Error Concealment for SVC Utilizing Spatial Enhancement Information

In this paper packet loss error concealment for video sequences compressed using spatial scalability is investigated. Slice support is implemented into the JSVM reference codec of the scalable extension to H.264/AVC video coding standard. The non-normative error concealment scheme introduced in the codec is developed further, adding to it the capability to also consider correctly received slice information from the same frame to conceal lost frame areas. In case of lost base layer slices further improvement on the reconstruction is then achieved by using the correctly received spatial enhancement information for the same frame. The proposed enhancements focus on packet loss concealment on the base layer of I and P-coded frames, where the greatest improvements to the original scheme were identified. Simulation results for given packet loss model indicate on average 2dB improvement over original scheme in the target error scenario.