Supporting binocular visual quality prediction using machine learning
Shanshan Wang, Feng Shao, Gangyi Jiang · 2014
We present a binocular visual quality prediction model using machine learning (ML). The model includes two steps: training and test phases. To be more specific, we first construct the feature vector from binocular energy response of stereoscopic images with different stimuli of orientations, spatial frequencies and phase shifts, and then use ML to handle the actual mapping of the feature vector into quality scores in training procedure. Finally, quality score is predicted by multiple iterations in test procedure. Experimental results on three publicly available 3D image quality assessment databases demonstrate that, in comparison with the most related existing methods, the proposed technique achieves comparatively consistent performance with subjective assessment.