Unsupervised view and rate invariant clustering of video sequences