Video coding using texture analysis and synthesis

A new approach to video coding is presented, where video scenes are classified into textures with subjectively relevant and irrelevant details. We apply this idea to improve video coding by using a texture analyzer and a texture synthesizer. The analyzer identifies the texture regions with no important subjective details and generates coarse masks as well as side information for the synthesizer at the decoder side. The synthesizer replaces the detail-irrelevant textures by inserting synthetic tex-tures into the identified regions. Texture analyzer and synthesizer are based on MPEG-7 descriptors. The approach has been integrated into an H.264/AVC codec. Bitrate savings up to 19.4 % are shown for a semi-automatic texture analyzer given similar subjective quality as the H.264/AVC codec without the presented approach