this is not correct - for eigendecomposition where U = V this works, but not for general SVD - you can't just add the diagonal matrix to the singular values. This is clear if you distribute U and V^T in the equation - you get the original SVD plus c times UV^T which does not equal I unless the SVD is an eigendecomposition