Feature analysis based super-resolution for video signal processing

HDTV flat panel displays are now available in many living rooms. However, a huge amount of video is still in standard definition format or has even a lower resolution and may suffer from severe coding artifacts. Displaying all this video on panels with high resolution grids makes methods mandatory which can significantly improve the video quality. Iterative reconstruction algorithms can fulfil this challenging task by utilizing internal image models and an integrated feed-back loop to control the output quality, enhancing resolution and reducing artifacts. The contribution of this paper is to depict the potential of iterative reconstruction algorithms by presenting a novel super-resolution algorithm, which is computationally very efficient and robust against motion estimation errors. Furthermore, integration of additional control mechanism into this algorithm is proposed, increasing its flexibility for processing input video with different quality levels, controlling the sharpness and detail level in the output sequence and strongly reducing side-effects caused by wrong motion estimation.