Nonintrusive Method Based on Neural Networks for Video Quality of Experience Assessment

Table 6

General overview of the correlations obtained for the study cases.

Correlation	PCC	SROCC	RMSE	OR
Y-PSNR versus MOS BPNN	0.9303	0.9713	0.2882	0.5833
VQM versus MOS BPNN	0.9412	0.9707	0.2647	0.1979
SSIM versus MOS BPNN	0.9446	0.9733	0.257	0.1483
QIBF versus MOS BPNN	0.937	0.9711	0.2739	0.1562
Y-PSNR versus	0.9698	0.9873	0.1796	0.5833
VQM versus	0.9645	0.9900	0.1948	0.1666
SSIM versus	0.9497	0.9889	0.2319	0.1354
QIBF versus	0.9254	0.9853	0.2824	0.1354

Note: PCC (Pearson correlation coefficient) for prediction accuracy.
SROCC (Spearman rank order correlation coefficient) for monotonicity.
RMSE (Root Mean Square Error) for correlation quality assessment.
OR (Outlier Ratio) for consistence.