Multiregion Object Segmentation Methods - A Survey

Abhisha Mano, B. P. Beno Ben, G. Jesly Marsha

Abstract: Image segmentation is the elementary step to investigate metaphors and haul out data from them. It is the field widely researched and still offers various challenges for the researchers. This paper tries to put glow on the critical ideology on the methods used to fragment an image. Image segmentation can be roughly be categorized as semi-interactive approach and fully habitual approach and the algorithms industrial lies in either of this approaches. Image segmentation is a critical step as it frankly influences the overall success to understand the image. This paper concentrates on the various approaches used to segment multiregion objects like Graph method, Lagrangian duality, Hausdorff distance, level set etc. A comparison was made between Lagrangian duality and level set approach in segmenting multi region objects. In these techniques it is found that level set approach outperforms the others in mltiregion object segmentation.

Keywords: Segmentation Methods, Image, Cluster, Graph-cut, Hybrid Methods