TY - GEN
T1 - An Area-latency-balanced Hardware Design for the Reference Pixel Management of VVC Intra Coding
AU - Huang, Chengkang
AU - Huang, Leilei
AU - Zhang, Taoyu
AU - Wang, Shuocheng
AU - Fan, Yibo
N1 - Publisher Copyright:
© 2026 IEEE.
PY - 2026
Y1 - 2026
N2 - In video encoding, prediction plays a crucial role in reducing redundant information. For intra block encoding, accurate prediction relies on the availability of reconstructed neighboring pixels as reference pixels. However, in hardware encoders, the rate-distortion optimization (RDO) process continuously updates the optimal reconstructed pixels during partition decisions, requiring a dedicated design for their real-time management. This challenge is further exacerbated by the Quad-Tree with nested Multi-Type Tree (QTMTT) partitioning scheme introduced in the Versatile Video Coding (VVC) standard, which significantly increases the complexity of reference pixel management. To address this, we propose an area-latency-balanced reference pixel management (RPM) design supporting QTMTT in VVC. This design integrates register-based and SRAM-based buffers to achieve an optimal trade-off between latency and area. The proposed hardware design is implemented using GF 28nm technology, achieving zero SRAM data transfer latency, with a hardware area cost of 62295 μm2.
AB - In video encoding, prediction plays a crucial role in reducing redundant information. For intra block encoding, accurate prediction relies on the availability of reconstructed neighboring pixels as reference pixels. However, in hardware encoders, the rate-distortion optimization (RDO) process continuously updates the optimal reconstructed pixels during partition decisions, requiring a dedicated design for their real-time management. This challenge is further exacerbated by the Quad-Tree with nested Multi-Type Tree (QTMTT) partitioning scheme introduced in the Versatile Video Coding (VVC) standard, which significantly increases the complexity of reference pixel management. To address this, we propose an area-latency-balanced reference pixel management (RPM) design supporting QTMTT in VVC. This design integrates register-based and SRAM-based buffers to achieve an optimal trade-off between latency and area. The proposed hardware design is implemented using GF 28nm technology, achieving zero SRAM data transfer latency, with a hardware area cost of 62295 μm2.
KW - QTMTT
KW - VVC
KW - hardware encoder
KW - rate-distortion optimization (RDO)
KW - reference pixel management
UR - https://www.scopus.com/pages/publications/105043462923
U2 - 10.1109/ISCAS66217.2026.11562741
DO - 10.1109/ISCAS66217.2026.11562741
M3 - 会议稿件
AN - SCOPUS:105043462923
T3 - Proceedings - IEEE International Symposium on Circuits and Systems
SP - 3216
EP - 3220
BT - ISCAS 2026 - 2026 IEEE International Symposium on Circuits and Systems
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2026 IEEE International Symposium on Circuits and Systems, ISCAS 2026
Y2 - 24 May 2026 through 27 May 2026
ER -