Survey: Foreground Segmentation of Live Videos Using Boundary Matting Technology

In this paper we to propose interactive methodology to extract foreground segmentation of objects from input videos using Boundary Matting Technology. Given an initial segmentation for the primary associated frame of a first and last video sequence, our objective is to propagate the segmentation from the middle frame in the sequence. Video frames are 1st divided into super pixels. As a final step the initial segmentation boundaries are improved to get an accurate object boundary. We then exploit these object boundaries in associate application for matting these videos. Here an effective algorithm for segmentation and then matting them is done accordingly which processes live videos with fuzzy boundaries. In our algorithm we generate Support Vector Machine (SVMs) and also algorithm which models local color distribution for both foreground and background video frames.