JCTVC-A022.docVIP

Joint Collaborative Team on Video Coding (JCT-VC) of ITU-T SG16 WP3 and ISO/IEC JTC1/SC29/WG11 1st Meeting: Dresden, DE, 15-23 April, 2010 Document: JCTVC-A022 Title: Multiple Predictor Sets for Intra Coding Status: Input Document to JCT-VC Purpose: Proposal, Information Author(s) orContact(s): Lingzhi Liu2330 Cental Expy.Santa Clara, CA 95050, USA Tel:Email: +1-408-3304996lzliu@ Source: Huawei Technologies _____________________________ Abstract To achieve better coding gain of intra (I) frames in H.264/MPEG AVC, multiple predictor sets for intra coding (MPSI) are proposed to improve the accuracy of intra prediction. A new predictor set is introduced with similar subsets as the original one in H.264/AVC, in which 8 are directional and 1 is non-directional. The same method in H.264/AVC is used to predict the intra prediction mode of current block. Experimental results show that the proposed method achieves 3.14 % average bit reduction for I frames at constraint set 1 and constraint set 2 with CABAC, and 3.89 % average bit reduction at constraint set 2 of CAVLC. The technique is compatible with other KTA tools like Adaptive Loop Filter, and achieves more bit reduction when used together. Introduction H.264/AVC uses intra prediction to reduce the spatial redundancy between spatially adjacent blocks, in which pixels from the already coded neighboring blocks are used to predict the current block. For luminance block, H.264 has a predictor set of 9 predictors for 4x4 and 8x8 blocks. The predictors can be divided into two subsets according to the texture feature of the predictors. One is directional subset with 8 directional predictors; another is non-directional with 1 DC predictor as shown in Fig. 1. Fig 1 Prediction Modes in H.264/MPEG AVC It has been proved that the above method works well for general sequences. However the weakness of the prediction method is that it is not able to represent sample prediction blocks with complex textures. I