No-Reference Light Field Image Quality Assessment Exploiting Saliency

In the near future, broadcasting scenario will be characterized by immersive content. One of systems for capturing 3D content a scene is Light Field imaging. The huge amount data and specific transmission impose strong constraints on services applications. Among others, evaluation quality received media cannot rely original signal but should based only data. this direction, we propose no-reference metric light field images which spatial angular characteristics. more details, estimated saliency cyclopean maps are exploited to extract features. consistency features are, instead, measured with use Global Luminance Distribution knowledge Weighted Local Binary Patterns operator Epipolar Plane Images. effectiveness proposed assessed comparing its performance state-of-the-art metrics using 4 datasets: SMART, Win5-LID, VALID 10-bit, 8-bit. Furthermore, analyzed in cross-datasets, different distortions, maps. achieved results show that model outperforms approaches perform well distortion types various models.