In the last decade, several image and video quality metrics have been proposed, which incorporate perceptual quality measures by considering the HVS characteristics. All this metrics do the differences pixel to pixel in image. Therefore a local fidelity of the colour is defined. However, the human visual system is rather sensitive to a global quality. In this paper, we propose a new objective full reference quality metric for colour images called Global Delta E and noted: ~ G DE . This metric is based on human visual system properties in order to obtain the best correspondence with judgments. Some experimentations and assessments prove the performance of our metrics and that interrelationship with the SVH.
In this work we model the interaction signature concepts in a consistent and compact manner as well as their related type checking rules. First, we begin by literally analyzing those concepts in order to bring unambiguous definitions out of them. Following this analysis we shall formalize those concepts by mapping them into UML language constructs. Secondly, we specify constraints imposed on computational interfaces interaction signatures related to the computational language typing and subtyping rules. We shall show how we can we literally redefine those rules in order to steadily formalize them. After rewriting those rules in a compact way, we make use of OCL 2.0 which provides the means to exploit those new definitions. Then we introduce the concept of Functional computational interface and a set of related concepts which unify signal and operation interfaces notions. Based on the new additional concepts introduced, we introduce two new important concepts, namely; QoS-definable interactions and QoS-capable interfaces. We then provide a UML metamodel of interfaces and interaction signatures. The final metamodel being a first step towards a QoS-capable computational metamodel. Finally, as an application of our modeling choices we define ODP QoScapable computational interfaces type checking rules and then specify them using OCL 2.0
The aim of this work is to reconstitute the state of a discrete-time nonlinear system representing a dynamical model of a harvested fish population. For this end, we are going to use a numerical method of building an interval observer for the consider discrete-time model fish population. We adapt to this model an algorithm called "Interval Moving Horizon State Estimation" (IMHSE) which gives an estimated interval of the system states. This algorithm is carried out in [8] and work well for a general class of discrete-time systems.
The component based approach and advanced separation of concerns are two important paradigms for software systems development. Although the two paradigms are complementary and looking for their synergy is a promising issue, only few research works are currently dedicated to their combination. This paper presents a comparative state of the art of the main research works that aim at the synergy of the two paradigms and proposes a new approach of combination that derives from the fact that all potential contributions that can be drawn, are tightly related with the well manipulation of aspects.
We propose in this paper an approach for document clustering. It consists of representing the corpus as a document graph, where the links are defined by some criteria. These links are quantified by simialrity measures. We aim join this context into the approach of classification to constitute small-worlds networks of homogeneous documents. The homogeneity of the clusters is measured according to the properties of small worlds. The clusters, as well as their proprietes, allow to rerank search results. Some experiments were done on a corpus provided by TREC and the obtained results show the contribution of small-worlds networks in information retrieval.
The introduction of high-performance applications such as multimedia applications into embedded systems led the manufacturers to offer embedded platforms able to offer an important computing power which makes it possible to answer the increasing requirements of future evolutions of these applications. One of the adopted solutions is the use of multiprocessor platforms. In this paper we propose an exploratory methodology of scheduling software modelled in the form of hierarchical, collaborative and interdependent hierarchical tasks with local deadlines. Scheduling answers (1) the local temporal requirements of tasks, and allows (2) an effective exploitation of multiprocessor platforms
The Fractal image compression has the advantage of presenting fast decoding and independent resolution but it suffers of slow encoding phase. In the present study, we propose to reduce the computational complexity by using two domain pools instead of one domain pool and encoding an image in two steps (AP2D approach). AP2D could be applied to classification methods or domain pool reduction methods leading to more reduction in encoding phase. Indeed, experimental results showed that AP2D speed up the encoding time. The time reduction obtained reached a percentage of more than 65% when AP2D was applied to Fisher classification and more than 72% when AP2D was applied to exhaustive search. The image quality was not altered by this approach while the compression ratio was slightly enhanced.
The work presented here aims to provide a composition model of semantic web services. This model is based on a semantic representation of domain concepts handled by web services, namely, operations and the static concepts used to describe static properties of Web services. Different levels of abstraction are given to the concept of operation to allow gradual access to concret services. Thus, two different levels of the composition plan are generated (abstract and concret). This will reuse plans already constructed to meet similar needs even with modified preferences.
