Options
2004
Conference Paper
Titel
Automatic detection of video synthesis related artifacts
Abstract
It is assumed that the textures in a video scene can be classified in two categories: textures with unimportant subjective details and the remainder. We use this idea for video coding with a texture analyzer at the encoder side and a texture synthesizer at the decoder side. The texture analyzer identifies the texture regions with unimportant subjective details and generates side information for the texture synthesizer, which in turn inserts synthetic textures at the specified locations. Our approach can be integrated in any video codec. This paper focuses on the texture synthesizer and its potentially subjectively annoying spatial artifacts. The automatic detector we developed yields an artifact identification rate of up to 79%. The former enables automatic online evaluation of the quality of synthesized frames, which allows eventual fallback on the reference video codec built on, for coding erroneous.