Computer analysis of images imposes a significant computational burden on the processing hardware. In dynamic vision, the requirement is also to reduce the latency of the processing, in order to allow realistic reaction times to events in the image. Flexible, massively parallel architectures hold the promise of fulfilling these requirements for low, medium and high level vision tasks, provided that robust algorithms can be implemented in an efficient manner. This paper describes a parallel model which is designed for use as a basis for implementation of edge tracking algorithms on parallel architectures. The model is independent of edge tracking algorithms. An implementation of the model is outlined using a tracking algorithm founded on features such as the mid-point, orientation and the length of edge segments, and using a modified form of the Kalman filter. The implementation is based on transputers and consists of three independent units each of which has been staged in a studied configuration. The tracking unit is based on a tree configuration and displays MIMD characteristics. The edge extraction unit performs the Canny operator, and two parallel programming models of \em data parallelism are examined to determine the most suitable and efficient topology and data routing method for this unit. The final unit is the host interface. Performance results for the implementation are presented.