This paper brings Hopgood's previous development of a distributed blackboard architecture to the computationally demanding application of image registration. Parallelisation is achieved by knowledge sources organised in a novel worker/manager model. The success of the approach is demonstrated in visual defect detection for quality assurance, and is now being applied to multimodal medical imaging, published elsewhere. The paper has won a Recognised Paper award at AI-2007 and will appear in a special issue of the Knowledge Based Systems journal in 2008. Knowledge-Based Systems, 21 (3), pp. 256-264.