Performance analysis of video segmentation
Harshit Aditya, Thota Gayatri, Taji Taramati Santosh, Shilpa Ankalaki, Jharna Majumdar · 2017
Segmentation plays a vital role in digital media processing, pattern recognition and computer vision. In the last four decades, extensive research has been done and a number of algorithms have been published in the literature. Each one has its own merits and demerits. This paper aims to make a comparative analysis of the most popularly known segmentation methods, namely K-Means, Region Growing, Mean shift and Watershed segmentation for video from different category. The contribution of the paper is twofold: Conventionally, the value of K in K-Means segmentation is not known a prior and given as input. In order to avoid manual input by the user, Region growing segmentation is used. The prominent regions come as output of the region growing method, is used as input for K-Means segmentation. The performance of the segmentation algorithms is determined using a set of Quality Metric (QM) parameters. Segmentation is done on RGB Color Video from Entertainment, Sports and Natural Scenery category. The results show the most suitable algorithm for segmentation for each category of video. UBUNTU C Version 16.04 LTS is used to implement the algorithms.