The aim of this work is to model the apparent motion in image sequences depicting textured motion patterns. We adopt the mixed-state Markov Random Fields (MRF) models recently introduced to represent the so-called motion textures. The approach consists in describing the spatial distribution of local motion measurements which exhibit values of two types: a discrete component related to the absence of motion and a continuous part for actual measurements. The former accounts for symbolic information that is beyond the null value of motion itself, providing crucial information on the dynamic content of the scene. We propose several significant extensions and we give theoretical results regarding this model, which are of great importance for its application to motion analysis. In this context, dynamic content recognition applications are analyzed. We have defined a motion texture classification scheme, and a motion texture segmentation method exploiting this modeling. Results on real examples demonstrate the accuracy and efficiency of our method.