Perceptual Video Quality Analysis for HEVC in Packet Loss Environment

HEVC provides a significant increase in compression efficiency over legacy standards, such as MPEG-2 and AVC. It is well known that an increase in compression efficiency typically results in decreased coding redundancy. This leads to subsequent vulnerability to error propagation resulting from transmission of stream(s) on a network under packet loss (PL) conditions. Here, incorporating/formulating a perceptual Video Quality (VQ) metric and analyzing VQ over a range of bit rates, PL profiles and a variety of content; would be of interest. Although a number of statistical/opinion models have been proposed and tested for H264 and earlier codecs in various packet loss environments, the parameters of such models have not been defined for HEVC. These statistical models have been designed on the basis of bit rates, packet loss rate, resolution and other parameters. It is assumed that the model parameters will remain same for the entire video sequence. These models do not take content specific properties into consideration, so these are not reliable; different profiles of spatial and temporal variations of content will affect the perceptual video quality differently. Even for the same content, effect of packet loss on perceptual video quality can be very much different depending upon the data lost with the packet. e.g. the effect of loss of a packet corresponding to an I frame can be higher than that of P or B frames because of temporal referencing. So, a new method has been devised and tested over the content with varying profiles of content, bit rate and other characteristics.

Published

2014-10

Content type

Original Research

Keywords

HEVC, Packet Loss, VQ metric

DOI

10.5594/M001539

ISBN

978-1-61482-954-6