摘要:
A system and method for performing candidate-based decoder-side motion vector determination (DMVD). Candidate motion vectors (MVs) may be rounded to the nearest whole or integer pixel. The rounded candidate MV having the best sum of absolute differences (SAD) may be identified. This may be used as the final MV. Alternatively, the un-rounded MV corresponding to this rounded candidate MV may be used as the final MV. Alternatively, a small range integer search may be performed around the chosen rounded candidate MV, and the best integer pixel in the search area may be identified and used to define the final MV. Alternatively, an intermediate MV may be chosen, where this MV is intermediate between the chosen rounded candidate MV and the MV corresponding to the best integer pixel in the search area.
摘要:
Methods and systems to apply motion estimation (ME) based on reconstructed reference pictures in a B frame or in a P frame at a video decoder. For a P frame, projective ME may be performed to obtain a motion vector (MV) for a current input block. In a B frame, both projective ME and mirror ME may be performed to obtain an MV for the current input block. The ME process can be performed on sub-partitions of the input block, which may reduce the prediction error without increasing the amount of MV information in the bitstream. Decoder-side ME can be applied for the prediction of existing inter frame coding modes, and traditional ME or the decoder-side ME can be adaptively selected to predict a coding mode based on a rate distribution optimization (RDO) criterion.
摘要:
Systems, devices and methods for performing low memory access candidate-based decoder-side motion vector determination (DMVD) are described. The number of candidate motion vectors (MVs) searched may be confined by limiting the range of pixels associated with candidate MVs to a pre-defined window. Reference windows may then be loaded into memory only once for both DMVD and motion compensation (MC) processing. Reference window size may be adapted to different PU sizes. Further, various schemes are described for determining reference window positions.
摘要:
To let decoder side motion vector derivation (DMVD) coded blocks be decoded in parallel, decoder side motion estimation (ME) dependency on spatially neighboring reconstructed pixels can be removed. Mirror ME and projective ME are only performed on two reference pictures, and the spatially neighboring reconstructed pixels will not be considered in the measurement metric of the decoder side ME. Also, at a video decoder, motion estimation for a target block in a current picture can be performed by calculating a motion vector for a spatially neighboring DMVD block, using the calculated motion vector to predict motion vectors of neighboring blocks of the DMVD block, and decoding the DMVD block and the target block in parallel. In addition, determining a best motion vector for a target block in a current picture can be performed by searching only candidate motion vectors in a search window, wherein candidate motion vectors are derived from a small range motion search around motion vectors of neighboring blocks.
摘要:
An adaptive Wiener filter may be applied to improve coding efficiency because of information lost during quantization of the video encoding process. The Wiener filter may be selectively applied globally to an entire picture or locally to portions of the picture. Histogram segmentation may be used to select pixels for Wiener filtering in some embodiments. The Wiener filter may be adaptively applied to histogram bins, improving coding efficiency in some cases.
摘要:
Systems, apparatus, articles, and methods are described including operations for 3D video coding including depth based disparity vector calibration.