An approach to multi-view subspace clustering that learns a joint subspace representation by constructing affinity matrix shared among all views is presented, relying on the importance of both low-rank and sparsity constraints in the construction of the affinity matrix.