Abstract
Video segmentation is an important part of computer vision research, which has been widely used in the fields of security monitoring, video compression and so on. The popular video segmentation methods can be divided into unsupervised ones and semi-supervised ones. Unsupervised video segmentation methods, which segment the saliency object, do not need the prior information given by users. The semi-supervised methods need usersto mark the object they want to segment in the first frame or some key frames. This paper proposes a new semi-supervised video segmentation method, which is based on random walk model. We need usersto mark the object they want to segment in the first frame, and then we use the spatio-temporal random walk to pass the segmentation results of current frame to the next frame. In this way, we can segment what we want of the whole video. Our experiments are performed on SegTrack database, and the experiment results have proved out the effectiveness of our method.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have