Proceedings of the Singaporean-French Ipal Symposium 2009

The following sections are included:

• PREFACE

• Organizing Committees

• CONTENTS

Synchronous programming languages like Esterel are widely used in safety-critical domains like avionics. However, it is only with the recent development of mature worst-case execution time (WCET) analysis tools that progress is being made on systematically studying the WCET analysis problem for languages like Esterel. In this context, we present techniques for methodically classifying and detecting different types of infeasible paths that arise while compiling Esterel programs into executable code, via high-level languages such as C. Our experimental results with well-known benchmarks show that the infeasible paths detected using our techniques result in as much as 36.5% reduction in the WCET estimates, compared to when no infeasible path detection is employed.

This paper proposes a new method for the protection of copyrighted multimedia content. Here the problems of compression and selective encryption (SE) have been simultaneously addressed for the state of the art video codec H.264/AVC. SE is performed in the context adaptive variable length coding (CAVLC) module of video codec. For this purpose, CAVLC is converted to an encryption cipher using permutation of equal length codes from a specific variable length coding (VLC) table. In our scheme, entropy coding engine serves the purpose of encryption step without affecting the coding efficiency of H.264/AVC by keeping the bitrate unchanged and generating completely compliant bit stream. Our scheme requires a negligible computational power. Nine different benchmark video sequences containing different combinations of motion, texture and objects are used for experimental evaluation of the proposed algorithm.

We present a scalable encoding strategy for the 3D facial data in various bandwidth scenarios. The scalability, needed to cater diverse clients, is achieved through the multiresolution characteristic of JPEG2000. The disparate 3D facial data is synchronously unified by the application of data hiding wherein the 2.5D facial model is embedded in the corresponding 2D texture in the discrete wavelet transform (DWT) domain. The unified file conforms to the JPEG2000 standard and thus no novel format is introduced. The method is effective and has the potential to be applied in videosurveillance and videoconference applications.

Content-Based Image Retrieval (CBIR) systems for diagnosis support, requires not only new tools but also new concepts and new approaches. Taking into account the domain knowledge is very challenging because of the particular importance of every detail surrounding any given topic, user, task and user's query. However, tackling domain knowledge in an abstract way is not sufficient: it is important also to tackle the context of use of this domain knowledge, and modeling context becomes the key problem. This problem has been encountered in artificial intelligence and fixed with the development of a conceptual framework for context modeling and the design and development of a context-based formalism of representation called "Contextual Graphs". This paper presents a preliminary modeling of breast cancer diagnosis in Contextual Graphs. We reproduce the modeling realized by Logan-Young and Hoffman (1995) in a context-based representation, and we discuss extensions offered by such a representation as an incremental enrichment by new knowledge and learning of new practices by the system, a presentation at different granularity, and the generation of various explanations.

The early screening and diagnosis is one of the most critical steps in the curing of skin cancers and mainly melanomas. We present here a complete platform which deals from screening to therapy via diagnosis in a collaborative manner. The screening is made with a computing collaborative platform consisting of a photograph of the incriminated skin region, its sending to a server devoted to image processing as an aid to the dermatologist who will examine the photograph and the results given by image processing. If there is a doubt, the patient will be proposed to go at a dermatologist's office for diagnosing the eventual cancer by using an optical probe developed in our laboratory. Then, in case of a cancer, the dermatologist will be able to cure the cancer by phototherapy by using the same optical probe.

Networked virtual environments have become increasingly popular in recent years, driven by online games and social applications such as World of War-craft and Second Life. Due to constraints in computing power, storage and bandwidth capacity, however, these networked virtual environments are not sufficiently realistic. In this position paper, we explore some research challenges in streaming of large-scale, realistic 3D scenes (natural and synthetic objects). We focus on progressive, multi-resolution representations of realistic 3D objects (e.g. plants) that are suitable for streaming. We give an overview of our recent research results in this area, and highlight possible research problems to tackle in the future.

This article discusses how to compare different image segmentation algorithms parameters in order to choose the most optimal algorithm parameters for a specific task(s) (e.g. feature extraction, spatial reasoning, topological analysis). Our method of comparison lets the user decide which segmentation algorithm/parameters are the most suitable for his system, this decision is obtained according to three indicators: the amount of relevant visual information as well as the noise in image regions, the average number of regions per images and the average number of regions per object.

In this paper, we describe an attempt to use a graph language modeling approach for the retrieval of images. Because photographic images are 2D data, we consider taking into account this specificity by representing relationships between image regions. Such a representation implies to use complex graph representations. Traditional sub-graph matching methods during retrieval do not scale-up, and language models have been quite successful for years now for text retrieval. That is why we propose in this paper to define an extension of language models for graphs, and to evaluate the interest of this approach on image indexing and retrieval. The results obtained show the potential interest by considering images or groups of images as queries or results.

In dermatology, the optical coherence tomography (OCT) is used to visualize the skin over few millimeters depth. These images are affected by speckle, which can alter their interpretation, but which also carries information that characterizes locally the visualized tissue. In this paper, we propose to differentiate the skin layers by modeling locally the speckle in OCT images. The performances of four probability density functions (Rayleigh, Lognormal, Nakagami and Generalized Gamma) to model the distribution of speckle in each skin layer are analyzed. From this study, we propose to classify the pixels of OCT images using the estimated parameters of the most appropriate distribution. Quantitative results with 30 images are compared to the manual delineations of five experts. The results confirm the potential of the method to generate useful data for robust segmentation.

The approximate joint diagonalization (AJD) is an important analytic tool at the base of numerous independent component analysis (ICA) and other blind source separation (BSS) methods, thus finding more and more applications in medical imaging analysis. In this work we present a new AJD algorithm named SDIAG (Spheric Diagonalization). It imposes no constraint either on the input matrices or on the joint diagonalizer to be estimated, thus it is very general. Whereas it is well grounded on the classical least-squares criterion, a new normalization reveals a very simple form of the solution matrix. Numerical simulations shown that the algorithm, named SDIAG (spheric diagonalization), behaves well as compared to state-of-the art AJD algorithms.

In the domain of radiation protection it is not always possible to perform an additional examination such as scanners or Magnetic Resonance Imaging (MRI) for a patient who has been accidentally radiated. However a medical diagnostic must be made as soon as possible to calculate the dosimetric balance. Currently this incidental calculation is based on the available voxelised phantoms which are 3D numerical reconstructions of the human body with the internal organs. The Case-Based Reasoning (CBR) is seen on the one hand like a problem solving method and on the other hand like a technology for the conception of intelligent systems. The ReEPh project (Research of Equivalent Phantom) strikes a new path in the field of problem solving methods in the radiation protection and uses the approach of the CBR to retrieve the set of the phantoms the most adapted to the irradiated victim. For this first version of ReEPh, the retrieval phase uses a Knn Algorithm (K nearest neighbours). We propose a measure of similarity and a confidence index to take into account the uncertainty implied by the possible missing characteristics of the victim. We have developed a graphic interface to view the cases retrieval, and to visually illustrate the combination of measures of similarity and confidence index.

In this paper, we devise a system that is able to take the mobile experience for users to a whole new level, through what has been coined as the Mobile Tourguide System. The system allows a user to find places of interest (e.g., cinemas, restaurants, etc.) using a camera-equipped mobile phone. Several challenges are addressed in the paper, which some more notable ones are the formulation of database for efficient retrieval of images, accuracy in wide-baseline matching and pose recovery. The system is validated by experiments in our campus, with images taken in different viewpoints, resolutions, weather and illumination conditions.

This work focuses on the viability of using a cell-phone as mobile museum guidance. The integrated cell-phone camera is used to recognize the paintings in the gallery.

The chosen solution is based on a client-server architecture and the object recognition is based on local features. The study focuses on the comparison, in terms of time and performance, of the Scale-Invariant Feature Transform (SIFT), the Speeded Up Robust Features (SURF), the Nearest Neighbor Search (NNS) match and a k-means trees based search. It was found that SIFT outperforms SURF in terms of performance but is dominated in terms of time. Finally, the combination of SIFT and k-means based search provides a good compromise for the low-resolution images necessary in this setup.

The study was performed using a windows mobile operated cell-phone and the 200 test images were taken on site from 4 different perspectives. The reference data set consisted of 1002 different art works of the Louvre.

The novel mobile scenarios encourage the design and development of context-aware middleware that provides on demand appropriate semantic services services to mobile user's access in a dynamic smart environment, since wireless hotspots and mobile devices are permeating our workplace, home and public places. As mobile users will be immersed in thousands of different kinds of pervasive services, it's of paramount importance to provide appropriate personalized services to the right person in the right form. In this paper we present a collaborative-based service provisioning approach which can enable effective service provision based on social semantic characteristics of user profile, device capability, situation context and services in WLAN enabled environment. We illustrate our ideas through a scenario of service discovery and provision in unfamiliar public places (e.g. airport), and present an OSGi-based middleware to instantiate the service provisioning process based on semantic matchmaking.

The following section are included:

• AUTHOR INDEX