Blind Image Inpainting with Mixture Noise Using and Total Regularization

Table 1

The quantitative performance of Figure 2 for the two compared methods with the corresponding noise setting, i.e., missing proportion for impulse noise NI and the a for Gaussian noise NG.

Image	Noise setting	ASInpaint [6]		PADMM
		PSNR	SSIM	PSNR	SSIM
Lena	50% missing σn = 15	21.82	0.6347	22.78	0.6650
Cameraman	30% missing σn = 19	19.13	0.5435	19.60	0.5630
Lena	10% missing σn = 15	24.29	0.7175	24.49	0.7240
Phantom	10% missing σn = 0.4	49.25	0.9657	49.35	0.9666
Satellite	10% missing σn = 15	22.41	0.6446	22.54	0.7662