This manuscript which constitutes a synthesis document of my research in preparation for my Habilitation degree (Habilitation à Diriger des Recherches) presents the most important outcomes of my research. Since my PhD degree in September 2005, I have been working on two main research themes which are the visual attention and saliency-based image editing. Before delving into the details of my research, a brief presentation of the visual attention and saliency-based image editing is made.Visual attention: our visual environment contains much more information than we are able to perceive at once. To deal with this large amount of data, human beings have developed biological mechanisms and visual strategies to optimize the visual treatment. Out of those, the visual attention is probably the most important one. It allows to concentrate our biological resources over the most important parts of the visual field. Two kinds of visual attention have been identified: the covert and the overt visual attention. The former does not involve eye movements and refers to the act of mentally focusing on a particular area. The latter, involving eye movements, is used both to explore complex visual scenes and to direct the gaze towards interesting spatial locations. A number of studies have shown that, in most circumstances, overt shifts of attention are mainly associated with the execution of saccadic eye movements. Overt attention of attention is often compared to a windows to the mind. Saccade targeting is indeed influenced by top-down factors (the task at hand, behavioral goals, motivational state) and bottom-up factors (both the local and global spatial properties of the visual scene). The bottom-up mechanism, also called stimulus-driven selection, is the core of my research dealing with the visual attention. It occurs when a target item effortlessly attracts the gaze. My research consists in understanding and modelling this mechanism.Saliency-based image editing: the high-level definition of image editing (as stated by Wikipedia) is the following: image editing encompasses the processes of altering images. These processes refer to color adjustments, histogram manipulation, noise reduction, inpainting, just to name a few. My research focuses on the use of the visual attention into image editing algorithms. As we will see, computational models of visual attention predict the most visually important areas within a scene. From an input picture, these models output a 2D saliency map which is a grey level map where the brighter areas indicate the highest saliency. Saliency-based image editing consists in altering images in function of the saliency map. To illustrate this general idea, an example is the retargeting approach (one of my former work which is not presented in this manuscript). The idea is to adapt automatically traditional contents to the specific constraints of small screen devices in order to provide users with the best possible viewing experience. The saliency map is in this case used to define the cropping window which should enclose as much as possible the saliency. Rather than displaying the whole content, only the content enclosed by the bounding box is displayed.