Optimization of Multisymbol Architecture of the Entropy Coder for H.265/HEVC Video Encoders

Video compression takes advantage of entropy coding based on binary arithmetic coding to achieve lower-bit rates. On the other hand, the throughput of hardware implementations is limited by intersymbol dependencies existing in the entropy coding. This article presents a series of optimizations appli...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on computer-aided design of integrated circuits and systems Jg. 44; H. 2; S. 458 - 468
1. Verfasser: Pastuszak, Grzegorz
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York IEEE 01.02.2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
ISSN:0278-0070, 1937-4151
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Video compression takes advantage of entropy coding based on binary arithmetic coding to achieve lower-bit rates. On the other hand, the throughput of hardware implementations is limited by intersymbol dependencies existing in the entropy coding. This article presents a series of optimizations applied to the architecture of the entropy coder presented in previous studies. The reference architecture uses several parallel processing paths for serialization, binarization, and probability model (PM) adaptation. Within this study, the paths are modified by increasing the number of symbols processed in one clock cycle and enlarging buffers. The throughput is also improved by more efficient multiplexing between paths. The parallelism of the PM adaptation is increased to eight and seven symbols per clock cycle for significance map and greater-than-1 contexts, respectively. Critical paths associated with state transitions are shortened by changing probability state representation from natural binary code to unary one. The optimized entropy coder increases the average symbol rate from 16.22 to 33.19 bins per clock cycle for high-quality low-delay H.265/HEVC compression. The optimized design for TSMC 65nm technology consumes 458.51k gates and can operate at 666 MHz, allowing for a throughput of 22 014 Mbins/s.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0278-0070
1937-4151
DOI:10.1109/TCAD.2024.3437339