Segmentation Based Video Steganalysis to Detect Motion Vector Modification

Table 4

Three categories’ detection accuracies (%) of improved NPELO and improved MVRBR.

Steganalysis	Steganography	BR (Mbps)	0.5		1		3		10
		ER	0.1	0.2	0.1	0.2	0.1	0.2	0.1	0.2
Improved NPELO	Cao’s	Slow	78.78	91.32	80.78	91.51	85.29	93.11	85.57	97.10
		Middle	81.92	95.19	82.28	95.47	93.03	97.53	93.38	98.88
		High	88.67	97.01	89.14	97.47	94.90	98.00	95.53	99.01
	Yao’s	Slow	91.01	96.10	92.05	97.96	91.94	99.66	97.13	99.60
		Middle	97.84	98.49	97.50	98.64	97.86	99.68	98.09	99.61
		High	98.66	99.41	98.91	99.17	98.69	99.82	98.72	99.90
	Wang’s	Slow	79.65	89.13	76.77	92.85	85.61	93.16	86.74	96.07
		Middle	81.28	94.68	81.26	93.79	88.30	97.86	90.11	98.31
		High	84.43	97.22	85.34	96.32	92.60	98.74	93.59	98.89
Improved MVRBR	Cao’s	Slow	71.70	83.99	73.83	84.16	77.03	88.21	82.07	89.88
		Middle	79.87	92.79	85.98	93.81	92.21	98.03	92.60	98.73
		High	85.44	94.11	89.83	96.60	94.68	98.66	93.80	99.38
	Yao’s	Slow	79.04	87.24	82.05	92.37	82.74	95.58	84.78	98.31
		Middle	87.36	95.99	91.22	96.85	92.03	97.96	93.53	99.59
		High	89.34	97.20	92.66	98.25	92.39	99.54	95.68	99.81
	Wang’s	Slow	68.81	75.14	70.28	81.27	74.31	86.04	76.09	88.36
		Middle	76.36	89.53	87.08	94.21	88.78	95.29	91.12	96.59
		High	86.40	91.56	89.62	96.78	91.73	97.30	93.88	98.62