Table 3. Average norm of sparse code errors of MSE–based OMPs and robust alternatives for different types of noise.

Best results are marked bold. K = K₀ = 10.

Noise	Gaussian	Laplacian	Student	Chi–squared	Exponential
OMP	5.92	5.69	7.14	5.22	4.43
GOMP	7.66	7.27	9.37	6.71	5.65
CMP	5.57	4.40	3.87	3.08	3.49
Cauchy	5.88	5.21	4.43	3.95	4.06
Fair	5.92	5.34	5.05	4.45	4.13
Huber	5.80	5.04	4.57	3.92	3.89
Tukey	5.85	4.78	3.80	3.05	3.64
Welsch	5.82	4.84	3.90	3.20	3.70