Packet dropping for H.264 videos considering both coding and packet-loss artifacts

Abstract

In the context of both compression artifacts and packet loss artifacts, we use generalized linear models to predict VQM quality scores. Using a network-based model, a router can estimate the visual importance of each incoming packet and decide which packet to drop when congestion happens. Considering a wide variety of bit reduction rates, we perform packet dropping experiments for combinations of video streams and examine the effects of video contents and different bit rates. By comparing with randomly dropping B slices or B frames, we conclude that our model gives a good performance on objective evaluation of packet importance.