Non-Orthogonal Sampling as the Basis for Video Data Compression and Reconstruction

Abstract

The study proposes video data compression and reconstruction methods based on various non-orthogonal spatiotemporal sub-Nyquist sampling patterns that provide 4 to 16 times compression, followed by reconstruction via interpolation with information loss levels below the visually perceptible threshold. The proposed methods can be implemented in real time, as demonstrated in the study, and used either alone or in combination with other video compression and coding techniques, providing additional compression.