EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Learning-Based Multiple Transform Selection Method for VVC Encoding

Caroline Souza Camargo, Bianca Silveira, Claudio Diniz, Daniel Palomino, Guilherme Corrˆea

Journal of Integrated Circuits and Systems · 12.04.2026 · 21 (1) · 1-12

DOI: 10.29292/jics.v21i1.1094 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Integrated Circuits and Systems, online erschienen am 12.04.2026. Verfasst von Caroline Souza Camargo und 4 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The Versatile Video Coding (VVC) standard achieves significant bitrate reduction compared to its predecessors, but at the cost of a substantial increase in computational complexity. A major contributor to this complexity is the Multiple Transform Selection (MTS) tool, which exhaustively evaluates multiple transform combinations to improve coding efficiency. This work proposes a fast transform selection method based on machine learning to reduce the computational cost of MTS in the VVC encoder. Two decision tree models are designed to operate sequentially within the transform evaluation loop, making early decisions to prune the search space. To train these models, a comprehensive dataset was built by extracting 25 features from various video sequences, using Reservoir Sampling to create a balanced and representative training set. Different block grouping strategies were evaluated, with the area-based grouping demonstrating the best trade-off between accuracy and model complexity. The proposed solution, integrated into the VVC Test Model (VTM 23.0), achieved an average encoding time reduction of 17.92% with a 1.32% BD-Rate increase in the All Intra configuration. These results demonstrate that the proposed method significantly reduces the complexity of the transform stage while maintaining a favorable balance between computational performance and compression efficiency.

Zitieren

APA

Camargo, C. S., Silveira, B., Diniz, C., Palomino, D., & Corrˆea, G. (2026). Learning-Based Multiple Transform Selection Method for VVC Encoding. Journal of Integrated Circuits and Systems, 21(1), 1-12. https://doi.org/10.29292/jics.v21i1.1094

Vancouver

Camargo CS, Silveira B, Diniz C, Palomino D, Corrˆea G. Learning-Based Multiple Transform Selection Method for VVC Encoding. Journal of Integrated Circuits and Systems. 2026;21(1):1-12. doi:10.29292/jics.v21i1.1094

BibTeX
@article{Camargo2026,
  author = {Camargo, Caroline Souza and Silveira, Bianca and Diniz, Claudio and Palomino, Daniel and Corrˆea, Guilherme},
  title = {Learning-Based Multiple Transform Selection Method for VVC Encoding},
  journal = {Journal of Integrated Circuits and Systems},
  year = {2026},
  volume = {21},
  number = {1},
  pages = {1--12},
  doi = {10.29292/jics.v21i1.1094}
}
Kostenlos starten →