Segmentation of Moving Object captured using Moving Camera

Abstract

Segmentation of video in to objects is must for retrieval of videos based on the contents or detection systems based on the concepts. A variety of video object segmentation algorithms, including semiautomatic and automatic, have been developed. Semiautomatic methods require involvement of human and are therefore inappropriate for numerous applications. Although various applications needs segmentation to be performed automatically, there is still scope for refinement. The goal of the work proposed here is to identify the gaps in existing segmentation systems as well as to provide viable solutions for overcoming such problems in order to develop a accurate and efficient video segmentation method. In this paper the work that is proposed, addresses issues related to segmentation of video automatically, like background that is uncovered, interim poses, and background’s global motion.