| Autor: |
Ei Ei Tun, Supavadee Aramvith, Takao Onoye |
| Jazyk: |
angličtina |
| Rok vydání: |
2022 |
| Předmět: |
|
| Zdroj: |
ICT Express, Vol 8, Iss 1, Pp 83-90 (2022) |
| Druh dokumentu: |
article |
| ISSN: |
2405-9595 |
| DOI: |
10.1016/j.icte.2021.08.018 |
| Popis: |
The newest video compression standard, called Versatile Video Coding (H.266/VVC), outperforms its predecessor High Efficiency Video Coding (HEVC) by up to 50% in the matter of coding performance. The achievement results from the integrating of several sophisticated tools such as 32 extended angular modes for intra coding, advanced partitioning structure, and additional transform selection. However, these tools negatively deliver the computational cost of the VVC Test Model (VTM) in all-intra encoding configuration. The existing fast intra coding algorithms achieve a significant saving in computational time with a comparable rate distortion (RD) performance of VVC plus additional computational complexity. In this paper, we study the original strategy of VVC intra mode decision and analyze the correlation among the sum of absolute transformed differences (SATD) costs of the rough mode decision (RMD) and the RD costs of the rate distortion optimization (RDO) processes. Based on the useful correlations with an effective threshold formulation, we propose a fast intra prediction mode selection for the most time-consuming RDO process by selecting a small candidate set based on the SATD costs and calculating the time-consuming RD costs for only intra prediction modes of the small candidate set without a significant overhead. According to the experimental results, our proposed system can reduce up to 27.38% and on average 21.07% of the computation complexity of VTM version 5.0 with an insignificant quality degradation. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
