Sparse and collaborative representations for computer vision door