Destroy the Robust Commercial Watermark via Deep Convolutional Encoder-Decoder Network

Table 4

Image quality of watermarked images with different intensities after being attacked.

		F16	Lena	Man	Pepper	Baboon
a = 1	PSNR	40.10	39.68	41.29	38.69	37.29
	SSIM	0.956 0	0.953 0	0.982 1	0.934 3	0.946 6
a = 2	PSNR	39.36	38.74	40.37	37.75	36.69
	SSIM	0.950 3	0.946 2	0.978 3	0.926 4	0.942 4
a = 3	PSNR	38.80	38.01	39.95	37.27	36.48
	SSIM	0.945 6	0.940 7	0.976 6	0.920 9	0.942 6
a = 4	PSNR	38.37	37.20	39.48	36.76	36.35
	SSIM	0.941 8	0.932 3	0.974 4	0.915 4	0.941 3