Hardware implementation of PSO-based approximate DST transform for VVC standard
| Autor: | Sonda Ben Jdidia, Fatma Belghith, Nouri Masmoudi, Maher Jridi, Amin Sallem |
|---|---|
| Přispěvatelé: | Vision et Analyse de Données (LABISEN-VISION-AD), Laboratoire ISEN (L@BISEN), Institut supérieur de l'électronique et du numérique (ISEN)-YNCREA OUEST (YO)-Institut supérieur de l'électronique et du numérique (ISEN)-YNCREA OUEST (YO), Laboratoire d'électronique et des technologies de l'Information [Sfax] (LETI), École Nationale d'Ingénieurs de Sfax | National School of Engineers of Sfax (ENIS) |
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
Optimization problem
Computational complexity theory business.industry Computer science Particle swarm optimization 020206 networking & telecommunications 02 engineering and technology Distortion Algorithmic efficiency 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing business Field-programmable gate array Encoder [SPI.SIGNAL]Engineering Sciences [physics]/Signal and Image processing Computer hardware Energy (signal processing) ComputingMilieux_MISCELLANEOUS Information Systems |
| Zdroj: | Journal of Real-Time Image Processing Journal of Real-Time Image Processing, Springer Verlag, 2021, ⟨10.1007/s11554-021-01160-5⟩ |
| ISSN: | 1861-8200 1861-8219 |
| DOI: | 10.1007/s11554-021-01160-5⟩ |
| Popis: | The H.266/Versatile Video Coding (VVC) standard, released in July 2020, has improved the encoder performance over the previous High Efficiency Video Coding (HEVC) with a significant increase in coding complexity. Enhancements on the transform module mainly involve the introduction of the Adaptive Multiple Transform (AMT) which has led to an additional computational complexity. This paper aims at reducing the transform module complexity by approximating the AMT core. The transform approximation has to reach a low MSE, a low total error energy, a low transform distortion and a high transform efficiency. The Particle Swarm Optimization (PSO) is used to solve the optimization problem modeled as a constrained one. The proposed approximate transforms preserve a good coding efficiency compared to the exact transforms and require a less arithmetic complexity as well. The hardware architectures of both the exact and the approximate versions of the 8, and 16-point DST VII transform are designed. The exact transforms are defined using multipliers and MCM-based designs. The approximate transforms are described using additions and bit-shifting operations. All the designs are implemented in the Arria 10 FPGA device. Synthesis results have shown that the proposed approximation saves more than 75% and 63% of logic utilization when compared to multipliers and MCM-based designs, respectively. The maximum operational frequency is of 180 MHz, supporting 2K and 4K videos at 231 and 58 fps, respectively. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink