PROJECT TITLE :

Tensor Oriented No-Reference Light Field Image Quality Assessment

ABSTRACT:

Immersive media acquisition, processing, and application are becoming more dependent on the quality of light field images (LFIs). A multi-dimensional difficulty is created by the fact that LFI quality assessment must take into account both the spatial and angular dimensions of quality decline. A new Tensor-NLFQ (Tensor-NLFQ) based on the idea of tenors has been proposed to evaluate the quality of light fields. Tucker decomposition is used to derive the principal components of four oriented sub-aperture view stacks because the LFI is considered a low-rank 4D tensor. LFI's global naturalness and local frequency attributes are taken into account when calculating the Principal Component Spatial Characteristic (PCSC). It has been recommended that TAVI, which looks at the distribution of structural similarity between the first principal component and each view in the view stack, be used to assess the quality of angular consistency. It has been demonstrated that the suggested Tensor-NLFQ model outperforms the current state-of-the-art algorithms for 2D, 3D, multi-view, and LFI quality assessments.