Video Coding with Super-Resolution Post-Processing

We propose a video coding technique using super-resolution image processing. The proposed method reduces spatial resolution of specific pictures in input video sequence at first, namely it produces a mixed-resolution video sequence. And then the proposed method encodes the mixed-resolution video sequence. The spatially reduced pictures are converted to original resolution pictures with a super-resolution image processing. The proposed method can reduce the bit rate compared to conventional methods such as H.264, since it encodes the video sequence in which several pictures are spatially reduced. Moreover, the proposed method can prevent deterioration in subjective quality, since the spatially reduced pictures are restored with the super-resolution image processing. We evaluate the proposed method by implementing it on an H.264-based video coding scheme. Experimental results confirm that compared to H.264, our implementation of the proposed method can achieve a 3.5 dB improvement in PSNR at a low bit rate.