The SVD of a matrix decomposes it into three matrices, one of which is diagonal, and the other two are orthogonal.