Objective No-Reference Stereoscopic Image Quality Prediction Based on 2D Image Features and Relative Disparity

Table 3

Methods’ evaluation results for training and testing with disparity.

Methods	Training
	CC	AAE	RMSE	OR
Proposed method	0.964	0.276	0.336	0.061
Only blockiness with disp.	0.867	0.529	0.664	0.086
Only zero-crossing with disp.	0.897	0.484	0.578	0.110
Conventional method with disp.	0.903	0.479	0.597	0.114
		Testing
Proposed method	0.940	0.339	0.413	0.040
Only blockiness with disp.	0.833	0.537	0.724	0.106
Only zero-crossing with disp.	0.804	0.578	0.707	0.102
Conventional method with disp.	0.881	0.473	0.577	0.127