Samy Ait-Aoudia

Remote sensing change detection (RSCD) is the process of identifying changes between scenes of the same location acquired at different times. This is an active research area with a broad range of applications. Many change detection methods have been described in the literature, ranging from simple differencing to machine learning techniques. Faced with this large number of suggested techniques, researchers have tried to classify them into different categories to simplify and systematize the topic. A number of different categorization schemes have been proposed, based on various dimensions, aspects or features of the techniques considered. Nevertheless, using these schemes to understand the different techniques and to select suitable techniques for a specific remote sensing change detection project remains a hard task for practitioners. In this context, we provide a critical study of proposed categorization schemes in order to extract the main dimensions used for their construction. Then, based on the obtained set of dimensions, we propose a new definition for the concept of a remote sensing change detection technique, which provides greater clarity and better guidance for researchers and practitioners. We then expand upon this description to incorporate non-computational aspects of RSCD in order to propose a comprehensive conceptual framework for remote sensing change detection projects.

Change detection (CD) from remote sensing data is a very challenging research problem, especially when we analyse an urban scene. Urban scenes are composed of many different types of objects, both natural and man-made. The building class is one of the important and most complex classes to analyse, important because it is useful for so many applications and complex because it exhibits many changes due to human activity and natural catastrophes. For these reasons, we focus our study on building change detection (BCD). In this paper we propose a classification scheme for BCD research according to several important dimensions including objective, input data, temporal resolution, analysis unit, target output unit, building features, processing technique, change categories, and assessment of results. This classification scheme can guide practitioners in choosing appropriate change detection methods to achieve their goals as well as informing new research efforts. Based on this multidimensional characterisation of BCD, we offer a number of suggestions for further work to be done in this field.

Many routine medical examinations produce images of patients suffering from various pathologies. With the huge number of medical images, the manual analysis and interpretation became a tedious task. Thus, automatic image segmentation became essential for diagnosis assistance. Segmentation consists in dividing the image into homogeneous and significant regions. We focus on hidden Markov random fields referred to as HMRF to model the problem of segmentation. This modelisation leads to a classical function minimisation problem. Broyden–Fletcher–Goldfarb–Shanno algorithm referred to as BFGS is one of the most powerful methods to solve unconstrained optimisation problem. In this paper, we investigate the combination of HMRF and BFGS algorithm to perform the segmentation operation. The proposed method shows very good segmentation results comparing with well-known approaches. The tests are conducted on brain magnetic resonance image databases (BrainWeb and IBSR) largely used to objectively confront the results obtained. The well-known Dice coefficient (DC) was used as similarity metric. The experimental results show that, in many cases, our proposed method approaches the perfect segmentation with a Dice Coefficient above .9. Moreover, it generally outperforms other methods in the tests conducted.

Bases tensorielles de Bernstein et solveurs Tensorial Bernstein bases and solvers