Video Packet Priority Assignment Based On Spatio-Temporal Perceptual Importance

A novel perceptually motivated two-stage algorithm for assigning priority to video packet data to\ud be transmitted over the internet is proposed. Priority assignment is based on temporal and spatial\ud features that are derived from low-level vision concepts. The motivation for a two-stage design is to\ud be able to handle different application settings. The first stage of the algorithm is computationally\ud very efficient and can be directly used in low-delay applications with limited computational resources.\ud The two-stage method performs exceedingly well across a variety of content and can be used in less\ud restrictive operating settings. The efficacy of the proposed algorithm (both stages) is demonstrated\ud using an intelligent packet drop application where it is compared with cumulative mean squared\ud error (cMSE) based priority assignment and random packet dropping. The proposed prioritization\ud algorithm allows for packet drops that result in significantly lower perceptual annoyance at the\ud receiver relative to the other methods considered