IMPROVED SUPERRESOLUTION-ENABLED (SRE) VIDEO CODEC

摘要

Systems and methods for transmitting source video (305) over a bandwidth-limited network (350). First, high frequency spectral content of the video is encoded into layer-1 data files (330). In parallel, frames of the video are downsampled (342) and compressed (333) using a lossy CODEC, to define layer-2 data files (335a, 335b) with high-frequency spectral content removed therefrom. The layer-1 and layer-2 data files are interleaved (345a, 345b) and transmitted over the network and are smaller than conventional lossy CODEC compressed files. After receipt, the layer-1 and layer-2 data files are deinterleaved. The layer-2 data is decompressed (353) and upsampled (362) to create a preliminary reconstructed video. The high frequency spectral content is extracted (351) from the layer-1 data files, and photometric warp superresolution processing (370) restores the high frequency spectral content into the preliminary reconstructed video to generate a final, output video (395a, 395b) at resolution at or near the source video for display to a viewer at or downstream from the receiving location.