Disguise of Steganography Behaviour: Steganography Using Image Processing with Generative Adversarial Network

Table 3

Testing error of the processed stego image generated from different image processing generation models against Ye-net.

Image processing	Algorithm	Payload (bpp)
		0.05	0.1	0.2	0.3	0.4	0.5
Histogram equalization	HILL	0.0101	0.0009	0.0002	0.0002	0	0
	DCGAN1	0.4290	0.4240	0.4024	0.3881	0.3640	0.3489
	WGAN-GP1	0.4427	0.4409	0.4388	0.3932	0.3881	0.3674
	S-UNIWARD	0.0018	0.0006	0.0002	0	0	0
	DCGAN2	0.4238	0.4186	0.4008	0.3889	0.3662	0.3497
	WGAN-GP2	0.4349	0.4343	0.4217	0.3924	0.3706	0.3515
	WOW	0.0008	0.0007	0	0	0	0
	DCGAN3	0.4116	0.4107	0.3872	0.3771	0.3654	0.3508
	WGAN-GP3	0.4345	0.4226	0.3918	0.3866	0.3694	0.3513
Sharpening	HILL	0.1417	0.0132	0.0017	0.0006	0	0
	DCGAN1	0.4351	0.4348	0.4104	0.3893	0.3689	0.3502
	WGAN-GP1	0.4501	0.4489	0.4376	0.3964	0.3883	0.3694
	S-UNIWARD	0.1106	0.0105	0.0003	0	0	0
	DCGAN2	0.4308	0.4296	0.4011	0.3998	0.3803	0.3617
	WGAN-GP2	0.4494	0.4363	0.4111	0.4034	0.3968	0.3708
	WOW	0.0979	0.0093	0.0018	0	0	0
	DCGAN3	0.4500	0.4127	0.4001	0.3802	0.3743	0.3529
	WGAN-GP3	0.4589	0.4311	0.4109	0.3956	0.3892	0.3705