This paper presents a survey of existing sonification systems used to represent visual scenes, analyzes their characteristics and proposes a taxonomy of this set of algorithms and devices. This classification is non-exclusive, since many parameters of any sonification procedure work as independent variables and can be recombined in any other set. Although many of these algorithms have been proposed in the field of assistive technology, and most of the examples come from that field, we will only focus on auditory aspects, avoiding an analysis in terms of mobility, rehabilitation or subjective perception. We propose two main categories to classify every sonification algorithm, the psychoacoustic and the artificial, and a third one mixing properties of each one of them. We use classic paradigms such as the pitch, piano and point transform, as well as some new subsets. A final summary of 25 different assistive products is given, with their classification following our scheme.