A new marker-based watershed algorithm

Abstract

The marker-based watershed approach is a very efficient means for image segmentation and has been widely used in recent years. The conventional marker-based algorithms are realized using hierarchical queues. A new marker-based watershed algorithm based on the disjoint set data structure is proposed in this paper. It consists of two steps: the flooding step and the resolving step. This algorithm has significantly lower memory requirement as compared with the conventional algorithms while maintaining the computational complexity of O(N) where N is the image size. Experimental results further show that the new algorithm implemented in C language runs much faster than the conventional algorithm based on the hierarchical queues as a result of savings from huge memory allocation and releasing operations.