extraction. The objective of this research was to improve the already existing marked point process model of ellipses to better fit the detection of boats in a harbor. The model involved two types of energy terms: a data term, used to determine the fidelity to the existing data (i.e. image) and a prior energy term, used to describe relationships between the objects. We proposed new energy components to model boat detection. The proposed model relied on a high number of parameters. While most of these parameters had an intuitive meaning and could be, thus, set manually, others were difficult to determine. We therefore used a parameter estimation method, based on the Stochastic Expectation - Maximization (SEM) algorithm, which proved to provide good results when combined with marked point processes. Furthermore, we proposed additional automatic procedures based on mathematical morphology to determine critical parameters of this model. Experimental results of boat detection are shown on Fig. .
This work was done in collaboration with Xavier Descombes (Morpheme team, Inria-SAM), Dr. Pierre Coureron and Dr. Christophe Proisy at IRD, UMR AMAP (http://
object detection, marked point processes, tree crowns, forest structure
This work aimed at providing information on the forest structure through the analysis of canopy properties as described by the spatial distribution and the crown size of dominant trees. Our approach was based on the Marked Point Processes (MPP) theory, which allows modeling tree crowns observed in remote sensing images by discs belonging to a two dimensional space. The potential of MPP to detect the trees crowns automatically was evaluated by using very high spatial resolution optical satellite images of both Eucalyptus plantations in Brazil and mangrove forest in French Guyana. LIDAR and simulated reflectance images were also analyzed for the mangrove application. Different adaptations (parameter settings, energy models) of the MPP method were tested and compared through the development of quantitative indices that allowed comparison between detection results and tree references derived from the field, photo-interpretation or the forest mockups.
In the case of mangroves, the estimated crown sizes from detections were consistent with the outputs from the available allometric models (Fig. (Left and Middle)). Other results indicated that tree detection by MPP allowed mapping the local density of trees of young Eucalyptus plantations (Fig. (Right), ) even if crown size was close to the image spatial resolution (0.5 m). However, the quality of detection by MPP decreased with canopy closeness. To improve the results, further work may involve MPP detection using objects with finer shapes and forest data measurements collected at the tree plant scale.
The source images were provided by the AYIN team itself for the study of folliculitis, and by an industrial leader in skin care for acne.
skin care, hyperpigmentation, acne, folliculitis, marked point process, mathematical morphology
Automatic detection of the skin hyperpigmentation helps in estimating the severity of some skin diseases like acne vulgaris and folliculitis. We compared two methods for studying acne and folliculitis lesions and hyperpigmentation of the skin. We adapted a model based on Marked Point Processes and initially developed for flamingo's population counting to dermatological images of acne and folliculitis. Then, we developed an algorithm which uses mathematical morphology together with volume and shadows compensation. Finally, we compared results in term of detection accuracy.
This work is supervised by Florent Lafarge (Geometrica team, Inria-SAM) in collaboration with the AYIN team.
Point processes have demonstrated both efficiency and relevance when addressing object recognition problems in vision. However, simulating these mathematical models is a difficult task, especially on large scenes. Existing samplers suffer from average performances in terms of computation time and stability. We propose a new sampling procedure based on a Monte Carlo formalism. Our algorithm exploits Markovian properties of point processes to perform the sampling in parallel. This procedure is embedded into a data-driven mechanism such that the points are non-uniformly distributed in the scene. The performances of the sampler are analyzed through a set of experiments on various object recognition problems from large scenes, and through comparisons to the existing algorithms.
This work was conducted in collaboration with DITEN, University of Genoa with Dr. Gabriele Moser and Prof. Sebastiano Serpico
(http://
Change detection, synthetic aperture radar, hypothesis test, likelihood ratio test, high resolution
Modern synthetic aperture radar (SAR) sensors represent an essential source of all-weather and 24-hour imagery with a fixed re-visit cycle at competitive high resolution. Two-date change detection from SAR images is a process that employs two SAR images acquired over the same geographical area with possibly the same (or close) acquisition characteristics at two different times to map the areas where changes occur between the two acquisition dates. The central disadvantage of the SAR imagery is given by an inherent multiplicative speckle noise, which restricts the direct application of optical-based change detection methods to SAR imagery.
We have developed a non-parametric statistical change detection approach. We avoided the ambiguity of choosing a restrictive clutter model by assuming no specific probability distribution function model for the statistics of SAR. We developed a modified hypothesis test which is based on the classical Wilcoxon two-sample test that verifies whether one of two samples of independent observations tends to have larger values than the other. The choice of the Wilcoxon statistic as compared to the other available goodness-of-fit test statistics, such as, e.g., that of Cramér-von Mises' test, is a compromise solution to have simultaneously an analytically tractable asymptotic distribution (which is needed to formulate the likelihood ratio test) and a non-parametric testing procedure. Furthermore, the experimental validation demonstrated the adequacy of this statistic to the considered problem. To be able to take a decision at each pixel of the coregistered image pair we considered samples originating from the local windows centered in each pixel. Finally, we constructed a likelihood ratio test on the image with Wilcoxon statistic values. This formulation allowed to overcome the limitation of a classical independency assumption for the Wilcoxon test which is violated (at least, locally) with the local window samples. The resulting technique is related to the statistical false discovery rate approach developed for “large-scale simultaneous hypothesis testing” problems, however the derivation and interpretation are different.
Encouraging detection results were obtained on XSAR and very high resolution COSMO-SkyMed images .
This work was partially funded by a contract with Galderma R&D
(http://
multispectral imaging, skin, hyperpigmentation, hypothesis tests, statistical inferences
One of the steps to evaluate the efficacy of a therapeutic solution is to test it on a clinical trial involving several populations of patients. Each population receives a studied treatment and a reference treatment for the disease.
For facial hyper-pigmentation, a group of
We proposed a methodology to estimate the efficacy a treatment by calculating three differential criteria: the darkness, the area and the homogeneity.
The darkness measures the average intensity of the disease on a gray scaled image
The Fig. illustrates the differential score calculated on a patient whose pathology decreases during the clinical trial. The proposed differential score has been tested in a full clinical study and provided results that agreed with the clinical analysis. This work have been patented, submitted to ISBI'13 conference and to the IEEE TMI journal, and published in Inria research reports , .
This work has been done in collaboration with Dr. Guillaume Charpiat (STARS team, Inria-SAM), Dr. Ludovic Brucker (NASA GSFC, USA) and Dr. James Tilton (NASA GSFC, USA).
hierarchical model, graph cut, segmentation, multiyear sea ice floes, shape analysis
The melting of sea ice is correlated to increases in sea surface temperature and associated climatic changes. Therefore, it is important to investigate how rapidly sea ice floes melt. We proposed two methods for segmentation of a time series of a melting sea ice floe. The first method employs hierarchical model for ice floe segmentation. Image features are extracted using morphological operators, and the floe of interest is marked based on AMSR-E satellite measurements. Then, hierarchical step-wise optimization segmentation is performed, by iteratively merging adjacent regions with the smallest dissimilarity criterion. We proposed to use area and shape parameters of the floe at two previous time moments as priors for computing a segmentation map at the next time moment.
Fig. (a) depicts a graph of the multiyear ice floe area as a function of time, computed by applying the proposed hierarchical model to the summer series of Moderate-Resolution Imaging Spectroradiometer (MODIS) images. While a multiyear ice floe can only melt in the summer period, peaks on the graph correspond to segmentation errors, which are a consequence of either a cloud cover or weakness of contrast between the multiyear ice and the surrounding young ice floes. These segmentation imprecisions can be avoided by simultaneously optimizing all segmentation maps in a time series. For this purpose, we developed a new method based on graph cuts for joint segmentation of monotonously shrinking (or growing) shapes. We impose shape shrinkage (or growth, respectively) constraint in graph cuts, and minimization of energy computed on the resulting graph of the image sequence yields globally optimal segmentation. Fig. (c-d) show examples of floe segmentations using the new approach. Fig. (b) presents a graph of the floe area as a function of time computed by performing the proposed graph cut-based method. The results are compared to those obtained by applying graph cut segmentation to each single image in the considered time series. It can be seen that the new approach yields results with continuous shrinkage of the shape size.
This work has been done in collaboration with Dr. James Tilton (NASA GSFC, USA).
hyperspectral images, classification, segmentation, geometrical features, rectangularity.
The recent advances in hyperspectral remote sensor technology makes it possible to acquire data with a very high spectral (hundreds of spectral channels) and spatial (order of a meter) resolution. The rich spectral information of the hyperspectral data leads to the potential of a more accurate classification, but also presents challenges of high-dimensional data analysis.
We developed a new method for spectral-spatial classification of hyperspectral images. The method is based on the integration of probabilistic classification and shape analysis within the hierarchical step-wise optimization algorithm. First, probabilistic support vector machines classification is applied. Then, at each iteration two neighboring regions with the smallest dissimilarity criterion are merged, and classification probabilities are recomputed. We proposed to estimate a dissimilarity criterion between regions as a function of statistical, classification and geometrical (area and rectangularity) features. Fig. shows the obtained classification results on a 102-band ROSIS image of the Center of Pavia, Italy, which are compared with Support Vector Machines (SVM) classification results. These results did show that the proposed method succeeded in taking advantage of both spatial and spectral information for accurate hyperspectral image classification.
This work has been conducted in collaboration with the French Space Agency CNES (http://
hyperspectral data, HYPXIM, data fusion, panchromatic image, segmentation
Hyperspectral imaging records a detailed spectrum for each pixel, opening new perspectives in classification. Currently, several hyperspectral satellite missions such as EnMAP (210 bands, GSR 30m) are under development. The future hyperspectral satellite missions PRISMA and HYPXIM also include a panchromatic channel with better spatial resolution. We explored if a panchromatic channel at a higher spatial resolution (factor 4) contributes for more accurate classification of hyperspectral images in space conditions.
We adapted and compared several classification methods for combined hyperspectral and panchromatic images, and conducted experiments on the simulated HYPXIM data provided by CNES. We fused both data sources using principal component and Gram-Schmidt fusion methods, as well as the vector stacking approach. We then applied Support Vector Machines (SVM) classification on the resulting feature sets. Furthermore, we considered spatial information for more accurate classification by: (1) including Haralick's texture features in the feature set; (2) segmenting an image into homogeneous regions using a Hierarchical Step-Wise Optimization (HSWO) technique, and assigning each segmented region to the dominant class within this region.
Classification results are illustrated in Fig. . We concluded that classification accuracies of the HYPXIM simulated data have been improved when including a panchromatic channel at a higher spatial resolution into a classification system. These results are close to hyperspectral aerial data classifications. For the image containing one-pixel regions and mixed pixels, standard spectral-spatial classification methods are not well adapted and thus do not improve accuracies when compared to pixelwise classification. In the future, we plan to develop methods which would use both spatial information and a spectral unmixing concept for efficient fusion of hyperspectral and panchromatic data.
This work is done in collaboration with Dr. Ian Jermyn of Durham University (United Kingdom, https://
object detection, shape prior, transformation invariance, higher-order active contours, energy minimization, non-convex energy, exact convex relaxation.
The problem under consideration is the multiple-instance object detection from imagery using prior shape knowledge. As mathematical and algorithmic framework, we have used the higher-order active contour (HOAC) model framework in order to incoporate prior shape knowledge about a class of objects of interest. On top of its robustness and its computational attractiveness (due to its parameter-estimation free method), the HOAC object-detection framework allows to incorporate shape knowledge about multiple occurrences of an object of interest in an image and to carry out object detection in a single algorithmic framework via the minimization of energy of the form:
where
In this work, we have developed a fourth-order active contour (FOAC) framework for incorporating prior shape knowledge about target shapes. Typically, we express a FOAC energy model as
where
We have then shown that shapes with arbitrary geometric complexity can be modeled
using such the FOAC framework , and we have developed a direct method
for the estimation of the parameters for a given class of shapes. In order to be able to detect multiple occurrences
of a target object in an image, one needs to re-express such an originally
contour-based energy by replacing appropriately in formula
the one-dimensional contour quantity
This work is part of LIRA Skin Care Project, which includes four key partners: Philips R&D (Netherlands, http://
image processing, feature extraction, pigmentation distributions, acne, cosmetology
Acne vulgaris is a highly prevalent skin disease, which has a significant life quality impact on sufferers. Although acne severity is readily observed by the human eye, it is an extremely challenging task to relate this visual inspection to measurable quantities of various skin tones and types. So far there is no golden standard for acne diagnosis in clinics, and it entirely depends on dermatologists' experience for evaluation of acne severity. But significant inter-rater variability among individual assessment may lead to less trustworthy diagnosis when several clinicians get involved in the study. In addition, less reproducibility of human evaluation makes comparison of acne changes over time difficult. Therefore, the long-term objective of this study is to construct an automatic acne grading system through applying spectroscopy imaging techniques and image processing methods, to objectively evaluate severity of skin disorder. Such a computer-based tool would also significantly benefit the development of better skin care products, if it can reliably characterize treatment effects of products in individual skin layers in agreement with physiological understanding.
Acne segmentation is normally considered as the first significant step in an automatic acne grading system, because segmentation accuracy directly influences the definition of acne pigmentation level, what has an impact on the goodness of acne severity evaluation. An initial unsupervised segmentation method is proposed for conventional RGB images, whose process is demonstrated in Fig. (a). After several pre-processing steps (background and skin hair removal, illumination corrections), nine pigmentation descriptors were extracted from three RGB channels based on colorimetric transformations and absorption spectroscopy of major chromophores. It has been proved that the derived hemoglobin, normalized red, and normalized green descriptors can properly characterize pigmentation distributions of acne, and they are used as segmentation features. Finally, an iterative unsupervised segmentation was performed to maximize pigmentation distributions between acne and normal skin. Fig. (b) shows an example of acne image on human face captured by a conventional RGB camera, while experimental result in Fig. (f) illustrates that suspicious acne areas and healthy human skin can be automatically discriminated by applying the proposed method. Moreover, it only takes 90.8 seconds to segment the example image with the size of
It should be noted that the segmentation method stated above is an initial approach. Shadows around non-flatten areas on human face (e.g. areas around nose) have a large influence on accuracy of automatic acne detection. However, based on the initial experimental results, it is difficult to entirely get rid of these effects using RGB channels only. Our finding is actually consistent with the existing studies, where researchers divided human face into several sub-regions and worked on these sub-regions individually to avoid shadow influence. Therefore, the next step study will compare acne segmentation results derived from RGB images and multi- or hyperspectral images, to investigate the most effective bands for describing acne pigmentation, as well as whether the introduction of multi- or hyperspectral analysis to the automatic acne detection and evaluation is necessary.
Contribution of multi and hyperspectral imaging to skin pigmentation evaluation. Contract #4383.
Development and validation of multitemporal image analysis methodologies for multirisk monitoring of critical structures and infrastructures. In collaboration with Gabriele Moser and Sebastiano Serpico[PI], from the University of Genoa (DITEN) and the Italian Space Agency (ISA).
Detection of objects in infrared imagery using phase field higher-order active contours. In collaboration with Ian Jermyn from the University of Durham (Dept of Mathematical Sciences).
Automatic object tracking on a sequence of images taken from a geostationary satellite. In collaboration with Pierre Del Moral from Inria Bordeaux (ALEA team) and Ecole Polytechnique (CMAPX) Palaiseau.
Paula Craciun and Josiane Zerubia met Antoine Mangin, Scientific
Director at ACRI-ST (http://
LIRA consortium
Partners: Philips R&D (Eindhoven), CWI (Amsterdam), Fraunhofer Institutes (Berlin, Stuttgart, Darmstadt), Inria-SAM
Skincare image and signal processing: Analysis, modeling and characterization of the condition of human skin
In July, during the visit of Prof.
Qiyin Fang from Mc Master University (http://
Siddharth Buddhiraju (from May 2012 until July 2012)
Subject: Satellite image classification using Bootstrap EM
Institution: IIT Bombay (India)
Paula Craciun (from March 2012 until August 2012)
Subject: Boats detection and counting in Mediterranean harbors
Institution: West University of Timisoara, Romania
Qiyin Fang (One week in September 2012)
Subject: New optical sensors for skin imaging and their biomedical applications
Institution: McMaster University (Canada)
Joseph Francos (One week in March and one week in July 2012)
Subject: Manifold embedding for geometric deformations estimation. Application to both remote sensing and skin imaging
Institution: Ben-Gurion University (Israel)
Ian Jermyn (One week in July 2012)
Subject: Object shape detection in images using prior shape information and higher order active contours
Institution: Durham University (UK)
Zoltan Kato (One week in July 2012)
Subject: Markov random fields for image segmentation
Institution: Sveged University (Hungary)
Nataliya Zagorodna (One month in July 2012)
Subject: Use of periodic or cyclic random processes for image processing, with application to both remote sensing and skin imaging
Institution: Ternopil Ivan Pul'uj Technical University (Ukraine)
Ikhlef Bechar was visiting Dr. Ian Jermyn at Durham University, UK from October 21, 2012 until November 19, 2012.
Yannick Verdie visited National Institute of Informatics (Nii) in Tokyo, Japan from February 15, 2012 to June 15, 2012, funded by Nii internship exchange program. He worked there on the topic of exact sub graph matching by mixed-integer linear problem.
Ikhlef Bechar and Yuliya Tarabalka presented on June 15, 2012 their work for visiting students from SupCom-Tunis to Inria.
Ikhlef Bechar presented his work at Astrium/EADS in Toulouse in January and at “ENVOL de la Recherche” day organized by the EADS foundation in March in Paris.
Vladimir Krylov presented a paper at the conference IEEE ICIP in September in Orlando, USA. He was a reviewer for the journals IEEE TIP, IEEE TGRS, IEEE GRSL, IEEE JSATRS, DSP and PRL.
Yuliya Tarabalka presented 2 papers at the conference IEEE IGARSS'2012 and a paper at the conference UkrObraz'2012. She also presented a poster at the Journées CNES Jeunes Chercheurs in Toulouse in October. She gave 4 seminars at Inria-SAM, CESBIO in Toulouse, University of Orleans and Télécom ParisTech in Paris (France). She was a reviewer for the journals IEEE TIP, IEEE TGRS, IEEE GRSL, Remote Sensing, IEEE JSTARS, and for the conferences ISPRS'2012, IEEE IGARSS'2012 and ICPRAM'2012. She was a part of the program committee for the conferences ISPRS'2012 and ICPRAM'2012. She chaired a session at the conference IEEE IGARSS'2012.
Yannick Verdie presented a paper at the conference ECCV'12 in Florence, Italy in October. He also presented his work at the National Institute of Informatics (Nii) in Tokyo, Japan.
Aurélie Voisin presented 2 papers at the conferences IS&T/SPIE Electronic Imaging'12 in San Francisco (USA) in January, and EUSIPCO'12 in Bucharest (Romania) in August. She presented a poster at Pleiades days organized by CNES in Toulouse in January. She was also invited to present her work by the Centre de Recherche en Automatique de Nancy (France) in June.
Josiane Zerubia chaired a session at IEEE ICIP in September in Orlando, USA, and attended the IVMSP-TC meeting there. She was a reviewer for TS (Traitement du Signal) and SFPT (Revue Française de Photogrammétrie et de Télédétection). She also was a reviewer and/or a program committee member for ICASSP'12, ISBI'12 and ICIP'12, as member of the IEEE BISP TC and IEEE IVMSP TC, and for SPIE-ISPRS'12 (`Image and Signal Processing for Remote Sensing') and ICPRAM'12.
She is an IEEE Fellow.
She was a member of the Biological Image and Signal Processing (BISP) Technical Committee till April 2012 and is a member of the Image, Video and Multidimensional Signal Processing (IVMSP) Technical Committee of the IEEE Signal Processing Society. She is an Associate Editor of the collection `Foundation and Trends in Signal Processing'
[http://
Master: Yuliya Tarabalka, Digital imaging, 15h eq. TD (2h of lectures + 12h of TD), M2 SVS ISAB, Université de Nice Sophia-Antipolis, France
Master et PhD: Yuliya Tarabalka, Computer vision, 8h of lectures = 12h eq. TD, M1 and M2, Ternopil Ivan Pul'uj National Technical University, Ukraine
Licence: Yuliya Tarabalka, Techniques for statistical data analysis, 39h eq. TD, L1, École Polytechnique de l'Université de Nice Sophia-Antipolis, France
Master: Josiane Zerubia, Deconvolution and denoising in confocal microscopy, 18h eq. TD (12h of lectures), M2 IFI BCC, Université de Nice Sophia-Antipolis, France. Furthermore she is also director of this course (total: 24h of lectures).
Master: Josiane Zerubia, Advanced techniques in signal and image processing, 30h eq. TD (20h of lectures), ISAE/SUPAERO, France. Furthermore she is also director of this course (total: 30h of lectures and 12.5h of TD). This course has been given to the third-year students of ISAE/SUPAERO and was also validated by Master 2 of Applied Mathematics of University Paul Sabatier of Toulouse.
Master: Josiane Zerubia, Introduction to image processing, 4.5h eq. TD (3h of lectures), M2 SVS ISAB, Université de Nice Sophia-Antipolis, France. Furthermore she is also director of the course “Digital imaging” at UNS, Master2 SVS ISAB, UE3 (total: 25h of lectures and 25h of TD).
Licence: Jia Zhou, Mathematics, 60h eq. TD, L2, IUT Montpellier, France.
PhD: Aurélie Voisin, Supervised classification of high-resolution remote sensing images including urban areas by using Markovian models, University of Nice-Sophia Antipolis, Defended on October 17th, 2012, Josiane Zerubia.
PhD: Sylvain Prigent, Contribution of multi and hyperspectral imaging to skin pigmentation evaluation, University of Nice-Sophia Antipolis, Defended on November 30th, 2012, Xavier Descombes and Josiane Zerubia.
PhD: Jia Zhou, Object identification on remote sensing images of tropical forest canopies - applications to the study of Eucalyptus plantation and mangrove forest, University of Montpellier 2, Defended on November 16th, 2012, Pierre Couteron and Josiane Zerubia.
PhD in progress: Yannick Verdié, Urban scene analysis from unstructured point data, University of Nice-Sophia Antipolis, started in November 2010, Florent Lafarge and Josiane Zerubia.
PhD in progress: Seong-Gyun Jeong, New image processing methods for skin condition evaluation, University of Nice-Sophia Antipolis, started in December 2012, Josiane Zerubia and Yuliya Tarabalka.
PhD in progress: Paula Craciun, Automatic object tracking on a sequence of images taken from a geostationary satellite, University of Nice-Sophia Antipolis, started in December 2012, Josiane Zerubia and Pierre Del Moral.
Josiane Zerubia was a reviewer of one HdR and of one PhD thesis, and member of 3 PhD committees.
Josiane Zerubia attended a meeting at ICIP'12 in Orlando organized by
IEEE women in engineering
(http://