A (natural) image is generally not a series of independent pixels (points with certain colors), but contains, for example, different objects, which are distinguished from one another and have a texture. Such objects (e.g. circles, lines, but also more complex objects) are often repeated in pictures, either as a self-similar copy or a transformed copy (e.g. tree leaves or slats in the fence). The same applies to other contents of multimedia data, such as the sound.
Furthermore, elements or objects in the multimedia data are often associated with each other or grouped, e.g. certain sounds belong often to specific objects (e.g. a C-hook), object belong to other objects (e.g. the C-hook to a duck) and several objects can be grouped into one object (e.g. a field with a group of ducks).