Clump splitting in Histopathological images based on concave points

摘要

This paper proposes a simple approach to split the clumps found in Histopathological Images. Watershed algorithm is generally used to segment and separate the clumps. Several studies have revealed that watershed algorithm suffer drawbacks leading to lesser accuracy in clump splitting. From the literature it is evident that most of the existing methods on clump splitting have been applied on binary image without considering the color information. In this paper we propose a novel clump splitting method by considering color features. The proposed method shows better accuracy when clumps are irregular in shape and have multiple splits.