Search

Classification or estimation problems deal with decisions among different hypotheses. In several applications, the set of hypotheses may evolve with time and/or with context. Then, this work focuses on the problem of dynamic estimation and update of the discernment frame in the framework of belief function theory.Belief function theory is widely used in decision systems because of its ability to model both the imprecision and the uncertainty. Now, the problem of the discernment frame estimation is even more critical as the set of handled hypotheses is the power set of the discernment frame.This study describes a solution to update and adjust the discernment frame in a sequential way as new sources provide new pieces of information. Besides incompleteness, it is assumed that the current discernment frame may contain duplicated or spurious hypotheses. We thus propose new update mechanisms and we show on a practical application, namely video surveillance, how these mechanisms may be applied.

International audience; Despite many proposed solutions, multi-object tracking remains a challenging problem in complex situations involving partial occlusions and non-uniform and abrupt illumination changes. Considering modular systems, the tracking performance strongly depends on the consistency of the different blocks relatively to error features. In this work, using the Belief Function framework, we take into account the reliability and the imprecision of the object detection and location to characterize objects and to derive a reliable descriptor. Since this latter is then estimated only on safe object subparts, even in case of crosses between objects, we use a distance between descriptor robust to partial occlusion, namely the recently proposed Bin-Ratio-Distance. Results obtained on various actual sequences underline the interest of the proposed algorithm by outperforming the tested alternative approaches.

Several video surveillance applications aim at detecting the object(s) appeared in the scene of interest. Assuming these objects may be of any kind, change detection algorithms are often used to detect objects or object subparts, that have then to be gathered to form the objects. We call 'object construction' this step of object fragments (elementary change detections) collecting, associating to the right object and fusing with other fragments. In this work, we model the uncertainty and the imprecision of the location of the detected fragments using Belief Function Theory (BFT). During object construction, two mechanisms compete, namely the data accumulation and their temporal removal or weighting. We show that BFT framework is suitable for implementing these mechanisms as well as the data association between the new detections that are unlabeled and the objects under construction. Tests on actual data were performed. They allow for the quantitative evaluation of the proposed method in term of robustness versus the object partial occultation and crossing. Proposed approach is also compared with several alternative approaches.

L’actinomycose est une infection chronique granulomateuse suppurative et fibrosante due a des germes anaerobies gram positif du genre Actinomyces israelii. Les localisations cervicales et thoraciques sont les plus frequentes. Les localisations abdominales sont rares et peuvent simuler une neoplasie posant des difficultes diagnostiques qui peuvent conduire a une intervention chirurgicale mutilante.

Summary Squamous cell carcinoma of the gallbladder is rare, accounting for only 0.5–12.7% of all malignant gallbladder tumors. Tumor progression is rapid but silent and therefore usually discovered at an advanced stage, hence its poor prognosis. We report the observation of a 69-year-old woman with known cholelithiasis, admitted for biliary colic. CT scan highlighted a thick-walled gallbladder harboring a tumor invading segments IV and V of the liver. The patient underwent cholecystectomy associated with bisegmental hepatic resection. Pathology showed a well-differentiated, keratinizing squamous cell carcinoma, infiltrating the entire gallbladder wall and the adjacent hepatic parenchyma.

This study focuses on the problem of object reconstruction through several frames of a video sequence. Elementary detections on which this reconstruction is based are assumed to be fragments of the objects. Belief function framework allows then the modelling of the uncertain and imprecise location of these object fragments within the image. We show that the two competing mechanisms for object reconstruction, namely the data accumulation and their temporal removal or weighting, can be implemented using belief function operators. Results illustrate the robustness of the proposed approach to object partial occultation and crossing.

In video surveillance application, objects e.g intruders should be detected in a reliable way from ‘plot’ detections in the images that are imprecise and uncertain. Belief function theory allows handling both the imprecision and the uncertainty in fusion and decision systems. In this work, specifically, we adress the problem of the dynamic estimation of the discernment frame as new information pieces are provided. Indeed, for our problem, discernment frame represents the set of the reliable objects. Then, we propose mechanisms to dynamically adjust a current discernment frame supposed incomplete and/or containing fictitious and/or duplicated hypotheses. The developed approach is validated on actual data in a video surveillance application.

Several video surveillance applications aim at counting the objects present in a scene. Using robust background substraction techniques, detections are unlabelled and often correspond to fragments of objects. Then, a key step for object counting is the association of the fragments representing subparts of a same object. In this work, we model the uncertainty and the imprecision of the location of the detected fragments using Belief Function Theory. Specifically to the case of a video sequence, we propose a data association method between the new detections and the objects already under construction. Tests on actual data were performed. In particular, they allow for the evaluation of the proposed method in term of robustness versus the objects moving.

International audience; The main focus of this paper is the optical flow computation for 2D movements of objects embedded on 3D surfaces. For spheric-shaped supports, object motion has only two degrees of liberty, thus the 3D optical flow constraint is not relevant. Constancy assumption is formulated using a suitable parametrization of the 3D surface, leading to a 2D equation. Input temporal sequence is also transformed according to the 3D surface parametrization. We build a complete 2D model, taking into account the underlying spherical surface. It has the merit to estimate at a lower cost velocity field in the temporal input sequence. In order to analyze motion computation results, we design an adapted visualization tool, instead of carrying out an inverse transformation for the velocity field. Adapted to the selected parametrization, it displays rapidly moving objects and velocity field and improves the understanding of the displayed information. We display optical flow computation results for 3D+t cell wall simulation sequences.

International audience; The main focus of this paper is 3D+t shape recovery from 3D spatial data and 2D+t temporal sequences. This reconstruction is particularly challenging due to the great deal of in-depth information loss observed on the 2D+t temporal sequence. Our approach embed a geometrical local constraint to handle the critical lack of information. This prior constraint is defined by a spherical topology because several applications may be concerned. It allows us to model relevantly the 3D- to-2D transformation that reduces each 3D image into a 2D frame. We then can build a 3D inaccurate inverse reconstruction of each 2D frame belonging to the video, i.e. 2D+t sequence. These inaccurate 3D images are enhanced by gradual motion compensation using a regularity criterion. Results on synthetic data are displayed.

International audience; In this article, we put forth the contribution of cartographic map projection in image analysis for 2D and 3D spheric-shaped data. In the 2D case, map projection of the underlying hemispherical surface has the merit to overcome deformations generated by perspective sights. In the 3D case, processing cartographic references of the original spheroid surfaces presents mainly two advantages. On one hand, it decreases significantly the amount of processed data. On the other hand, we avoid extension of exhaustive processes to a higher dimension. We make use of this geographical transformation in several classical image processing applications such as temporal tracking, information visualization, and feature extraction, such as segmentation. These applications are described and results of cell wall simulations are displayed.