Robust Object Segmentation using Split-and-Merge

Abstract

In spite of simplicity and effectiveness in segmenting homogeneous regions in an image, Split-and-Merge (SM) algorithm is unable to segment all types of objects due to huge number of objects with myriad variations among them and due to high dependability on the threshold values used in splitting and merging techniques. Addressing these issues, a novel Robust Object Segmentation using Split-and-Merge (ROSSM) is proposed in this paper considering image feature stability, inter- and intra-object variability, and human visual perception. The qualitative analysis proves the superior performance of ROSSM in comparison with the basic SM algorithm and a recently developed shape-based fuzzy clustering algorithm namely Object-based image Segmentation using Fuzzy clustering (OSF).