TY - JOUR
T1 - Three-dimensional motion vector translation for geometric partitioning mode in panoramic video coding using H.266/VVC
AU - Li, Ning
AU - Wan, Shuai
N1 - Publisher Copyright:
© The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.
PY - 2024/7
Y1 - 2024/7
N2 - Geometric partitioning mode (GPM) is introduced into H.266/versatile video coding (VVC) as a new inter-frame prediction technology, in which one block can be divided into two non-rectangular blocks to increase partitioning precision of moving objects. In this paper, a three-dimensional (3D) motion vector (MV) translation based on spherical coordinate transformation is introduced into H.266/VVC and applied to GPM, in which the MV candidate list is constructed separately for each GPM partition. Firstly, the number of MVs in the GPM MV candidate list is expanded to 12, and all MVs in the merge mode candidate list are reserved, which can prevent the MVs with the best encoding performance from being removed. Secondly, the MVs of neighboring blocks of the current block are transformed by 3D MV translation and put into the blank positions of the newly constructed GPM partition MV candidate list. Thirdly, an independent motion compensation prediction based on the MV of each GPM partition is performed to improve the prediction accuracy. The experimental results demonstrated the effectiveness of the proposed method, achieving up to 0.44% reduction in BD rate in rate-distortion performance compared to H.266/VVC.
AB - Geometric partitioning mode (GPM) is introduced into H.266/versatile video coding (VVC) as a new inter-frame prediction technology, in which one block can be divided into two non-rectangular blocks to increase partitioning precision of moving objects. In this paper, a three-dimensional (3D) motion vector (MV) translation based on spherical coordinate transformation is introduced into H.266/VVC and applied to GPM, in which the MV candidate list is constructed separately for each GPM partition. Firstly, the number of MVs in the GPM MV candidate list is expanded to 12, and all MVs in the merge mode candidate list are reserved, which can prevent the MVs with the best encoding performance from being removed. Secondly, the MVs of neighboring blocks of the current block are transformed by 3D MV translation and put into the blank positions of the newly constructed GPM partition MV candidate list. Thirdly, an independent motion compensation prediction based on the MV of each GPM partition is performed to improve the prediction accuracy. The experimental results demonstrated the effectiveness of the proposed method, achieving up to 0.44% reduction in BD rate in rate-distortion performance compared to H.266/VVC.
KW - 3D MV translation
KW - Geometric partitioning mode
KW - H.266/VVC
KW - Panoramic video
UR - http://www.scopus.com/inward/record.url?scp=85186626138&partnerID=8YFLogxK
U2 - 10.1007/s11760-024-03073-9
DO - 10.1007/s11760-024-03073-9
M3 - 文章
AN - SCOPUS:85186626138
SN - 1863-1703
VL - 18
SP - 4303
EP - 4312
JO - Signal, Image and Video Processing
JF - Signal, Image and Video Processing
IS - 5
ER -