Regularized supervised Bayesian approach for image deconvolution with regularization parameter estimation

EURASIP Journal on Advances in Signal Processing

Table 2 PSNR and SSIM values obtained by five methods for seven different images degraded by kernel 2

Image	Size	Measure	Blurred	Method
				RL	Wiener	TV	BTV	MAP- H¹
Lena	256×256	PSNR	23.14	26.52	26.45	26.91	26.94	27.88
		SSIM	0.6794	0.7720	0.7824	0.7943	0.7967	0.8410
Cameraman	512×512	PSNR	24.38	31.24	30.76	30.29	30.38	31.58
		SSIM	0.7737	0.8959	0.8987	0.8895	0.8910	0.9153
House	256×256	PSNR	24.74	28.93	28.60	29.18	29.26	30.73
		SSIM	0.7264	0.7861	0.7860	0.7982	0.8011	0.8469
Couple	512×512	PSNR	23.60	26.83	26.97	26.75	26.77	27.44
		SSIM	0.5625	0.7341	0.7431	0.7292	0.7293	0.7653
Pirate	512×512	PSNR	24.67	28.10	28.06	28.09	28.16	28.83
		SSIM	0.6282	0.7778	0.7852	0.7739	0.7766	0.8068
Boat	512×512	PSNR	23.63	27.31	27.39	27.08	27.10	27.81
		SSIM	0.5939	0.7546	0.7623	0.7454	0.7457	0.7774
Fingerprint	512×512	PSNR	18.07	25.64	25.96	26.23	26.37	27.40
		SSIM	0.3938	0.8568	0.8642	0.8660	0.8683	0.8959

For each test setting, six results are provided: blurred, Richardson-Lucy algorithm, Wiener filter, TV method, BTV approach, and our proposed model. Bold format: the best score in each line