Comparative Analysis of Cluster Based Superpixel Segmentation Techniques

Superpixel, becomes popular for use in computer vision applications, is an image patch that adheres well to image boundaries. It has better alignment with intensity edges than rectangular patch and also capable of carrying more information. Intuitively, superpixels segment similar pixels to create visually meaningful entities while completely reducing the number of primitives for consequent image processing tasks. The main objective of superpixel segmentation, also known as image over-segmentation, is to create a consistent pixels grouping. This paper aims to present an insight of existing cluster based superpixel segmentation algorithms. Additionally, the parametric comparison of these algorithms is highlighted to establish possible research direction.