Finding connected components in digital images by aggressive reuse of labels

摘要

An efficient algorithm is presented to label the connected components in array representation of very huge 2-D images that one works with in remote sensing applications. A new data structure is suggested to maintain the equivalence table that makes use of a modified version of Union-Find algorithms. The operations supported by the equivalence table enable aggressive reuse of labels and bound the size of the table by ⌊N/2⌋+1, for a N×N image. It is also shown that the maintenance of the table has an overall linear amortized cost.