Connected components labeling based on union-find operations applied to connected branches

This paper presents an efficient run-based algorithm for labeling connected components in a binary image. By means of two tables called index table and equivalent information table respectively, the algorithm assign labels only to run sets which are composed of a series of runs connected to each other in adjacent scan rows, then the same index in equivalent information table can be assigned to all runs of a set in index table at one time. Further, the algorithm organizes all equivalence information in equivalent information table in a tree structure, and employee the union-find technique to merge connected components rapidly. The algorithm is very simple, and fewer operations are needed to check and to label runs of a connected branch. Furthermore, it is not necessary to calculate the minimum label while combining two runs, and is able to construct a union-find tree with smaller height. Thus the algorithm is able to provide a high speed to label connected components for general images.