The objective of the PhD thesis is to develop new algorithms and models to predict the emotion a user may feel when watching a video. The validation of such subjective algorithms is a challenging area of research that should not be underestimated. Issues remain in terms of database availability. This PhD includes the following research contributions: creation and validation of a new video database that best fits the addressed emotional space and solve sharing issues, proposition and implementation of a computational model of emotional impact for video material based on biological and psychological concepts and using new perceptual features in order to facilitate bridging the semantic gap between low-level signal features and the high-level semantic concepts that are emotions.