A Novel Intra Refreshment Algorithm for ROI

The importance of video sequences varied with the regions, and human visual sensitivity too varied with the importance of region of video sequences, so the subjective quality requirement of region of interest (ROI) was not the same as non-ROI. Traditional intra refreshment algorithm didn't consider the human visual characteristics, and it cost much more bits, but the subjective quality was still very bad in the decoder, so it is necessary for the codec to consider the human visual attention. So the paper proposed the RDO intra refreshment algorithm based the human visual attention, and it used the end to end overall distortion theory to analysis the ROI and non-ROI, and it would given more intra refreshment chances to the marcoblock of ROI. As the bit rate was constrained, so the bit rate of the ROI and non-ROI must be reallocated. At last, the algorithm was implemented in reference software JM12.2 of JVT. The many simulation results showed that, comparing with the traditional RDO intra refreshment algorithms, the proposed intra refreshment algorithm based human visual attention-ROI could obtain a better subjective quality of the reconstructed video in the region of human eyes interest.