No-reference metrics for video streaming applications

In this paper we present no-reference metrics to evaluate the following two major distortions in streaming of compressed video over packet-switched networks: i) block-edge impairment and ii) the effect of packet loss. The first metric proposed measures block-edge impairments in reconstructed video frames and is based on the idea that these impairments are best observed in regions with low spatial activity. The second metric proposed in this paper evaluates distortion in reconstructed video frames due to packet loss by exploiting the structure of the artifact. Both metrics have low computational complexity and can be used for real-time monitoring of streaming video in a multimedia transmission scenario. Further these measures could be used as feedback to help a streaming server perform dynamic rate adaptation or dynamic selection of scalable video streams in case of increasing network congestion.