A New Boosting Algorithm for Shrinkage Curve Learning

Table 1

Comparison of several denoising results under different noise levels. Test images are from Figure 1.

Image
			Ours	Impv			Ours	Impv			Ours	Impv
Cameraman	35.91	35.91	36.19	0.28	31.60	31.60	32.01	0.41	29.49	29.50	29.89	0.40
Lena	36.82	36.96	37.91	1.09	33.29	33.35	34.65	1.35	31.64	31.65	32.81	1.17
Barbara	36.57	36.63	36.87	0.30	32.54	32.56	32.90	0.36	30.45	30.46	30.85	0.39
Boat	36.33	36.36	36.57	0.24	32.48	32.48	33.04	0.56	30.58	30.57	31.11	0.53
Couple	35.18	34.83	35.47	0.29	30.44	3dsl0.28	30.99	0.54	28.06	28.00	28.59	0.53
Fingerprint	36.10	36.03	36.67	0.57	31.86	31.86	32.43	0.57	29.59	29.60	30.10	0.51
Hill	35.06	34.89	35.44	0.37	30.59	30.48	30.96	0.36	28.55	28.51	28.82	0.27
House	36.55	36.57	38.34	1.79	32.88	32.93	34.71	1.83	31.20	31.23	32.74	1.54
Man	35.03	34.77	35.31	0.28	30.33	30.21	30.72	0.39	28.10	28.01	28.24	0.14
Montage	36.75	36.76	37.06	0.32	32.82	32.83	33.20	0.38	30.78	30.79	31.20	0.42
Peppers	36.26	36.31	36.96	0.70	32.30	32.33	33.07	0.78	30.33	30.34	31.00	0.68
Image
			Ours	Impv			Ours	Impv			Ours	Impv
Cameraman	28.10	28.12	28.50	0.40	27.04	27.06	27.45	0.40	23.45	23.41	24.34	0.89
Lena	30.54	30.53	31.50	0.96	29.68	29.64	30.46	0.78	26.86	26.82	27.16	0.30
Barbara	29.02	29.02	29.36	0.34	27.91	27.90	28.15	0.24	24.46	24.48	24.10	-0.35
Boat	29.31	29.30	29.75	0.44	28.34	28.33	28.71	0.37	25.34	25.33	25.64	0.30
Couple	26.56	26.53	27.04	0.48	25.50	25.48	25.92	0.42	22.72	22.71	22.84	0.11
Fingerprint	28.01	28.02	28.54	0.53	26.77	26.79	27.35	0.58	22.70	22.75	23.55	0.86
Hill	27.30	27.30	27.50	0.21	26.42	26.44	26.57	0.15	24.06	24.05	24.17	0.12
House	30.08	30.10	31.41	1.33	29.22	29.23	30.41	1.19	26.47	26.46	27.43	0.96
Man	26.74	26.69	26.70	-0.04	25.79	25.76	25.68	-0.11	23.28	23.27	23.20	-0.07
Montage	29.36	29.37	29.84	0.49	28.21	28.23	28.80	0.59	24.39	24.30	25.49	1.10
Peppers	29.00	29.01	29.60	0.60	27.96	27.97	28.52	0.56	24.51	24.46	25.09	0.58