A profusion of methods exist to evaluate image quality, often issued from the toolbox of the video industry. Signal processing modelling, information theoretic approaches, conformation with psycho-visual performances of human visual system, are among the useful theories to address the problem. In the last years a plethora of studies have adopted the machine-learning approach to gain access, beyond the objective quality criteria, to some subjective evaluation of the image attractiveness. These works reopen the door to two-thousand year old study about Aesthetics. However beside the contribution of philosophical authorities, this topic is also concerned by the most recent studies of neurophysiologists who suggest different explanations to justify the human interest to observe such or such image. Although the domain of Image Aesthetics is still quite controversial, it is of major importance in many emerging domains (data-base management, automatic image indexing, web browsing, archive exploring, etc.).