Image component labelling on the star graph using divide and conquer

Since its introduction in the literature, the star interconnection network has become a popular architecture for connecting parallel processors. It compares favourably with the famous hypercube as far as basic network attributes are concerned. Its degree and diameter are sublogarithmic with respect to the total number of processors and it enjoys fault tolerance and symmetry properties. It remains to be seen whether known efficient algorithms that have been developed for other interconnection network architectures can be implemented efficiently on the star, and how well these will perform in the presence of faults in the network. The authors address this question for image analysis algorithms, specifically for the image component labelling problem. Component labelling is a fundamental and basic task common to virtually all image analysis problems. The star interconnection network is modelled with a graph, an innovative grid embedding into the star is used and an efficient algorithm to solve the image component labelling problem on the star graph is developed. The authors then analyse the performance of the algorithm in the fault-free star graph