Sparse approximation with an orthogonal complementary matching pursuit algorithm

摘要: This paper presents the orthogonal extension of recently introduced complementary matching pursuit (CMP) algorithm for sparse approximation [1]. The CMP is analogous to (MP) but done in row-space dictionary matrix. It suffers from a similar sub-optimality as MP. (OCMP) presented here tries remove this by updating coefficients all selected atoms at each iteration. Its development follows same procedure (OMP). In contrast with OMP, residual errors resulting OCMP may not be up respective Though energy increase over OMP during first iterations, it shown that, compared convergence speed increased subsequent iterations and sparsity solution vector improved.