Motion compensated interpolation for advanced standards conversion and noise reduction

The paper is concerned with modifications and extensions in order to adapt HDTV-to-film conversion to applications of purely 2-D interlace sampling grid conversions without considering the problem of calculating intermediate frames. The degradations of conventional methods based on purely 2-D spatial interpolation are analyzed. Assuming a reliable motion estimation the spatial resolution can be improved significantly by use of a 3-D motion compensated interpolation filter. Improved concepts of noise reduction become more important because HDTV cameras show, compared with TV cameras, an increased amount of noise up to about 16 dB. The limits and degradations of motion uncompensating methods for noise reduction are analyzed. A motion compensated concept for 2-D interlace sampling grid conversion applications and noise reduction is proposed, which is adapted to the reliability and accuracy of motion vectors. The concealment is performed by adapting the vertical/temporal bandwidth of a 3-D interpolation filter.