A 4-D Sparse FIR Hyperfan Filter for Volumetric Refocusing of Light Fields by Hard Thresholding
Sanduni U. Premaratne, Chamira U. S. Edussooriya, Chamith Wijenayake, Len T. Bruton, P. Agathoklis · 2018
A low-complexity 4-D sparse FIR hyperfan filter is proposed for volumetric refocusing of light fields. By exploiting the partial separability of the spectral region of support of a light field, the proposed filter is designed as a cascade of two 4-D hyperfan filters. The sparsity of the filter coefficients is achieved by hard thresholding the nonsparse filter coefficients. The experimental results confirm that the proposed 4-D sparse FIR hyperfan filter provides 72% mean reduction of computational complexity compared to a 4-D nonsparse FIR hyperfan filter withoudeteriorating the fidelity of volumetric refocused light fields. In particular, the mean structure similarity (SSIM) index between the volumetric refocused light fields by the proposed sparse filter and the nonsparse filter is 0.989.