Real-time Foreground Segmentation and Boundary matting for Live Videos using SVM technique

Computer engineering

Segmentation is nothing but to simplify and change the representation of video into something that is more meaningful and easier to analyze. Foreground segmentation is nothing but the extraction of desired foreground object from input videos. Large amount of efforts are taken on this topic but there is still lack of significant algorithm that work on live videos of object with fuzzy boundaries captured by movable camera. This paper presents an algorithm that can work in real-time for freely moving cameras. In this we use two competing one class support vector machine at each pixel location they can work according to local color distribution for both foreground and background. By using this machine learning method and addressing foreground segmentation and boundary matting in integrated manner this algorithm gives complete extraction of foreground object at processing a wide range of videos with complex background from freely moving cameras in real-time. In this also calculate the real-time processing speed in the form of frames per seconds as well as graph frames.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.