In qualitative motion representation, frames of reference play an important role as well in measuring of the motion data as in representation and application of algorithms. This paper discusses motion representation in egocentric and allocentric frames of reference and begins with some general considerations on motion representation through qualitative distances and directions that apply to both technical and biological systems. An approach that involves incremental numeric generalization of a numerically represented motion track and subsequent transformation in a qualitative representation has advantages for technical systems, though. Last, algorithms for generalizing qualitatively represented motion tracks in egocentric and allocentric frames of reference are presented.